Torchscript

# Conflicts:
#	.github/workflows/build.yml
#	kmath-core/src/commonMain/kotlin/space/kscience/kmath/nd/StructureND.kt
#	kmath-tensors/src/commonMain/kotlin/space/kscience/kmath/tensors/core/internal/TensorLinearStructure.kt

kmath-memory/src/commonMain/kotlin/space/kscience/kmath/memory/DeferScope.kt

@@ -0,0 +1,7 @@
+package space.kscience.kmath.memory
+
+public expect class DeferScope {
+    public inline fun defer(crossinline block: () -> Unit)
+}
+
+public expect inline fun <R> withDeferScope(block: DeferScope.() -> R): R

kmath-memory/src/jsMain/kotlin/space/kscience/kmath/memory/DeferScopeJS.kt

@@ -0,0 +1,30 @@
+package space.kscience.kmath.memory
+
+private typealias Deferred = () -> Unit
+
+public actual class DeferScope {
+    @PublishedApi
+    internal val deferred: MutableList<Deferred> = mutableListOf()
+
+    @PublishedApi
+    internal fun executeAllDeferred() {
+        deferred.forEach(Deferred::invoke)
+        deferred.clear()
+    }
+
+    public actual inline fun defer(crossinline block: () -> Unit) {
+        deferred += {
+            try {
+                block()
+            } catch (ignored: Throwable) {
+            }
+        }
+    }
+}
+
+public actual inline fun <R> withDeferScope(block: DeferScope.() -> R): R {
+    val ds = DeferScope()
+    val r = ds.block()
+    ds.executeAllDeferred()
+    return r
+}

kmath-memory/src/jvmMain/kotlin/space/kscience/kmath/memory/DeferScopeJvm.kt

@@ -0,0 +1,30 @@
+package space.kscience.kmath.memory
+
+private typealias Deferred = () -> Unit
+
+public actual class DeferScope {
+    @PublishedApi
+    internal val deferred: MutableList<Deferred> = mutableListOf()
+
+    @PublishedApi
+    internal fun executeAllDeferred() {
+        deferred.forEach(Deferred::invoke)
+        deferred.clear()
+    }
+
+    public actual inline fun defer(crossinline block: () -> Unit) {
+        deferred += {
+            try {
+                block()
+            } catch (ignored: Throwable) {
+            }
+        }
+    }
+}
+
+public actual inline fun <R> withDeferScope(block: DeferScope.() -> R): R {
+    val ds = DeferScope()
+    val r = ds.block()
+    ds.executeAllDeferred()
+    return r
+}

kmath-memory/src/nativeMain/kotlin/space/kscience/kmath/memory/DeferScopeNative.kt

@@ -0,0 +1,7 @@
+package space.kscience.kmath.memory
+
+import kotlinx.cinterop.memScoped
+
+public actual typealias DeferScope = kotlinx.cinterop.DeferScope
+
+public actual inline fun <R> withDeferScope(block: DeferScope.() -> R): R = memScoped(block)

kmath-torch/README.md

@@ -0,0 +1,66 @@
+# LibTorch extension (`kmath-torch`)
+
+This is a `Kotlin/Native` & `JVM` module, with only `linuxX64` supported so far. The library wraps some of
+the [PyTorch C++ API](https://pytorch.org/cppdocs), focusing on integrating `Aten` & `Autograd` with `KMath`.
+
+## Installation
+
+To install the library, you have to build & publish locally `kmath-core`, `kmath-memory` with `kmath-torch`:
+
+```
+./gradlew -q :kmath-core:publishToMavenLocal :kmath-memory:publishToMavenLocal :kmath-torch:publishToMavenLocal
+```
+
+This builds `ctorch` a C wrapper and `jtorch` a JNI wrapper for `LibTorch`, placed inside:
+
+`~/.konan/third-party/kmath-torch-0.2.0/cpp-build`
+
+You will have to link against it in your own project.
+
+## Usage
+
+Tensors are implemented over the `MutableNDStructure`. They can only be created through provided factory methods
+and require scoping within a `TensorAlgebra` instance:
+
+```kotlin
+TorchTensorRealAlgebra {
+
+    val realTensor: TorchTensorReal = copyFromArray(
+        array = (1..10).map { it + 50.0 }.toList().toDoubleArray(),
+        shape = intArrayOf(2, 5)
+    )
+    println(realTensor)
+
+    val gpuRealTensor: TorchTensorReal = copyFromArray(
+        array = (1..8).map { it * 2.5 }.toList().toDoubleArray(),
+        shape = intArrayOf(2, 2, 2),
+        device = Device.CUDA(0)
+    )
+    println(gpuRealTensor)
+}
+```
+
+High performance automatic differentiation engine is available:
+
+```kotlin
+TorchTensorRealAlgebra {
+    val dim = 10
+    val device = Device.CPU //or Device.CUDA(0)
+    
+    val tensorX = randNormal(shape = intArrayOf(dim), device = device)
+    val randFeatures = randNormal(shape = intArrayOf(dim, dim), device = device)
+    val tensorSigma = randFeatures + randFeatures.transpose(0, 1)
+    val tensorMu = randNormal(shape = intArrayOf(dim), device = device)
+
+    // expression to differentiate w.r.t. x evaluated at x = tensorX
+    val expressionAtX = withGradAt(tensorX, { x ->
+        0.5 * (x dot (tensorSigma dot x)) + (tensorMu dot x) + 25.9
+    })
+
+    // value of the gradient at x = tensorX
+    val gradientAtX = expressionAtX.grad(tensorX, retainGraph = true)
+    // value of the hessian at x = tensorX
+    val hessianAtX = expressionAtX hess tensorX
+}
+```
+Contributed by [Roland Grinis](https://github.com/rgrit91)

kmath-torch/api/kmath-torch.api

@@ -0,0 +1,397 @@
+public abstract class space/kscience/kmath/torch/Device {
+	public static final field Companion Lspace/kscience/kmath/torch/Device$Companion;
+	public final fun toInt ()I
+}
+
+public final class space/kscience/kmath/torch/Device$CPU : space/kscience/kmath/torch/Device {
+	public static final field INSTANCE Lspace/kscience/kmath/torch/Device$CPU;
+	public fun toString ()Ljava/lang/String;
+}
+
+public final class space/kscience/kmath/torch/Device$CUDA : space/kscience/kmath/torch/Device {
+	public fun <init> (I)V
+	public final fun component1 ()I
+	public final fun copy (I)Lspace/kscience/kmath/torch/Device$CUDA;
+	public static synthetic fun copy$default (Lspace/kscience/kmath/torch/Device$CUDA;IILjava/lang/Object;)Lspace/kscience/kmath/torch/Device$CUDA;
+	public fun equals (Ljava/lang/Object;)Z
+	public final fun getIndex ()I
+	public fun hashCode ()I
+	public fun toString ()Ljava/lang/String;
+}
+
+public final class space/kscience/kmath/torch/Device$Companion {
+	public final fun fromInt (I)Lspace/kscience/kmath/torch/Device;
+}
+
+public abstract interface class space/kscience/kmath/torch/TorchTensor : space/kscience/kmath/tensors/TensorStructure {
+	public abstract fun elements ()Lkotlin/sequences/Sequence;
+	public abstract fun getDevice ()Lspace/kscience/kmath/torch/Device;
+	public abstract fun getSize ()I
+	public abstract fun getStrides ()[I
+}
+
+public final class space/kscience/kmath/torch/TorchTensor$DefaultImpls {
+	public static fun elements (Lspace/kscience/kmath/torch/TorchTensor;)Lkotlin/sequences/Sequence;
+	public static fun getDimension (Lspace/kscience/kmath/torch/TorchTensor;)I
+	public static fun value (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+}
+
+public abstract interface class space/kscience/kmath/torch/TorchTensorAlgebra : space/kscience/kmath/tensors/TensorAlgebra {
+	public abstract fun copy (Lspace/kscience/kmath/torch/TorchTensor;)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun copyFromArray (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun copyToArray (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+	public abstract fun copyToDevice (Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun cudaAvailable ()Z
+	public abstract fun full (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun getChecks ()Z
+	public abstract fun getNumThreads ()I
+	public abstract fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun randIntegral (Lspace/kscience/kmath/torch/TorchTensor;JJ)Lspace/kscience/kmath/torch/TorchTensor;
+	public abstract fun randIntegralAssign (Lspace/kscience/kmath/torch/TorchTensor;JJ)V
+	public abstract fun setChecks (Z)V
+	public abstract fun setNumThreads (I)V
+	public abstract fun setSeed (I)V
+	public abstract fun swap (Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/TorchTensor;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorAlgebra$DefaultImpls {
+	public static synthetic fun copyFromArray$default (Lspace/kscience/kmath/torch/TorchTensorAlgebra;Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;ILjava/lang/Object;)Lspace/kscience/kmath/torch/TorchTensor;
+	public static synthetic fun randIntegral$default (Lspace/kscience/kmath/torch/TorchTensorAlgebra;JJ[ILspace/kscience/kmath/torch/Device;ILjava/lang/Object;)Lspace/kscience/kmath/torch/TorchTensor;
+}
+
+public abstract class space/kscience/kmath/torch/TorchTensorAlgebraJVM : space/kscience/kmath/torch/TorchTensorAlgebra {
+	public synthetic fun <init> (Lspace/kscience/kmath/memory/DeferScope;Lkotlin/jvm/internal/DefaultConstructorMarker;)V
+	public synthetic fun abs (Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun abs (Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun absAssign (Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun absAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun copy (Lspace/kscience/kmath/torch/TorchTensor;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copy (Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun copyToDevice (Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copyToDevice (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public fun cudaAvailable ()Z
+	public synthetic fun diagonalEmbedding (Lspace/kscience/kmath/tensors/TensorStructure;III)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun diagonalEmbedding (Lspace/kscience/kmath/torch/TorchTensorJVM;III)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun dot (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun dot (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun dotAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun dotAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun dotRightAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun dotRightAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public fun getChecks ()Z
+	public fun getNumThreads ()I
+	public synthetic fun minus (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun minus (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun minusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun minusAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun plus (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun plus (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun plusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun plusAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun randIntegral (Lspace/kscience/kmath/torch/TorchTensor;JJ)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun randIntegral (Lspace/kscience/kmath/torch/TorchTensorJVM;JJ)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun randIntegralAssign (Lspace/kscience/kmath/torch/TorchTensor;JJ)V
+	public fun randIntegralAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;JJ)V
+	public fun setChecks (Z)V
+	public fun setNumThreads (I)V
+	public fun setSeed (I)V
+	public synthetic fun sum (Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun sum (Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun sumAssign (Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun sumAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun swap (Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/TorchTensor;)V
+	public fun swap (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun times (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun times (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun timesAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun timesAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;Lspace/kscience/kmath/torch/TorchTensorJVM;)V
+	public synthetic fun transpose (Lspace/kscience/kmath/tensors/TensorStructure;II)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun transpose (Lspace/kscience/kmath/torch/TorchTensorJVM;II)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun transposeAssign (Lspace/kscience/kmath/tensors/TensorStructure;II)V
+	public fun transposeAssign (Lspace/kscience/kmath/torch/TorchTensorJVM;II)V
+	public synthetic fun unaryMinus (Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun unaryMinus (Lspace/kscience/kmath/torch/TorchTensorJVM;)Lspace/kscience/kmath/torch/TorchTensorJVM;
+	public synthetic fun view (Lspace/kscience/kmath/tensors/TensorStructure;[I)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun view (Lspace/kscience/kmath/torch/TorchTensorJVM;[I)Lspace/kscience/kmath/torch/TorchTensorJVM;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorAlgebraJVMKt {
+	public static final fun TorchTensorFloatAlgebra (Lkotlin/jvm/functions/Function1;)Ljava/lang/Object;
+	public static final fun TorchTensorIntAlgebra (Lkotlin/jvm/functions/Function1;)Ljava/lang/Object;
+	public static final fun TorchTensorLongAlgebra (Lkotlin/jvm/functions/Function1;)Ljava/lang/Object;
+	public static final fun TorchTensorRealAlgebra (Lkotlin/jvm/functions/Function1;)Ljava/lang/Object;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorAlgebraKt {
+	public static final fun checkDeviceCompatible (Lspace/kscience/kmath/torch/TorchTensorAlgebra;Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/TorchTensor;)V
+	public static final fun checkDotOperation (Lspace/kscience/kmath/torch/TorchTensorAlgebra;Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/TorchTensor;)V
+	public static final fun checkLinearOperation (Lspace/kscience/kmath/torch/TorchTensorAlgebra;Lspace/kscience/kmath/torch/TorchTensor;Lspace/kscience/kmath/torch/TorchTensor;)V
+	public static final fun withChecks (Lspace/kscience/kmath/torch/TorchTensorAlgebra;Lkotlin/jvm/functions/Function1;)V
+	public static final fun withGradAt (Lspace/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra;Lspace/kscience/kmath/torch/TorchTensorOverField;Lkotlin/jvm/functions/Function2;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorFloat : space/kscience/kmath/torch/TorchTensorOverFieldJVM {
+	public fun get ([I)Ljava/lang/Float;
+	public synthetic fun get ([I)Ljava/lang/Object;
+	public fun item ()Ljava/lang/Float;
+	public synthetic fun item ()Ljava/lang/Object;
+	public fun set ([IF)V
+	public synthetic fun set ([ILjava/lang/Object;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorFloatAlgebra : space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebraJVM {
+	public fun <init> (Lspace/kscience/kmath/memory/DeferScope;)V
+	public synthetic fun copyFromArray (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copyFromArray ([F[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun copyToArray (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+	public fun copyToArray (Lspace/kscience/kmath/torch/TorchTensorFloat;)[F
+	public fun full (F[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun full (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun minus (FLspace/kscience/kmath/torch/TorchTensorFloat;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun minus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun minus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun minus (Lspace/kscience/kmath/torch/TorchTensorFloat;F)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun minusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun minusAssign (Lspace/kscience/kmath/torch/TorchTensorFloat;F)V
+	public fun plus (FLspace/kscience/kmath/torch/TorchTensorFloat;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun plus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun plus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun plus (Lspace/kscience/kmath/torch/TorchTensorFloat;F)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun plusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun plusAssign (Lspace/kscience/kmath/torch/TorchTensorFloat;F)V
+	public synthetic fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public fun randNormal ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun randNormal ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun randUniform ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun randUniform ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun times (FLspace/kscience/kmath/torch/TorchTensorFloat;)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun times (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun times (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun times (Lspace/kscience/kmath/torch/TorchTensorFloat;F)Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public synthetic fun timesAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun timesAssign (Lspace/kscience/kmath/torch/TorchTensorFloat;F)V
+	public synthetic fun wrap$kmath_torch (J)Lspace/kscience/kmath/torch/TorchTensorJVM;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorInt : space/kscience/kmath/torch/TorchTensorOverFieldJVM {
+	public fun get ([I)Ljava/lang/Integer;
+	public synthetic fun get ([I)Ljava/lang/Object;
+	public fun item ()Ljava/lang/Integer;
+	public synthetic fun item ()Ljava/lang/Object;
+	public fun set ([II)V
+	public synthetic fun set ([ILjava/lang/Object;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorIntAlgebra : space/kscience/kmath/torch/TorchTensorAlgebraJVM {
+	public fun <init> (Lspace/kscience/kmath/memory/DeferScope;)V
+	public synthetic fun copyFromArray (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copyFromArray ([I[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun copyToArray (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+	public fun copyToArray (Lspace/kscience/kmath/torch/TorchTensorInt;)[I
+	public fun full (I[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun full (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun minus (ILspace/kscience/kmath/torch/TorchTensorInt;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun minus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun minus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun minus (Lspace/kscience/kmath/torch/TorchTensorInt;I)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun minusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun minusAssign (Lspace/kscience/kmath/torch/TorchTensorInt;I)V
+	public fun plus (ILspace/kscience/kmath/torch/TorchTensorInt;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun plus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun plus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun plus (Lspace/kscience/kmath/torch/TorchTensorInt;I)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun plusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun plusAssign (Lspace/kscience/kmath/torch/TorchTensorInt;I)V
+	public synthetic fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public fun times (ILspace/kscience/kmath/torch/TorchTensorInt;)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun times (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun times (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun times (Lspace/kscience/kmath/torch/TorchTensorInt;I)Lspace/kscience/kmath/torch/TorchTensorInt;
+	public synthetic fun timesAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun timesAssign (Lspace/kscience/kmath/torch/TorchTensorInt;I)V
+	public synthetic fun wrap$kmath_torch (J)Lspace/kscience/kmath/torch/TorchTensorJVM;
+}
+
+public abstract class space/kscience/kmath/torch/TorchTensorJVM : space/kscience/kmath/torch/TorchTensorMemoryHolder, space/kscience/kmath/torch/TorchTensor {
+	public synthetic fun <init> (Lspace/kscience/kmath/memory/DeferScope;JLkotlin/jvm/internal/DefaultConstructorMarker;)V
+	protected fun close ()V
+	public final fun copyToDouble ()Lspace/kscience/kmath/torch/TorchTensorReal;
+	public final fun copyToFloat ()Lspace/kscience/kmath/torch/TorchTensorFloat;
+	public final fun copyToInt ()Lspace/kscience/kmath/torch/TorchTensorInt;
+	public final fun copyToLong ()Lspace/kscience/kmath/torch/TorchTensorLong;
+	public fun elements ()Lkotlin/sequences/Sequence;
+	public fun getDevice ()Lspace/kscience/kmath/torch/Device;
+	public fun getDimension ()I
+	public fun getShape ()[I
+	public fun getSize ()I
+	public fun getStrides ()[I
+	public fun toString ()Ljava/lang/String;
+	public fun value ()Ljava/lang/Object;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorLong : space/kscience/kmath/torch/TorchTensorOverFieldJVM {
+	public fun get ([I)Ljava/lang/Long;
+	public synthetic fun get ([I)Ljava/lang/Object;
+	public fun item ()Ljava/lang/Long;
+	public synthetic fun item ()Ljava/lang/Object;
+	public fun set ([IJ)V
+	public synthetic fun set ([ILjava/lang/Object;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorLongAlgebra : space/kscience/kmath/torch/TorchTensorAlgebraJVM {
+	public fun <init> (Lspace/kscience/kmath/memory/DeferScope;)V
+	public synthetic fun copyFromArray (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copyFromArray ([J[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun copyToArray (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+	public fun copyToArray (Lspace/kscience/kmath/torch/TorchTensorLong;)[J
+	public fun full (J[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun full (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun minus (JLspace/kscience/kmath/torch/TorchTensorLong;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun minus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun minus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun minus (Lspace/kscience/kmath/torch/TorchTensorLong;J)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun minusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun minusAssign (Lspace/kscience/kmath/torch/TorchTensorLong;J)V
+	public fun plus (JLspace/kscience/kmath/torch/TorchTensorLong;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun plus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun plus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun plus (Lspace/kscience/kmath/torch/TorchTensorLong;J)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun plusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun plusAssign (Lspace/kscience/kmath/torch/TorchTensorLong;J)V
+	public synthetic fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public fun times (JLspace/kscience/kmath/torch/TorchTensorLong;)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun times (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun times (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun times (Lspace/kscience/kmath/torch/TorchTensorLong;J)Lspace/kscience/kmath/torch/TorchTensorLong;
+	public synthetic fun timesAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun timesAssign (Lspace/kscience/kmath/torch/TorchTensorLong;J)V
+	public synthetic fun wrap$kmath_torch (J)Lspace/kscience/kmath/torch/TorchTensorJVM;
+}
+
+public abstract class space/kscience/kmath/torch/TorchTensorMemoryHolder {
+	protected abstract fun close ()V
+	public fun equals (Ljava/lang/Object;)Z
+	public final fun getScope ()Lspace/kscience/kmath/memory/DeferScope;
+	public fun hashCode ()I
+}
+
+public abstract interface class space/kscience/kmath/torch/TorchTensorOverField : space/kscience/kmath/torch/TorchTensor {
+	public abstract fun getRequiresGrad ()Z
+	public abstract fun setRequiresGrad (Z)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorOverField$DefaultImpls {
+	public static fun elements (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lkotlin/sequences/Sequence;
+	public static fun getDimension (Lspace/kscience/kmath/torch/TorchTensorOverField;)I
+	public static fun value (Lspace/kscience/kmath/torch/TorchTensorOverField;)Ljava/lang/Object;
+}
+
+public abstract class space/kscience/kmath/torch/TorchTensorOverFieldJVM : space/kscience/kmath/torch/TorchTensorJVM, space/kscience/kmath/torch/TorchTensorOverField {
+	public synthetic fun <init> (Lspace/kscience/kmath/memory/DeferScope;JLkotlin/jvm/internal/DefaultConstructorMarker;)V
+	public fun getRequiresGrad ()Z
+	public fun setRequiresGrad (Z)V
+}
+
+public abstract interface class space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra : space/kscience/kmath/tensors/TensorPartialDivisionAlgebra, space/kscience/kmath/torch/TorchTensorAlgebra {
+	public abstract fun detachFromGraph (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun grad (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun grad (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;Z)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun hess (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun randNormal (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun randNormal ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun randNormalAssign (Lspace/kscience/kmath/torch/TorchTensorOverField;)V
+	public abstract fun randUniform (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun randUniform ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public abstract fun randUniformAssign (Lspace/kscience/kmath/torch/TorchTensorOverField;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra$DefaultImpls {
+	public static fun grad (Lspace/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra;Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public static synthetic fun grad$default (Lspace/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra;Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;ZILjava/lang/Object;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public static synthetic fun randNormal$default (Lspace/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra;[ILspace/kscience/kmath/torch/Device;ILjava/lang/Object;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public static synthetic fun randUniform$default (Lspace/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra;[ILspace/kscience/kmath/torch/Device;ILjava/lang/Object;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+}
+
+public abstract class space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebraJVM : space/kscience/kmath/torch/TorchTensorAlgebraJVM, space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebra {
+	public synthetic fun <init> (Lspace/kscience/kmath/memory/DeferScope;Lkotlin/jvm/internal/DefaultConstructorMarker;)V
+	public synthetic fun detachFromGraph (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun detachFromGraph (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun div (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun div (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun divAssign (Lspace/kscience/kmath/tensors/TensorStructure;Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun divAssign (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)V
+	public synthetic fun exp (Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun exp (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun expAssign (Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun expAssign (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)V
+	public synthetic fun grad (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public synthetic fun grad (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;Z)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun grad (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public fun grad (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Z)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun hess (Lspace/kscience/kmath/torch/TorchTensorOverField;Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun hess (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun log (Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun log (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun logAssign (Lspace/kscience/kmath/tensors/TensorStructure;)V
+	public fun logAssign (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)V
+	public synthetic fun randNormal (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun randNormal (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun randNormalAssign (Lspace/kscience/kmath/torch/TorchTensorOverField;)V
+	public fun randNormalAssign (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)V
+	public synthetic fun randUniform (Lspace/kscience/kmath/torch/TorchTensorOverField;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun randUniform (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;
+	public synthetic fun randUniformAssign (Lspace/kscience/kmath/torch/TorchTensorOverField;)V
+	public fun randUniformAssign (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)V
+	public synthetic fun svd (Lspace/kscience/kmath/tensors/TensorStructure;)Lkotlin/Triple;
+	public fun svd (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;)Lkotlin/Triple;
+	public synthetic fun symEig (Lspace/kscience/kmath/tensors/TensorStructure;Z)Lkotlin/Pair;
+	public fun symEig (Lspace/kscience/kmath/torch/TorchTensorOverFieldJVM;Z)Lkotlin/Pair;
+}
+
+public final class space/kscience/kmath/torch/TorchTensorReal : space/kscience/kmath/torch/TorchTensorOverFieldJVM {
+	public fun get ([I)Ljava/lang/Double;
+	public synthetic fun get ([I)Ljava/lang/Object;
+	public fun item ()Ljava/lang/Double;
+	public synthetic fun item ()Ljava/lang/Object;
+	public fun set ([ID)V
+	public synthetic fun set ([ILjava/lang/Object;)V
+}
+
+public final class space/kscience/kmath/torch/TorchTensorRealAlgebra : space/kscience/kmath/torch/TorchTensorPartialDivisionAlgebraJVM {
+	public fun <init> (Lspace/kscience/kmath/memory/DeferScope;)V
+	public synthetic fun copyFromArray (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun copyFromArray ([D[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun copyToArray (Lspace/kscience/kmath/torch/TorchTensor;)Ljava/lang/Object;
+	public fun copyToArray (Lspace/kscience/kmath/torch/TorchTensorReal;)[D
+	public fun full (D[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun full (Ljava/lang/Object;[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun minus (DLspace/kscience/kmath/torch/TorchTensorReal;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun minus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun minus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun minus (Lspace/kscience/kmath/torch/TorchTensorReal;D)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun minusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun minusAssign (Lspace/kscience/kmath/torch/TorchTensorReal;D)V
+	public fun plus (DLspace/kscience/kmath/torch/TorchTensorReal;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun plus (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun plus (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun plus (Lspace/kscience/kmath/torch/TorchTensorReal;D)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun plusAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun plusAssign (Lspace/kscience/kmath/torch/TorchTensorReal;D)V
+	public synthetic fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensor;
+	public fun randIntegral (JJ[ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun randNormal ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun randNormal ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun randUniform ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorOverField;
+	public fun randUniform ([ILspace/kscience/kmath/torch/Device;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public fun times (DLspace/kscience/kmath/torch/TorchTensorReal;)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun times (Ljava/lang/Object;Lspace/kscience/kmath/tensors/TensorStructure;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public synthetic fun times (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)Lspace/kscience/kmath/tensors/TensorStructure;
+	public fun times (Lspace/kscience/kmath/torch/TorchTensorReal;D)Lspace/kscience/kmath/torch/TorchTensorReal;
+	public synthetic fun timesAssign (Lspace/kscience/kmath/tensors/TensorStructure;Ljava/lang/Object;)V
+	public fun timesAssign (Lspace/kscience/kmath/torch/TorchTensorReal;D)V
+	public synthetic fun wrap$kmath_torch (J)Lspace/kscience/kmath/torch/TorchTensorJVM;
+}
+

kmath-torch/build.gradle.kts

@@ -0,0 +1,194 @@
+import de.undercouch.gradle.tasks.download.Download
+import org.jetbrains.kotlin.gradle.plugin.mpp.KotlinNativeTarget
+import org.gradle.api.JavaVersion.VERSION_11
+
+
+plugins {
+    id("ru.mipt.npm.gradle.mpp")
+    id("de.undercouch.download")
+}
+
+java {
+    sourceCompatibility = VERSION_11
+    targetCompatibility = VERSION_11
+}
+
+val home = System.getProperty("user.home")
+val javaHome = System.getProperty("java.home")
+val thirdPartyDir = "$home/.konan/third-party/kmath-torch-${project.property("version")}"
+val cppBuildDir = "$thirdPartyDir/cpp-build"
+val cppSources = projectDir.resolve("src/cppMain")
+
+val cudaHome: String? = System.getenv("CUDA_HOME")
+val cudaDefault = file("/usr/local/cuda").exists()
+val cudaFound = cudaHome?.isNotEmpty() ?: false or cudaDefault
+
+val cmakeArchive = "cmake-3.19.2-Linux-x86_64"
+val torchArchive = "libtorch"
+
+val cmakeCmd = "$thirdPartyDir/$cmakeArchive/bin/cmake"
+val ninjaCmd = "$thirdPartyDir/ninja"
+
+val downloadCMake by tasks.registering(Download::class) {
+    val tarFile = "$cmakeArchive.tar.gz"
+    src("https://github.com/Kitware/CMake/releases/download/v3.19.2/$tarFile")
+    dest(File(thirdPartyDir, tarFile))
+    overwrite(false)
+}
+
+val downloadNinja by tasks.registering(Download::class) {
+    src("https://github.com/ninja-build/ninja/releases/download/v1.10.2/ninja-linux.zip")
+    dest(File(thirdPartyDir, "ninja-linux.zip"))
+    overwrite(false)
+}
+
+val downloadTorch by tasks.registering(Download::class) {
+    val abiMeta = "$torchArchive-cxx11-abi-shared-with-deps-1.7.1%2B"
+    val cudaUrl = "https://download.pytorch.org/libtorch/cu110/${abiMeta}cu110.zip"
+    val cpuUrl = "https://download.pytorch.org/libtorch/cpu/${abiMeta}cpu.zip"
+    val url = if (cudaFound) cudaUrl else cpuUrl
+    src(url)
+    dest(File(thirdPartyDir, "$torchArchive.zip"))
+    overwrite(false)
+}
+
+val extractCMake by tasks.registering(Copy::class) {
+    dependsOn(downloadCMake)
+    from(tarTree(resources.gzip(downloadCMake.get().dest)))
+    into(thirdPartyDir)
+}
+
+val extractTorch by tasks.registering(Copy::class) {
+    dependsOn(downloadTorch)
+    from(zipTree(downloadTorch.get().dest))
+    into(thirdPartyDir)
+}
+
+val extractNinja by tasks.registering(Copy::class) {
+    dependsOn(downloadNinja)
+    from(zipTree(downloadNinja.get().dest))
+    into(thirdPartyDir)
+}
+
+val configureCpp by tasks.registering {
+    dependsOn(extractCMake)
+    dependsOn(extractNinja)
+    dependsOn(extractTorch)
+    onlyIf { !file(cppBuildDir).exists() }
+    doLast {
+        exec {
+            workingDir(thirdPartyDir)
+            commandLine("mkdir", "-p", cppBuildDir)
+        }
+        exec {
+            workingDir(cppBuildDir)
+            commandLine(
+                cmakeCmd,
+                cppSources,
+                "-GNinja",
+                "-DCMAKE_MAKE_PROGRAM=$ninjaCmd",
+                "-DCMAKE_PREFIX_PATH=$thirdPartyDir/$torchArchive",
+                "-DJAVA_HOME=$javaHome",
+                "-DCMAKE_BUILD_TYPE=Release"
+            )
+        }
+    }
+}
+
+val cleanCppBuild by tasks.registering {
+    onlyIf { file(cppBuildDir).exists() }
+    doLast {
+        exec {
+            workingDir(thirdPartyDir)
+            commandLine("rm", "-rf", cppBuildDir)
+        }
+    }
+}
+
+val buildCpp by tasks.registering {
+    dependsOn(configureCpp)
+    doLast {
+        exec {
+            workingDir(cppBuildDir)
+            commandLine(cmakeCmd, "--build", ".", "--config", "Release")
+        }
+    }
+}
+
+val generateJNIHeader by tasks.registering {
+    doLast {
+        exec {
+            workingDir(projectDir.resolve("src/jvmMain/java/space/kscience/kmath/torch"))
+            commandLine("$javaHome/bin/javac", "-h", cppSources.resolve("include") , "JTorch.java")
+        }
+    }
+}
+
+kotlin {
+    explicitApiWarning()
+
+    jvm {
+        withJava()
+    }
+
+    val nativeTarget = linuxX64("native")
+    nativeTarget.apply {
+        binaries {
+            all {
+                linkerOpts(
+                    "-L$cppBuildDir",
+                    "-Wl,-rpath=$cppBuildDir",
+                    "-lctorch"
+                )
+                optimized = true
+                debuggable = false
+            }
+        }
+    }
+
+    val main by nativeTarget.compilations.getting {
+        cinterops {
+            val libctorch by creating {
+                includeDirs(cppSources.resolve("include"))
+            }
+        }
+    }
+
+    val test by nativeTarget.compilations.getting
+
+    sourceSets {
+        val commonMain by getting {
+            dependencies {
+                api(project(":kmath-core"))
+            }
+        }
+
+        val nativeMain by getting {
+            dependencies {
+                api(project(":kmath-core"))
+            }
+        }
+
+        val jvmMain by getting {
+            dependencies {
+                api(project(":kmath-core"))
+            }
+        }
+    }
+}
+
+val native: KotlinNativeTarget by kotlin.targets
+tasks[native.compilations["main"].cinterops["libctorch"].interopProcessingTaskName]
+    .dependsOn(buildCpp)
+
+tasks["jvmProcessResources"].dependsOn(buildCpp)
+
+tasks {
+    withType<Test>{
+        systemProperty("java.library.path", cppBuildDir)
+    }
+}
+
+// No JS implementation
+project.gradle.startParameter.excludedTaskNames.add("jsTest")
+project.gradle.startParameter.excludedTaskNames.add("jsBrowserTest")

kmath-torch/src/commonMain/kotlin/space/kscience/kmath/torch/Device.kt

@@ -0,0 +1,24 @@
+package space.kscience.kmath.torch
+
+
+public sealed class Device {
+    public object CPU: Device() {
+        override fun toString(): String {
+            return "CPU"
+        }
+    }
+    public data class CUDA(val index: Int): Device()
+    public fun toInt(): Int {
+        when(this) {
+            is CPU -> return 0
+            is CUDA -> return this.index + 1
+        }
+    }
+    public companion object {
+        public fun fromInt(deviceInt: Int): Device {
+            return if (deviceInt == 0) CPU else CUDA(
+                deviceInt - 1
+            )
+        }
+    }
+}

kmath-torch/src/commonMain/kotlin/space/kscience/kmath/torch/TorchTensor.kt

@@ -0,0 +1,26 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.tensors.*
+
+public interface TorchTensor<T> : TensorStructure<T> {
+
+    public val strides: IntArray
+    public val size: Int
+    public val device: Device
+
+    override fun elements(): Sequence<Pair<IntArray, T>> {
+        if (dimension == 0) {
+            return emptySequence()
+        }
+        val indices = (1..size).asSequence().map { indexFromOffset(it - 1, strides, dimension) }
+        return indices.map { it to get(it) }
+    }
+}
+
+public interface TorchTensorOverField<T>: TorchTensor<T>
+{
+    public var requiresGrad: Boolean
+}
+

kmath-torch/src/commonMain/kotlin/space/kscience/kmath/torch/TorchTensorAlgebra.kt

@@ -0,0 +1,105 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.tensors.*
+
+public interface TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>> :
+    TensorAlgebra<T, TorchTensorType> {
+
+    public fun getNumThreads(): Int
+    public fun setNumThreads(numThreads: Int): Unit
+    public fun cudaAvailable(): Boolean
+    public fun setSeed(seed: Int): Unit
+
+    public var checks: Boolean
+
+    public fun copyFromArray(
+        array: PrimitiveArrayType,
+        shape: IntArray,
+        device: Device = Device.CPU
+    ): TorchTensorType
+
+    public fun TorchTensorType.copyToArray(): PrimitiveArrayType
+
+    public fun full(value: T, shape: IntArray, device: Device): TorchTensorType
+
+    public fun randIntegral(
+        low: Long, high: Long, shape: IntArray,
+        device: Device = Device.CPU
+    ): TorchTensorType
+    public fun TorchTensorType.randIntegral(low: Long, high: Long): TorchTensorType
+    public fun TorchTensorType.randIntegralAssign(low: Long, high: Long): Unit
+
+    public fun TorchTensorType.copy(): TorchTensorType
+    public fun TorchTensorType.copyToDevice(device: Device): TorchTensorType
+    public infix fun TorchTensorType.swap(other: TorchTensorType)
+
+}
+
+public interface TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType : TorchTensorOverField<T>> :
+    TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>, TensorPartialDivisionAlgebra<T, TorchTensorType> {
+
+    public fun randUniform(shape: IntArray, device: Device = Device.CPU): TorchTensorType
+    public fun randNormal(shape: IntArray, device: Device = Device.CPU): TorchTensorType
+    public fun TorchTensorType.randUniform(): TorchTensorType
+    public fun TorchTensorType.randUniformAssign(): Unit
+    public fun TorchTensorType.randNormal(): TorchTensorType
+    public fun TorchTensorType.randNormalAssign(): Unit
+
+    public fun TorchTensorType.grad(variable: TorchTensorType, retainGraph: Boolean = false): TorchTensorType
+    public infix fun TorchTensorType.grad(variable: TorchTensorType): TorchTensorType =
+        this.grad(variable, false)
+
+    public infix fun TorchTensorType.hess(variable: TorchTensorType): TorchTensorType
+    public fun TorchTensorType.detachFromGraph(): TorchTensorType
+}
+
+
+public inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.withChecks(block: TorchTensorAlgebraType.() -> Unit): Unit {
+    val state = this.checks
+    this.checks = true
+    this.block()
+    this.checks = state
+}
+
+public inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.checkDeviceCompatible(
+    a: TorchTensorType, b: TorchTensorType
+): Unit =
+    check(a.device == b.device) {
+        "Tensors must be on the same device"
+    }
+
+
+public inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.checkLinearOperation(
+    a: TorchTensorType,
+    b: TorchTensorType
+) {
+    if (a.isNotValue() and b.isNotValue()) {
+        checkDeviceCompatible(a, b)
+        checkShapeCompatible(a, b)
+    }
+}
+
+public inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.checkDotOperation(a: TorchTensorType, b: TorchTensorType): Unit {
+    checkDeviceCompatible(a, b)
+    checkDot(a,b)
+}
+
+public inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensorOverField<T>,
+        TorchTensorDivisionAlgebraType : TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorDivisionAlgebraType.withGradAt(
+    tensor: TorchTensorType,
+    block: TorchTensorDivisionAlgebraType.(TorchTensorType) -> TorchTensorType
+): TorchTensorType {
+    tensor.requiresGrad = true
+    return this.block(tensor)
+}

kmath-torch/src/commonMain/kotlin/space/kscience/kmath/torch/TorchTensorMemoryHolder.kt

@@ -0,0 +1,15 @@
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.memory.DeferScope
+
+public abstract class TorchTensorMemoryHolder internal constructor(
+    public val scope: DeferScope
+) {
+    init {
+        scope.defer(::close)
+    }
+    protected abstract fun close(): Unit
+
+    override fun equals(other: Any?): Boolean = false
+    override fun hashCode(): Int = 0
+}

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/BenchmarkMatMult.kt

@@ -0,0 +1,24 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import kotlin.time.measureTime
+
+internal inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensorOverField<T>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkMatMul(
+    scale: Int,
+    numWarmUp: Int,
+    numIter: Int,
+    fieldName: String,
+    device: Device = Device.CPU
+): Unit {
+    println("Benchmarking $scale x $scale $fieldName matrices on $device: ")
+    setSeed(SEED)
+    val lhs = randNormal(shape = intArrayOf(scale, scale), device = device)
+    val rhs = randNormal(shape = intArrayOf(scale, scale), device = device)
+    repeat(numWarmUp) { lhs dotAssign rhs }
+    val measuredTime = measureTime { repeat(numIter) { lhs dotAssign rhs } }
+    println("   ${measuredTime / numIter} p.o. with $numIter iterations")
+}
+

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/BenchmarkRandomGenerators.kt

@@ -0,0 +1,132 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import kotlin.time.measureTime
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkRand(
+    samples: Int,
+    numWarmUp: Int,
+    numIter: Int,
+    device: Device,
+    distName: String,
+    initBock: TorchTensorAlgebraType.(IntArray, Device) -> TorchTensorType,
+    runBlock: TorchTensorAlgebraType.(TorchTensorType) -> Unit
+): Unit{
+    println("Benchmarking generation of $samples $distName samples on $device: ")
+    setSeed(SEED)
+    val shape = intArrayOf(samples)
+    val tensor = this.initBock(shape,device)
+    repeat(numWarmUp) { this.runBlock(tensor) }
+    val measuredTime = measureTime { repeat(numIter) { this.runBlock(tensor) } }
+    println("   ${measuredTime / numIter} p.o. with $numIter iterations")
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkRandNormal(
+    samples: Int,
+    numWarmUp: Int,
+    numIter: Int,
+    device: Device = Device.CPU): Unit{
+    benchmarkRand(
+        samples,
+        numWarmUp,
+        numIter,
+        device,
+        "Normal",
+        {sh, dc -> randNormal(shape = sh, device = dc)},
+        {ten -> ten.randNormalAssign() }
+    )
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkRandUniform(
+    samples: Int,
+    numWarmUp: Int,
+    numIter: Int,
+    device: Device = Device.CPU): Unit{
+    benchmarkRand(
+        samples,
+        numWarmUp,
+        numIter,
+        device,
+        "Uniform",
+        {sh, dc -> randUniform(shape = sh, device = dc)},
+        {ten -> ten.randUniformAssign() }
+    )
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkRandIntegral(
+    samples: Int,
+    numWarmUp: Int,
+    numIter: Int,
+    device: Device = Device.CPU): Unit{
+    benchmarkRand(
+        samples,
+        numWarmUp,
+        numIter,
+        device,
+        "integer [0,100]",
+        {sh, dc -> randIntegral(0, 100, shape = sh, device = dc)},
+        {ten -> ten.randIntegralAssign(0, 100) }
+    )
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkingRand1(): Unit {
+    benchmarkRandNormal(10, 10, 100000)
+    benchmarkRandUniform(10, 10, 100000)
+    benchmarkRandIntegral(10, 10, 100000)
+    if(cudaAvailable()) {
+        benchmarkRandNormal(10, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandUniform(10, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandIntegral(10, 10, 100000, device = Device.CUDA(0))
+    }
+
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkingRand3(): Unit {
+    benchmarkRandNormal(1000, 10, 10000)
+    benchmarkRandUniform(1000, 10, 10000)
+    benchmarkRandIntegral(1000, 10, 10000)
+    if(cudaAvailable()) {
+        benchmarkRandNormal(1000, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandUniform(1000, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandIntegral(1000, 10, 100000, device = Device.CUDA(0))
+    }
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkingRand5(): Unit {
+    benchmarkRandNormal(100000, 5, 100)
+    benchmarkRandUniform(100000, 5, 100)
+    benchmarkRandIntegral(100000, 5, 100)
+    if(cudaAvailable()){
+        benchmarkRandNormal(100000, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandUniform(100000, 10, 100000, device = Device.CUDA(0))
+        benchmarkRandIntegral(100000, 10, 100000, device = Device.CUDA(0))
+    }
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.benchmarkingRand7(): Unit {
+    benchmarkRandNormal(10000000, 3, 20)
+    benchmarkRandUniform(10000000, 3, 20)
+    benchmarkRandIntegral(10000000, 3, 20)
+    if(cudaAvailable()){
+        benchmarkRandNormal(10000000, 10, 10000, device = Device.CUDA(0))
+        benchmarkRandUniform(10000000, 10, 10000, device = Device.CUDA(0))
+        benchmarkRandIntegral(10000000, 10, 10000, device = Device.CUDA(0))
+    }
+}

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/TestAutograd.kt

@@ -0,0 +1,53 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import kotlin.test.assertTrue
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingAutoGrad(device: Device = Device.CPU): Unit {
+    setSeed(SEED)
+    val dim = 3
+    val tensorX = randNormal(shape = intArrayOf(dim), device = device)
+    val randFeatures = randNormal(shape = intArrayOf(dim, dim), device = device)
+    val tensorSigma = randFeatures + randFeatures.transpose(0, 1)
+    val tensorMu = randNormal(shape = intArrayOf(dim), device = device)
+
+    val expressionAtX = withGradAt(tensorX, { x ->
+        0.5f * (x dot (tensorSigma dot x)) + (tensorMu dot x) + 25.9f
+    })
+
+    val gradientAtX = expressionAtX.grad(tensorX, retainGraph = true)
+    val hessianAtX = expressionAtX hess tensorX
+    val expectedGradientAtX = (tensorSigma dot tensorX) + tensorMu
+
+    val error = (gradientAtX - expectedGradientAtX).abs().sum().value() +
+            (hessianAtX - tensorSigma).abs().sum().value()
+    assertTrue(error < TOLERANCE)
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingBatchedAutoGrad(device: Device = Device.CPU): Unit {
+    setSeed(SEED)
+    val batch = intArrayOf(2)
+    val dim = 2
+    val tensorX = randNormal(shape = batch + intArrayOf(1, dim), device = device)
+    val randFeatures = randNormal(shape = batch + intArrayOf(dim, dim), device = device)
+    val tensorSigma = randFeatures + randFeatures.transpose(-2, -1)
+    val tensorMu = randNormal(shape = batch + intArrayOf(1, dim), device = device)
+
+    val expressionAtX = withGradAt(tensorX, { x ->
+        val xt = x.transpose(-1, -2)
+        0.5f * (x dot (tensorSigma dot xt)) + (tensorMu dot xt) + 58.2f
+    })
+    expressionAtX.sumAssign()
+
+    val gradientAtX = expressionAtX grad tensorX
+    val expectedGradientAtX = (tensorX dot tensorSigma) + tensorMu
+
+    val error = (gradientAtX - expectedGradientAtX).abs().sum().value()
+    assertTrue(error < TOLERANCE)
+}
+

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/TestTorchTensor.kt

@@ -0,0 +1,53 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import kotlin.test.assertEquals
+import kotlin.test.assertTrue
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingCopying(device: Device = Device.CPU): Unit {
+    val array = (1..24).map { 10f * it * it }.toFloatArray()
+    val shape = intArrayOf(2, 3, 4)
+    val tensor = copyFromArray(array, shape = shape, device = device)
+    val copyOfTensor = tensor.copy()
+    tensor[intArrayOf(1, 2, 3)] = 0.1f
+    assertTrue(copyOfTensor.copyToArray() contentEquals array)
+    assertEquals(0.1f, tensor[intArrayOf(1, 2, 3)])
+    if(device != Device.CPU){
+        val normalCpu = randNormal(intArrayOf(2, 3))
+        val normalGpu = normalCpu.copyToDevice(device)
+        assertTrue(normalCpu.copyToArray() contentEquals normalGpu.copyToArray())
+
+        val uniformGpu = randUniform(intArrayOf(3,2),device)
+        val uniformCpu = uniformGpu.copyToDevice(Device.CPU)
+        assertTrue(uniformGpu.copyToArray() contentEquals uniformCpu.copyToArray())
+    }
+}
+
+internal inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensorOverField<T>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.testingRequiresGrad(): Unit {
+    val tensor = randNormal(intArrayOf(3))
+    assertTrue(!tensor.requiresGrad)
+    tensor.requiresGrad = true
+    assertTrue(tensor.requiresGrad)
+    tensor.requiresGrad = false
+    assertTrue(!tensor.requiresGrad)
+    tensor.requiresGrad = true
+    val detachedTensor = tensor.detachFromGraph()
+    assertTrue(!detachedTensor.requiresGrad)
+}
+
+internal inline fun <TorchTensorType : TorchTensor<Int>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<Int, IntArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingViewWithNoCopy(device: Device = Device.CPU) {
+    val tensor = copyFromArray(intArrayOf(1, 2, 3, 4, 5, 6), shape = intArrayOf(6), device)
+    val viewTensor = tensor.view(intArrayOf(2, 3))
+    assertTrue(viewTensor.shape contentEquals intArrayOf(2, 3))
+    viewTensor[intArrayOf(0, 0)] = 10
+    assertEquals(tensor[intArrayOf(0)], 10)
+}
+
+

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/TestTorchTensorAlgebra.kt

@@ -0,0 +1,133 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.linear.RealMatrixContext
+import space.kscience.kmath.linear.Matrix
+import space.kscience.kmath.operations.invoke
+import kotlin.math.*
+import kotlin.test.*
+
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingScalarProduct(device: Device = Device.CPU): Unit {
+    val lhs = randUniform(shape = intArrayOf(3), device = device)
+    val rhs = randUniform(shape = intArrayOf(3), device = device)
+    val product = lhs dot rhs
+    var expected = 0.0
+    lhs.elements().forEach {
+        expected += it.second * rhs[it.first]
+    }
+    assertTrue(abs(expected - product.value()) < TOLERANCE)
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingMatrixMultiplication(device: Device = Device.CPU): Unit {
+    setSeed(SEED)
+
+    val lhsTensor = randNormal(shape = intArrayOf(3, 3), device = device)
+    val rhsTensor = randNormal(shape = intArrayOf(3, 3), device = device)
+    val product = lhsTensor dot rhsTensor
+
+    val expected: Matrix<Double> = RealMatrixContext {
+        val lhs = produce(3, 3) { i, j -> lhsTensor[intArrayOf(i, j)] }
+        val rhs = produce(3, 3) { i, j -> rhsTensor[intArrayOf(i, j)] }
+        lhs dot rhs
+    }
+
+    val lhsTensorCopy = lhsTensor.copy()
+    val rhsTensorCopy = rhsTensor.copy()
+
+    lhsTensorCopy dotAssign rhsTensor
+    lhsTensor dotRightAssign rhsTensorCopy
+
+    var error = 0.0
+    product.elements().forEach {
+        error += abs(expected[it.first] - it.second) +
+                abs(expected[it.first] - lhsTensorCopy[it.first]) +
+                abs(expected[it.first] - rhsTensorCopy[it.first])
+    }
+    assertTrue(error < TOLERANCE)
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingLinearStructure(device: Device = Device.CPU): Unit {
+
+    val shape = intArrayOf(3)
+    val tensorA = full(value = -4.5, shape = shape, device = device)
+    val tensorB = full(value = 10.9, shape = shape, device = device)
+    val tensorC = full(value = 789.3, shape = shape, device = device)
+    val tensorD = full(value = -72.9, shape = shape, device = device)
+    val tensorE = full(value = 553.1, shape = shape, device = device)
+    val result = 15.8 * tensorA - 1.5 * tensorB * (-tensorD) + 0.02 * tensorC / tensorE - 39.4
+    val expected = copyFromArray(
+        array = (1..3).map {
+            15.8 * (-4.5) - 1.5 * 10.9 * 72.9 + 0.02 * 789.3 / 553.1 - 39.4
+        }
+            .toDoubleArray(),
+        shape = shape,
+        device = device
+    )
+
+    val assignResult = full(value = 0.0, shape = shape, device = device)
+    tensorA *= 15.8
+    tensorB *= 1.5
+    tensorB *= -tensorD
+    tensorC *= 0.02
+    tensorC /= tensorE
+    assignResult += tensorA
+    assignResult -= tensorB
+    assignResult += tensorC
+    assignResult += -39.4
+
+    val error = (expected - result).abs().sum().value() +
+            (expected - assignResult).abs().sum().value()
+    assertTrue(error < TOLERANCE)
+
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingTensorTransformations(device: Device = Device.CPU): Unit {
+    setSeed(SEED)
+    val tensor = randNormal(shape = intArrayOf(3, 3), device = device)
+    val result = tensor.exp().log()
+    val assignResult = tensor.copy()
+    assignResult.transposeAssign(0, 1)
+    assignResult.expAssign()
+    assignResult.logAssign()
+    assignResult.transposeAssign(0, 1)
+    val error = tensor - result
+    error.absAssign()
+    error.sumAssign()
+    error += (tensor - assignResult).abs().sum()
+    assertTrue(error.value() < TOLERANCE)
+
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingBatchedSVD(device: Device = Device.CPU): Unit {
+    val tensor = randNormal(shape = intArrayOf(7, 5, 3), device = device)
+    val (tensorU, tensorS, tensorV) = tensor.svd()
+    val error = tensor - (tensorU dot (diagonalEmbedding(tensorS) dot tensorV.transpose(-2, -1)))
+    assertTrue(error.abs().sum().value() < TOLERANCE)
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Double>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Double, DoubleArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingBatchedSymEig(device: Device = Device.CPU): Unit {
+    val tensor = randNormal(shape = intArrayOf(5, 5), device = device)
+    val tensorSigma = tensor + tensor.transpose(-2, -1)
+    val (tensorS, tensorV) = tensorSigma.symEig()
+    val error = tensorSigma - (tensorV dot (diagonalEmbedding(tensorS) dot tensorV.transpose(-2, -1)))
+    assertTrue(error.abs().sum().value() < TOLERANCE)
+}
+
+
+
+
+

kmath-torch/src/commonTest/kotlin/space/kscience/kmath/torch/TestUtils.kt

@@ -0,0 +1,39 @@
+@file:Suppress("NOTHING_TO_INLINE")
+
+package space.kscience.kmath.torch
+
+import kotlin.test.assertEquals
+
+internal val SEED = 987654
+internal val TOLERANCE = 1e-6
+
+internal inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.withCuda(block: TorchTensorAlgebraType.(Device) -> Unit): Unit {
+    this.block(Device.CPU)
+    if (cudaAvailable()) this.block(Device.CUDA(0))
+}
+
+internal inline fun <T, PrimitiveArrayType, TorchTensorType : TorchTensor<T>,
+        TorchTensorAlgebraType : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType>>
+        TorchTensorAlgebraType.testingSetNumThreads(): Unit {
+    val numThreads = 2
+    setNumThreads(numThreads)
+    assertEquals(numThreads, getNumThreads())
+}
+
+internal inline fun <TorchTensorType : TorchTensorOverField<Float>,
+        TorchTensorAlgebraType : TorchTensorPartialDivisionAlgebra<Float, FloatArray, TorchTensorType>>
+        TorchTensorAlgebraType.testingSetSeed(device: Device = Device.CPU): Unit {
+    setSeed(SEED)
+    val integral = randIntegral(0, 100, IntArray(0), device = device).value()
+    val normal = randNormal(IntArray(0), device = device).value()
+    val uniform = randUniform(IntArray(0), device = device).value()
+    setSeed(SEED)
+    val nextIntegral = randIntegral(0, 100, IntArray(0), device = device).value()
+    val nextNormal = randNormal(IntArray(0), device = device).value()
+    val nextUniform = randUniform(IntArray(0), device = device).value()
+    assertEquals(normal, nextNormal)
+    assertEquals(uniform, nextUniform)
+    assertEquals(integral, nextIntegral)
+}

kmath-torch/src/cppMain/CMakeLists.txt

@@ -0,0 +1,34 @@
+cmake_minimum_required(VERSION 3.12)
+
+project(CTorch LANGUAGES C CXX)
+
+# Require C++17
+set(CMAKE_CXX_STANDARD 17)
+
+# Build configuration
+if(NOT CMAKE_BUILD_TYPE)
+  set(CMAKE_BUILD_TYPE Release CACHE STRING "Build type" FORCE)
+endif()
+message(STATUS "Build type: ${CMAKE_BUILD_TYPE}")
+
+find_package(Torch REQUIRED)
+find_package(JNI REQUIRED)
+
+add_library(ctorch SHARED src/ctorch.cc)
+target_include_directories(ctorch PRIVATE include)
+target_link_libraries(ctorch PRIVATE torch)
+target_compile_options(ctorch PRIVATE -Wall -Wextra -Wpedantic -O3 -fPIC)
+
+add_library(jtorch SHARED src/jtorch.cc)
+target_include_directories(jtorch PRIVATE include ${JNI_INCLUDE_DIRS})
+target_link_libraries(jtorch PRIVATE torch)
+target_compile_options(jtorch PRIVATE -Wall -Wextra -Wpedantic -O3 -fPIC)
+
+include(GNUInstallDirs)
+
+set_target_properties(ctorch PROPERTIES PUBLIC_HEADER include/ctorch.h)
+install(TARGETS ctorch
+    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+    PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+
+install(TARGETS jtorch LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})

kmath-torch/src/cppMain/include/ctorch.h

@@ -0,0 +1,160 @@
+#ifndef CTORCH
+#define CTORCH
+
+#include <stdbool.h>
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+
+    typedef void *TorchTensorHandle;
+
+    int get_num_threads();
+
+    void set_num_threads(int num_threads);
+
+    bool cuda_is_available();
+
+    void set_seed(int seed);
+
+    TorchTensorHandle empty_tensor();
+
+    TorchTensorHandle from_blob_double(double *data, int *shape, int dim, int device, bool copy);
+    TorchTensorHandle from_blob_float(float *data, int *shape, int dim, int device, bool copy);
+    TorchTensorHandle from_blob_long(long *data, int *shape, int dim, int device, bool copy);
+    TorchTensorHandle from_blob_int(int *data, int *shape, int dim, int device, bool copy);
+    TorchTensorHandle copy_tensor(TorchTensorHandle tensor_handle);
+    TorchTensorHandle copy_to_device(TorchTensorHandle tensor_handle, int device);
+    TorchTensorHandle copy_to_double(TorchTensorHandle tensor_handle);
+    TorchTensorHandle copy_to_float(TorchTensorHandle tensor_handle);
+    TorchTensorHandle copy_to_long(TorchTensorHandle tensor_handle);
+    TorchTensorHandle copy_to_int(TorchTensorHandle tensor_handle);
+    void swap_tensors(TorchTensorHandle lhs_handle, TorchTensorHandle rhs_handle);
+    TorchTensorHandle view_tensor(TorchTensorHandle tensor_handle, int *shape, int dim);
+
+    char *tensor_to_string(TorchTensorHandle tensor_handle);
+    void dispose_char(char *ptr);
+    void dispose_tensor(TorchTensorHandle tensor_handle);
+
+    int get_dim(TorchTensorHandle tensor_handle);
+    int get_numel(TorchTensorHandle tensor_handle);
+    int get_shape_at(TorchTensorHandle tensor_handle, int d);
+    int get_stride_at(TorchTensorHandle tensor_handle, int d);
+    int get_device(TorchTensorHandle tensor_handle);
+
+    double *get_data_double(TorchTensorHandle tensor_handle);
+    float *get_data_float(TorchTensorHandle tensor_handle);
+    long *get_data_long(TorchTensorHandle tensor_handle);
+    int *get_data_int(TorchTensorHandle tensor_handle);
+
+    double get_item_double(TorchTensorHandle tensor_handle);
+    float get_item_float(TorchTensorHandle tensor_handle);
+    long get_item_long(TorchTensorHandle tensor_handle);
+    int get_item_int(TorchTensorHandle tensor_handle);
+
+    double get_double(TorchTensorHandle tensor_handle, int *index);
+    float get_float(TorchTensorHandle tensor_handle, int *index);
+    long get_long(TorchTensorHandle tensor_handle, int *index);
+    int get_int(TorchTensorHandle tensor_handle, int *index);
+    void set_double(TorchTensorHandle tensor_handle, int *index, double value);
+    void set_float(TorchTensorHandle tensor_handle, int *index, float value);
+    void set_long(TorchTensorHandle tensor_handle, int *index, long value);
+    void set_int(TorchTensorHandle tensor_handle, int *index, int value);
+
+    TorchTensorHandle rand_double(int *shape, int shape_size, int device);
+    TorchTensorHandle randn_double(int *shape, int shape_size, int device);
+    TorchTensorHandle rand_float(int *shape, int shape_size, int device);
+    TorchTensorHandle randn_float(int *shape, int shape_size, int device);
+
+    TorchTensorHandle randint_double(long low, long high, int *shape, int shape_size, int device);
+    TorchTensorHandle randint_float(long low, long high, int *shape, int shape_size, int device);
+    TorchTensorHandle randint_long(long low, long high, int *shape, int shape_size, int device);
+    TorchTensorHandle randint_int(long low, long high, int *shape, int shape_size, int device);
+
+    TorchTensorHandle rand_like(TorchTensorHandle tensor_handle);
+    void rand_like_assign(TorchTensorHandle tensor_handle);
+    TorchTensorHandle randn_like(TorchTensorHandle tensor_handle);
+    void randn_like_assign(TorchTensorHandle tensor_handle);
+    TorchTensorHandle randint_like(TorchTensorHandle tensor_handle, long low, long high);
+    void randint_like_assign(TorchTensorHandle tensor_handle, long low, long high);
+  
+
+    TorchTensorHandle full_double(double value, int *shape, int shape_size, int device);
+    TorchTensorHandle full_float(float value, int *shape, int shape_size, int device);
+    TorchTensorHandle full_long(long value, int *shape, int shape_size, int device);
+    TorchTensorHandle full_int(int value, int *shape, int shape_size, int device);
+
+    TorchTensorHandle times_double(double value, TorchTensorHandle other);
+    TorchTensorHandle times_float(float value, TorchTensorHandle other);
+    TorchTensorHandle times_long(long value, TorchTensorHandle other);
+    TorchTensorHandle times_int(int value, TorchTensorHandle other);
+
+    void times_double_assign(double value, TorchTensorHandle other);
+    void times_float_assign(float value, TorchTensorHandle other);
+    void times_long_assign(long value, TorchTensorHandle other);
+    void times_int_assign(int value, TorchTensorHandle other);
+
+    TorchTensorHandle plus_double(double value, TorchTensorHandle other);
+    TorchTensorHandle plus_float(float value, TorchTensorHandle other);
+    TorchTensorHandle plus_long(long value, TorchTensorHandle other);
+    TorchTensorHandle plus_int(int value, TorchTensorHandle other);
+
+    void plus_double_assign(double value, TorchTensorHandle other);
+    void plus_float_assign(float value, TorchTensorHandle other);
+    void plus_long_assign(long value, TorchTensorHandle other);
+    void plus_int_assign(int value, TorchTensorHandle other);
+
+    TorchTensorHandle times_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void times_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    TorchTensorHandle div_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void div_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    TorchTensorHandle plus_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void plus_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    TorchTensorHandle minus_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void minus_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    TorchTensorHandle unary_minus(TorchTensorHandle tensor_handle);
+
+    TorchTensorHandle abs_tensor(TorchTensorHandle tensor_handle);
+    void abs_tensor_assign(TorchTensorHandle tensor_handle);
+
+    TorchTensorHandle transpose_tensor(TorchTensorHandle tensor_handle, int i, int j);
+    void transpose_tensor_assign(TorchTensorHandle tensor_handle, int i, int j);
+
+    TorchTensorHandle exp_tensor(TorchTensorHandle tensor_handle);
+    void exp_tensor_assign(TorchTensorHandle tensor_handle);
+
+    TorchTensorHandle log_tensor(TorchTensorHandle tensor_handle);
+    void log_tensor_assign(TorchTensorHandle tensor_handle);
+
+    TorchTensorHandle sum_tensor(TorchTensorHandle tensor_handle);
+    void sum_tensor_assign(TorchTensorHandle tensor_handle);
+
+    TorchTensorHandle matmul(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void matmul_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+    void matmul_right_assign(TorchTensorHandle lhs, TorchTensorHandle rhs);
+
+    TorchTensorHandle diag_embed(TorchTensorHandle diags_handle, int offset, int dim1, int dim2);
+
+    void svd_tensor(TorchTensorHandle tensor_handle,
+                    TorchTensorHandle U_handle,
+                    TorchTensorHandle S_handle,
+                    TorchTensorHandle V_handle);
+
+    void symeig_tensor(TorchTensorHandle tensor_handle,
+                       TorchTensorHandle S_handle,
+                       TorchTensorHandle V_handle,
+                       bool eigenvectors);
+
+    bool requires_grad(TorchTensorHandle tensor_handle);
+    void requires_grad_(TorchTensorHandle tensor_handle, bool status);
+    TorchTensorHandle detach_from_graph(TorchTensorHandle tensor_handle);
+    TorchTensorHandle autograd_tensor(TorchTensorHandle value, TorchTensorHandle variable, bool retain_graph);
+    TorchTensorHandle autohess_tensor(TorchTensorHandle value, TorchTensorHandle variable);
+    TorchTensorHandle autohess_tensor_given_grad(TorchTensorHandle value, TorchTensorHandle variable, TorchTensorHandle gradient);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif //CTORCH

kmath-torch/src/cppMain/include/space_kscience_kmath_torch_JTorch.h

@@ -0,0 +1,813 @@
+/* DO NOT EDIT THIS FILE - it is machine generated */
+#include <jni.h>
+/* Header for class space_kscience_kmath_torch_JTorch */
+
+#ifndef _Included_space_kscience_kmath_torch_JTorch
+#define _Included_space_kscience_kmath_torch_JTorch
+#ifdef __cplusplus
+extern "C" {
+#endif
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getNumThreads
+ * Signature: ()I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getNumThreads
+  (JNIEnv *, jclass);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setNumThreads
+ * Signature: (I)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setNumThreads
+  (JNIEnv *, jclass, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    cudaIsAvailable
+ * Signature: ()Z
+ */
+JNIEXPORT jboolean JNICALL Java_space_kscience_kmath_torch_JTorch_cudaIsAvailable
+  (JNIEnv *, jclass);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setSeed
+ * Signature: (I)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setSeed
+  (JNIEnv *, jclass, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    emptyTensor
+ * Signature: ()J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_emptyTensor
+  (JNIEnv *, jclass);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fromBlobDouble
+ * Signature: ([D[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobDouble
+  (JNIEnv *, jclass, jdoubleArray, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fromBlobFloat
+ * Signature: ([F[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobFloat
+  (JNIEnv *, jclass, jfloatArray, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fromBlobLong
+ * Signature: ([J[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobLong
+  (JNIEnv *, jclass, jlongArray, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fromBlobInt
+ * Signature: ([I[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobInt
+  (JNIEnv *, jclass, jintArray, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyTensor
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyToDevice
+ * Signature: (JI)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToDevice
+  (JNIEnv *, jclass, jlong, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyToDouble
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToDouble
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyToFloat
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToFloat
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyToLong
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToLong
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    copyToInt
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToInt
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    swapTensors
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_swapTensors
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    viewTensor
+ * Signature: (J[I)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_viewTensor
+  (JNIEnv *, jclass, jlong, jintArray);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    tensorToString
+ * Signature: (J)Ljava/lang/String;
+ */
+JNIEXPORT jstring JNICALL Java_space_kscience_kmath_torch_JTorch_tensorToString
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    disposeTensor
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_disposeTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getDim
+ * Signature: (J)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getDim
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getNumel
+ * Signature: (J)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getNumel
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getShapeAt
+ * Signature: (JI)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getShapeAt
+  (JNIEnv *, jclass, jlong, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getStrideAt
+ * Signature: (JI)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getStrideAt
+  (JNIEnv *, jclass, jlong, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getDevice
+ * Signature: (J)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getDevice
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getItemDouble
+ * Signature: (J)D
+ */
+JNIEXPORT jdouble JNICALL Java_space_kscience_kmath_torch_JTorch_getItemDouble
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getItemFloat
+ * Signature: (J)F
+ */
+JNIEXPORT jfloat JNICALL Java_space_kscience_kmath_torch_JTorch_getItemFloat
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getItemLong
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_getItemLong
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getItemInt
+ * Signature: (J)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getItemInt
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getDouble
+ * Signature: (J[I)D
+ */
+JNIEXPORT jdouble JNICALL Java_space_kscience_kmath_torch_JTorch_getDouble
+  (JNIEnv *, jclass, jlong, jintArray);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getFloat
+ * Signature: (J[I)F
+ */
+JNIEXPORT jfloat JNICALL Java_space_kscience_kmath_torch_JTorch_getFloat
+  (JNIEnv *, jclass, jlong, jintArray);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getLong
+ * Signature: (J[I)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_getLong
+  (JNIEnv *, jclass, jlong, jintArray);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    getInt
+ * Signature: (J[I)I
+ */
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getInt
+  (JNIEnv *, jclass, jlong, jintArray);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setDouble
+ * Signature: (J[ID)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setDouble
+  (JNIEnv *, jclass, jlong, jintArray, jdouble);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setFloat
+ * Signature: (J[IF)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setFloat
+  (JNIEnv *, jclass, jlong, jintArray, jfloat);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setLong
+ * Signature: (J[IJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setLong
+  (JNIEnv *, jclass, jlong, jintArray, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setInt
+ * Signature: (J[II)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setInt
+  (JNIEnv *, jclass, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randDouble
+ * Signature: ([II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randDouble
+  (JNIEnv *, jclass, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randnDouble
+ * Signature: ([II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnDouble
+  (JNIEnv *, jclass, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randFloat
+ * Signature: ([II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randFloat
+  (JNIEnv *, jclass, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randnFloat
+ * Signature: ([II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnFloat
+  (JNIEnv *, jclass, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintDouble
+ * Signature: (JJ[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintDouble
+  (JNIEnv *, jclass, jlong, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintFloat
+ * Signature: (JJ[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintFloat
+  (JNIEnv *, jclass, jlong, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintLong
+ * Signature: (JJ[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintLong
+  (JNIEnv *, jclass, jlong, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintInt
+ * Signature: (JJ[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintInt
+  (JNIEnv *, jclass, jlong, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randLike
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randLike
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randLikeAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randLikeAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randnLike
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnLike
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randnLikeAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randnLikeAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintLike
+ * Signature: (JJJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintLike
+  (JNIEnv *, jclass, jlong, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    randintLikeAssign
+ * Signature: (JJJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randintLikeAssign
+  (JNIEnv *, jclass, jlong, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fullDouble
+ * Signature: (D[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullDouble
+  (JNIEnv *, jclass, jdouble, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fullFloat
+ * Signature: (F[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullFloat
+  (JNIEnv *, jclass, jfloat, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fullLong
+ * Signature: (J[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullLong
+  (JNIEnv *, jclass, jlong, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    fullInt
+ * Signature: (I[II)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullInt
+  (JNIEnv *, jclass, jint, jintArray, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesDouble
+ * Signature: (DJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesDouble
+  (JNIEnv *, jclass, jdouble, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesFloat
+ * Signature: (FJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesFloat
+  (JNIEnv *, jclass, jfloat, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesLong
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesLong
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesInt
+ * Signature: (IJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesInt
+  (JNIEnv *, jclass, jint, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesDoubleAssign
+ * Signature: (DJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesDoubleAssign
+  (JNIEnv *, jclass, jdouble, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesFloatAssign
+ * Signature: (FJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesFloatAssign
+  (JNIEnv *, jclass, jfloat, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesLongAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesLongAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesIntAssign
+ * Signature: (IJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesIntAssign
+  (JNIEnv *, jclass, jint, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusDouble
+ * Signature: (DJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusDouble
+  (JNIEnv *, jclass, jdouble, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusFloat
+ * Signature: (FJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusFloat
+  (JNIEnv *, jclass, jfloat, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusLong
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusLong
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusInt
+ * Signature: (IJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusInt
+  (JNIEnv *, jclass, jint, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusDoubleAssign
+ * Signature: (DJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusDoubleAssign
+  (JNIEnv *, jclass, jdouble, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusFloatAssign
+ * Signature: (FJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusFloatAssign
+  (JNIEnv *, jclass, jfloat, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusLongAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusLongAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusIntAssign
+ * Signature: (IJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusIntAssign
+  (JNIEnv *, jclass, jint, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesTensor
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesTensor
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    timesTensorAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesTensorAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    divTensor
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_divTensor
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    divTensorAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_divTensorAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusTensor
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusTensor
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    plusTensorAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusTensorAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    minusTensor
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_minusTensor
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    minusTensorAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_minusTensorAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    unaryMinus
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_unaryMinus
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    absTensor
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_absTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    absTensorAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_absTensorAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    transposeTensor
+ * Signature: (JII)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_transposeTensor
+  (JNIEnv *, jclass, jlong, jint, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    transposeTensorAssign
+ * Signature: (JII)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_transposeTensorAssign
+  (JNIEnv *, jclass, jlong, jint, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    expTensor
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_expTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    expTensorAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_expTensorAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    logTensor
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_logTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    logTensorAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_logTensorAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    sumTensor
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_sumTensor
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    sumTensorAssign
+ * Signature: (J)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_sumTensorAssign
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    matmul
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_matmul
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    matmulAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_matmulAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    matmulRightAssign
+ * Signature: (JJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_matmulRightAssign
+  (JNIEnv *, jclass, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    diagEmbed
+ * Signature: (JIII)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_diagEmbed
+  (JNIEnv *, jclass, jlong, jint, jint, jint);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    svdTensor
+ * Signature: (JJJJ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_svdTensor
+  (JNIEnv *, jclass, jlong, jlong, jlong, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    symeigTensor
+ * Signature: (JJJZ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_symeigTensor
+  (JNIEnv *, jclass, jlong, jlong, jlong, jboolean);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    requiresGrad
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_space_kscience_kmath_torch_JTorch_requiresGrad
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    setRequiresGrad
+ * Signature: (JZ)V
+ */
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setRequiresGrad
+  (JNIEnv *, jclass, jlong, jboolean);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    detachFromGraph
+ * Signature: (J)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_detachFromGraph
+  (JNIEnv *, jclass, jlong);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    autogradTensor
+ * Signature: (JJZ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_autogradTensor
+  (JNIEnv *, jclass, jlong, jlong, jboolean);
+
+/*
+ * Class:     space_kscience_kmath_torch_JTorch
+ * Method:    autohessTensor
+ * Signature: (JJ)J
+ */
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_autohessTensor
+  (JNIEnv *, jclass, jlong, jlong);
+
+#ifdef __cplusplus
+}
+#endif
+#endif

kmath-torch/src/cppMain/include/utils.hh

@@ -0,0 +1,151 @@
+#include <torch/torch.h>
+
+namespace ctorch
+{
+
+    using TorchTensorHandle = void *;
+
+    template <typename Dtype>
+    inline c10::ScalarType dtype()
+    {
+        return torch::kFloat64;
+    }
+
+    template <>
+    inline c10::ScalarType dtype<float>()
+    {
+        return torch::kFloat32;
+    }
+
+    template <>
+    inline c10::ScalarType dtype<long>()
+    {
+        return torch::kInt64;
+    }
+
+    template <>
+    inline c10::ScalarType dtype<int>()
+    {
+        return torch::kInt32;
+    }
+
+    template <typename Handle>
+    inline torch::Tensor &cast(const Handle &tensor_handle)
+    {
+        return *static_cast<torch::Tensor *>((TorchTensorHandle)tensor_handle);
+    }
+
+    template <typename Handle>
+    inline void dispose_tensor(const Handle &tensor_handle)
+    {
+        delete static_cast<torch::Tensor *>((TorchTensorHandle)tensor_handle);
+    }
+
+    inline std::string tensor_to_string(const torch::Tensor &tensor)
+    {
+        std::stringstream bufrep;
+        bufrep << tensor;
+        return bufrep.str();
+    }
+
+    inline char *tensor_to_char(const torch::Tensor &tensor)
+    {
+        auto rep = tensor_to_string(tensor);
+        char *crep = (char *)malloc(rep.length() + 1);
+        std::strcpy(crep, rep.c_str());
+        return crep;
+    }
+
+    inline int device_to_int(const torch::Tensor &tensor)
+    {
+        return (tensor.device().type() == torch::kCPU) ? 0 : 1 + tensor.device().index();
+    }
+
+    inline torch::Device int_to_device(int device_int)
+    {
+        return (device_int == 0) ? torch::kCPU : torch::Device(torch::kCUDA, device_int - 1);
+    }
+
+    inline std::vector<int64_t> to_vec_int(int *arr, int arr_size)
+    {
+        auto vec = std::vector<int64_t>(arr_size);
+        vec.assign(arr, arr + arr_size);
+        return vec;
+    }
+
+    inline std::vector<at::indexing::TensorIndex> to_index(int *arr, int arr_size)
+    {
+        std::vector<at::indexing::TensorIndex> index;
+        for (int i = 0; i < arr_size; i++)
+        {
+            index.emplace_back(arr[i]);
+        }
+        return index;
+    }
+
+    template <typename Dtype>
+    inline torch::Tensor from_blob(Dtype *data, const std::vector<int64_t> &shape, torch::Device device, bool copy)
+    {
+        return torch::from_blob(data, shape, dtype<Dtype>()).to(torch::TensorOptions().layout(torch::kStrided).device(device), false, copy);
+    }
+
+    template <typename NumType>
+    inline NumType get(const torch::Tensor &tensor, int *index)
+    {
+        return tensor.index(to_index(index, tensor.dim())).item<NumType>();
+    }
+
+    template <typename NumType>
+    inline void set(const torch::Tensor &tensor, int *index, NumType value)
+    {
+        tensor.index(to_index(index, tensor.dim())) = value;
+    }
+
+    template <typename Dtype>
+    inline torch::Tensor randn(const std::vector<int64_t> &shape, torch::Device device)
+    {
+        return torch::randn(shape, torch::TensorOptions().dtype(dtype<Dtype>()).layout(torch::kStrided).device(device));
+    }
+
+    template <typename Dtype>
+    inline torch::Tensor rand(const std::vector<int64_t> &shape, torch::Device device)
+    {
+        return torch::rand(shape, torch::TensorOptions().dtype(dtype<Dtype>()).layout(torch::kStrided).device(device));
+    }
+
+    template <typename Dtype>
+    inline torch::Tensor randint(long low, long high, const std::vector<int64_t> &shape, torch::Device device)
+    {
+        return torch::randint(low, high, shape, torch::TensorOptions().dtype(dtype<Dtype>()).layout(torch::kStrided).device(device));
+    }
+
+    template <typename Dtype>
+    inline torch::Tensor full(Dtype value, const std::vector<int64_t> &shape, torch::Device device)
+    {
+        return torch::full(shape, value, torch::TensorOptions().dtype(dtype<Dtype>()).layout(torch::kStrided).device(device));
+    }
+
+    inline torch::Tensor hessian(const torch::Tensor &value, const torch::Tensor &variable)
+    {
+        auto nelem = variable.numel();
+        auto hess = value.new_zeros({nelem, nelem});
+        auto grad = torch::autograd::grad({value}, {variable}, {}, torch::nullopt, true)[0].view(nelem);
+        int i = 0;
+        for (int j = 0; j < nelem; j++)
+        {
+            auto row = grad[j].requires_grad()
+                           ? torch::autograd::grad({grad[i]}, {variable}, {}, true, true, true)[0].view(nelem).slice(0, j, nelem)
+                           : grad[j].new_zeros(nelem - j);
+            hess[i].slice(0, i, nelem).add_(row.type_as(hess));
+            i++;
+        }
+        auto ndim = variable.dim();
+        auto sizes = variable.sizes().data();
+        auto shape = std::vector<int64_t>(ndim);
+        shape.assign(sizes, sizes + ndim);
+        shape.reserve(2 * ndim);
+        std::copy_n(shape.begin(), ndim, std::back_inserter(shape));
+        return (hess + torch::triu(hess, 1).t()).view(shape);
+    }
+
+} // namespace ctorch

kmath-torch/src/cppMain/src/ctorch.cc

@@ -0,0 +1,465 @@
+#include <torch/torch.h>
+#include <iostream>
+#include <stdlib.h>
+
+#include "ctorch.h"
+#include "utils.hh"
+
+int get_num_threads()
+{
+  return torch::get_num_threads();
+}
+
+void set_num_threads(int num_threads)
+{
+  torch::set_num_threads(num_threads);
+}
+
+bool cuda_is_available()
+{
+  return torch::cuda::is_available();
+}
+
+void set_seed(int seed)
+{
+  torch::manual_seed(seed);
+}
+
+TorchTensorHandle empty_tensor()
+{
+  return new torch::Tensor;
+}
+
+TorchTensorHandle from_blob_double(double *data, int *shape, int dim, int device, bool copy)
+{
+  return new torch::Tensor(ctorch::from_blob<double>(data, ctorch::to_vec_int(shape, dim), ctorch::int_to_device(device), copy));
+}
+TorchTensorHandle from_blob_float(float *data, int *shape, int dim, int device, bool copy)
+{
+  return new torch::Tensor(ctorch::from_blob<float>(data, ctorch::to_vec_int(shape, dim), ctorch::int_to_device(device), copy));
+}
+TorchTensorHandle from_blob_long(long *data, int *shape, int dim, int device, bool copy)
+{
+  return new torch::Tensor(ctorch::from_blob<long>(data, ctorch::to_vec_int(shape, dim), ctorch::int_to_device(device), copy));
+}
+TorchTensorHandle from_blob_int(int *data, int *shape, int dim, int device, bool copy)
+{
+  return new torch::Tensor(ctorch::from_blob<int>(data, ctorch::to_vec_int(shape, dim), ctorch::int_to_device(device), copy));
+}
+TorchTensorHandle copy_tensor(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).clone());
+}
+TorchTensorHandle copy_to_device(TorchTensorHandle tensor_handle, int device)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::int_to_device(device), false, true));
+}
+TorchTensorHandle copy_to_double(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<double>(), false, true));
+}
+TorchTensorHandle copy_to_float(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<float>(), false, true));
+}
+TorchTensorHandle copy_to_long(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<long>(), false, true));
+}
+TorchTensorHandle copy_to_int(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<int>(), false, true));
+}
+void swap_tensors(TorchTensorHandle lhs_handle, TorchTensorHandle rhs_handle)
+{
+  std::swap(ctorch::cast(lhs_handle), ctorch::cast(rhs_handle));
+}
+TorchTensorHandle view_tensor(TorchTensorHandle tensor_handle, int *shape, int dim)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).view(ctorch::to_vec_int(shape, dim)));
+}
+
+char *tensor_to_string(TorchTensorHandle tensor_handle)
+{
+  return ctorch::tensor_to_char(ctorch::cast(tensor_handle));
+}
+void dispose_char(char *ptr)
+{
+  free(ptr);
+}
+void dispose_tensor(TorchTensorHandle tensor_handle)
+{
+  ctorch::dispose_tensor(tensor_handle);
+}
+
+int get_dim(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).dim();
+}
+int get_numel(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).numel();
+}
+int get_shape_at(TorchTensorHandle tensor_handle, int d)
+{
+  return ctorch::cast(tensor_handle).size(d);
+}
+int get_stride_at(TorchTensorHandle tensor_handle, int d)
+{
+  return ctorch::cast(tensor_handle).stride(d);
+}
+int get_device(TorchTensorHandle tensor_handle)
+{
+  return ctorch::device_to_int(ctorch::cast(tensor_handle));
+}
+
+double *get_data_double(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).data_ptr<double>();
+}
+float *get_data_float(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).data_ptr<float>();
+}
+long *get_data_long(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).data_ptr<long>();
+}
+int *get_data_int(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).data_ptr<int>();
+}
+
+double get_item_double(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).item<double>();
+}
+float get_item_float(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).item<float>();
+}
+long get_item_long(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).item<long>();
+}
+int get_item_int(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).item<int>();
+}
+
+double get_double(TorchTensorHandle tensor_handle, int *index)
+{
+  return ctorch::get<double>(ctorch::cast(tensor_handle), index);
+}
+float get_float(TorchTensorHandle tensor_handle, int *index)
+{
+  return ctorch::get<float>(ctorch::cast(tensor_handle), index);
+}
+long get_long(TorchTensorHandle tensor_handle, int *index)
+{
+  return ctorch::get<long>(ctorch::cast(tensor_handle), index);
+}
+int get_int(TorchTensorHandle tensor_handle, int *index)
+{
+  return ctorch::get<int>(ctorch::cast(tensor_handle), index);
+}
+void set_double(TorchTensorHandle tensor_handle, int *index, double value)
+{
+  ctorch::set<double>(ctorch::cast(tensor_handle), index, value);
+}
+void set_float(TorchTensorHandle tensor_handle, int *index, float value)
+{
+  ctorch::set<float>(ctorch::cast(tensor_handle), index, value);
+}
+void set_long(TorchTensorHandle tensor_handle, int *index, long value)
+{
+  ctorch::set<long>(ctorch::cast(tensor_handle), index, value);
+}
+void set_int(TorchTensorHandle tensor_handle, int *index, int value)
+{
+  ctorch::set<int>(ctorch::cast(tensor_handle), index, value);
+}
+
+TorchTensorHandle rand_double(int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::rand<double>(ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle randn_double(int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randn<double>(ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle rand_float(int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::rand<float>(ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle randn_float(int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randn<float>(ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+
+TorchTensorHandle randint_double(long low, long high, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randint<double>(low, high, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle randint_float(long low, long high, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randint<float>(low, high, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle randint_long(long low, long high, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randint<long>(low, high, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle randint_int(long low, long high, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::randint<int>(low, high, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+
+TorchTensorHandle rand_like(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(torch::rand_like(ctorch::cast(tensor_handle)));
+}
+void rand_like_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = torch::rand_like(ctorch::cast(tensor_handle));
+}
+TorchTensorHandle randn_like(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(torch::randn_like(ctorch::cast(tensor_handle)));
+}
+void randn_like_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = torch::randn_like(ctorch::cast(tensor_handle));
+}
+TorchTensorHandle randint_like(TorchTensorHandle tensor_handle, long low, long high)
+{
+  return new torch::Tensor(torch::randint_like(ctorch::cast(tensor_handle), low, high));
+}
+void randint_like_assign(TorchTensorHandle tensor_handle, long low, long high)
+{
+  ctorch::cast(tensor_handle) = torch::randint_like(ctorch::cast(tensor_handle), low, high);
+}
+
+TorchTensorHandle full_double(double value, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::full<double>(value, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle full_float(float value, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::full<float>(value, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle full_long(long value, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::full<long>(value, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+TorchTensorHandle full_int(int value, int *shape, int shape_size, int device)
+{
+  return new torch::Tensor(ctorch::full<int>(value, ctorch::to_vec_int(shape, shape_size), ctorch::int_to_device(device)));
+}
+
+TorchTensorHandle times_double(double value, TorchTensorHandle other)
+{
+  return new torch::Tensor(value * ctorch::cast(other));
+}
+TorchTensorHandle times_float(float value, TorchTensorHandle other)
+{
+  return new torch::Tensor(value * ctorch::cast(other));
+}
+TorchTensorHandle times_long(long value, TorchTensorHandle other)
+{
+  return new torch::Tensor(value * ctorch::cast(other));
+}
+TorchTensorHandle times_int(int value, TorchTensorHandle other)
+{
+  return new torch::Tensor(value * ctorch::cast(other));
+}
+void times_double_assign(double value, TorchTensorHandle other)
+{
+  ctorch::cast(other) *= value;
+}
+void times_float_assign(float value, TorchTensorHandle other)
+{
+  ctorch::cast(other) *= value;
+}
+void times_long_assign(long value, TorchTensorHandle other)
+{
+  ctorch::cast(other) *= value;
+}
+void times_int_assign(int value, TorchTensorHandle other)
+{
+  ctorch::cast(other) *= value;
+}
+
+TorchTensorHandle plus_double(double value, TorchTensorHandle other)
+{
+  return new torch::Tensor(ctorch::cast(other) + value);
+}
+TorchTensorHandle plus_float(float value, TorchTensorHandle other)
+{
+  return new torch::Tensor(ctorch::cast(other) + value);
+}
+TorchTensorHandle plus_long(long value, TorchTensorHandle other)
+{
+  return new torch::Tensor(ctorch::cast(other) + value);
+}
+TorchTensorHandle plus_int(int value, TorchTensorHandle other)
+{
+  return new torch::Tensor(ctorch::cast(other) + value);
+}
+void plus_double_assign(double value, TorchTensorHandle other)
+{
+  ctorch::cast(other) += value;
+}
+void plus_float_assign(float value, TorchTensorHandle other)
+{
+  ctorch::cast(other) += value;
+}
+void plus_long_assign(long value, TorchTensorHandle other)
+{
+  ctorch::cast(other) += value;
+}
+void plus_int_assign(int value, TorchTensorHandle other)
+{
+  ctorch::cast(other) += value;
+}
+
+TorchTensorHandle times_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  return new torch::Tensor(ctorch::cast(lhs) * ctorch::cast(rhs));
+}
+void times_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(lhs) *= ctorch::cast(rhs);
+}
+TorchTensorHandle div_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  return new torch::Tensor(ctorch::cast(lhs) / ctorch::cast(rhs));
+}
+void div_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(lhs) /= ctorch::cast(rhs);
+}
+TorchTensorHandle plus_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  return new torch::Tensor(ctorch::cast(lhs) + ctorch::cast(rhs));
+}
+void plus_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(lhs) += ctorch::cast(rhs);
+}
+TorchTensorHandle minus_tensor(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  return new torch::Tensor(ctorch::cast(lhs) - ctorch::cast(rhs));
+}
+void minus_tensor_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(lhs) -= ctorch::cast(rhs);
+}
+TorchTensorHandle unary_minus(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(-ctorch::cast(tensor_handle));
+}
+
+TorchTensorHandle abs_tensor(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).abs());
+}
+void abs_tensor_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).abs();
+}
+
+TorchTensorHandle transpose_tensor(TorchTensorHandle tensor_handle, int i, int j)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).transpose(i, j));
+}
+void transpose_tensor_assign(TorchTensorHandle tensor_handle, int i, int j)
+{
+  ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).transpose(i, j);
+}
+
+TorchTensorHandle exp_tensor(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).exp());
+}
+void exp_tensor_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).exp();
+}
+
+TorchTensorHandle log_tensor(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).log());
+}
+void log_tensor_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).log();
+}
+
+TorchTensorHandle sum_tensor(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).sum());
+}
+void sum_tensor_assign(TorchTensorHandle tensor_handle)
+{
+  ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).sum();
+}
+
+TorchTensorHandle matmul(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  return new torch::Tensor(torch::matmul(ctorch::cast(lhs), ctorch::cast(rhs)));
+}
+void matmul_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(lhs) = ctorch::cast(lhs).matmul(ctorch::cast(rhs));
+}
+void matmul_right_assign(TorchTensorHandle lhs, TorchTensorHandle rhs)
+{
+  ctorch::cast(rhs) = ctorch::cast(lhs).matmul(ctorch::cast(rhs));
+}
+
+TorchTensorHandle diag_embed(TorchTensorHandle diags_handle, int offset, int dim1, int dim2)
+{
+  return new torch::Tensor(torch::diag_embed(ctorch::cast(diags_handle), offset, dim1, dim2));
+}
+
+void svd_tensor(TorchTensorHandle tensor_handle,
+                TorchTensorHandle U_handle,
+                TorchTensorHandle S_handle,
+                TorchTensorHandle V_handle)
+{
+  auto [U, S, V] = torch::svd(ctorch::cast(tensor_handle));
+  ctorch::cast(U_handle) = U;
+  ctorch::cast(S_handle) = S;
+  ctorch::cast(V_handle) = V;
+}
+
+void symeig_tensor(TorchTensorHandle tensor_handle,
+                   TorchTensorHandle S_handle,
+                   TorchTensorHandle V_handle,
+                   bool eigenvectors)
+{
+  auto [S, V] = torch::symeig(ctorch::cast(tensor_handle), eigenvectors);
+  ctorch::cast(S_handle) = S;
+  ctorch::cast(V_handle) = V;
+}
+
+bool requires_grad(TorchTensorHandle tensor_handle)
+{
+  return ctorch::cast(tensor_handle).requires_grad();
+}
+void requires_grad_(TorchTensorHandle tensor_handle, bool status)
+{
+  ctorch::cast(tensor_handle).requires_grad_(status);
+}
+TorchTensorHandle detach_from_graph(TorchTensorHandle tensor_handle)
+{
+  return new torch::Tensor(ctorch::cast(tensor_handle).detach());
+}
+TorchTensorHandle autograd_tensor(TorchTensorHandle value, TorchTensorHandle variable, bool retain_graph)
+{
+  return new torch::Tensor(torch::autograd::grad({ctorch::cast(value)}, {ctorch::cast(variable)}, {}, retain_graph)[0]);
+}
+TorchTensorHandle autohess_tensor(TorchTensorHandle value, TorchTensorHandle variable)
+{
+  return new torch::Tensor(ctorch::hessian(ctorch::cast(value), ctorch::cast(variable)));
+}

kmath-torch/src/cppMain/src/jtorch.cc

@@ -0,0 +1,572 @@
+#include <torch/torch.h>
+#include <iostream>
+#include <stdlib.h>
+
+#include "space_kscience_kmath_torch_JTorch.h"
+#include "utils.hh"
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getNumThreads(JNIEnv *, jclass)
+{
+    return torch::get_num_threads();
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setNumThreads(JNIEnv *, jclass, jint num_threads)
+{
+    torch::set_num_threads(num_threads);
+}
+
+JNIEXPORT jboolean JNICALL Java_space_kscience_kmath_torch_JTorch_cudaIsAvailable(JNIEnv *, jclass)
+{
+    return torch::cuda::is_available();
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setSeed(JNIEnv *, jclass, jint seed)
+{
+    torch::manual_seed(seed);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_emptyTensor(JNIEnv *, jclass)
+{
+    return (long)new torch::Tensor;
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobDouble(JNIEnv *env, jclass, jdoubleArray data, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::from_blob<double>(
+            env->GetDoubleArrayElements(data, 0),
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device), true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobFloat(JNIEnv *env, jclass, jfloatArray data, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::from_blob<float>(
+            env->GetFloatArrayElements(data, 0),
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device), true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobLong(JNIEnv *env, jclass, jlongArray data, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::from_blob<long>(
+            env->GetLongArrayElements(data, 0),
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device), true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fromBlobInt(JNIEnv *env, jclass, jintArray data, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::from_blob<int>(
+            env->GetIntArrayElements(data, 0),
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device), true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).clone());
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToDevice(JNIEnv *, jclass, jlong tensor_handle, jint device)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::int_to_device(device), false, true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToDouble(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<double>(), false, true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToFloat(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<float>(), false, true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToLong(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<long>(), false, true));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_copyToInt(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).to(ctorch::dtype<int>(), false, true));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_swapTensors(JNIEnv *, jclass, jlong lhs_handle, jlong rhs_handle)
+{
+    std::swap(ctorch::cast(lhs_handle), ctorch::cast(rhs_handle));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_viewTensor(JNIEnv *env, jclass, jlong tensor_handle, jintArray shape)
+{
+    return (long)new torch::Tensor(
+        ctorch::cast(tensor_handle).view(ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape))));
+}
+
+JNIEXPORT jstring JNICALL Java_space_kscience_kmath_torch_JTorch_tensorToString(JNIEnv *env, jclass, jlong tensor_handle)
+{
+    return env->NewStringUTF(ctorch::tensor_to_string(ctorch::cast(tensor_handle)).c_str());
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_disposeTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::dispose_tensor(tensor_handle);
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getDim(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).dim();
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getNumel(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).numel();
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getShapeAt(JNIEnv *, jclass, jlong tensor_handle, jint d)
+{
+    return ctorch::cast(tensor_handle).size(d);
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getStrideAt(JNIEnv *, jclass, jlong tensor_handle, jint d)
+{
+    return ctorch::cast(tensor_handle).stride(d);
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getDevice(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::device_to_int(ctorch::cast(tensor_handle));
+}
+
+JNIEXPORT jdouble JNICALL Java_space_kscience_kmath_torch_JTorch_getItemDouble(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).item<double>();
+}
+
+JNIEXPORT jfloat JNICALL Java_space_kscience_kmath_torch_JTorch_getItemFloat(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).item<float>();
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_getItemLong(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).item<long>();
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getItemInt(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).item<int>();
+}
+
+JNIEXPORT jdouble JNICALL Java_space_kscience_kmath_torch_JTorch_getDouble(JNIEnv *env, jclass, jlong tensor_handle, jintArray index)
+{
+    return ctorch::get<double>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0));
+}
+
+JNIEXPORT jfloat JNICALL Java_space_kscience_kmath_torch_JTorch_getFloat(JNIEnv *env, jclass, jlong tensor_handle, jintArray index)
+{
+    return ctorch::get<float>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_getLong(JNIEnv *env, jclass, jlong tensor_handle, jintArray index)
+{
+    return ctorch::get<long>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0));
+}
+
+JNIEXPORT jint JNICALL Java_space_kscience_kmath_torch_JTorch_getInt(JNIEnv *env, jclass, jlong tensor_handle, jintArray index)
+{
+    return ctorch::get<int>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setDouble(JNIEnv *env, jclass, jlong tensor_handle, jintArray index, jdouble value)
+{
+    ctorch::set<double>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0), value);
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setFloat(JNIEnv *env, jclass, jlong tensor_handle, jintArray index, jfloat value)
+{
+    ctorch::set<float>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0), value);
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setLong(JNIEnv *env, jclass, jlong tensor_handle, jintArray index, jlong value)
+{
+    ctorch::set<long>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0), value);
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setInt(JNIEnv *env, jclass, jlong tensor_handle, jintArray index, jint value)
+{
+    ctorch::set<int>(ctorch::cast(tensor_handle), env->GetIntArrayElements(index, 0), value);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randDouble(JNIEnv *env, jclass, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::rand<double>(
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnDouble(JNIEnv *env, jclass, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randn<double>(
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randFloat(JNIEnv *env, jclass, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::rand<float>(
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnFloat(JNIEnv *env, jclass, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randn<float>(
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintDouble(JNIEnv *env, jclass, jlong low, jlong high, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randint<double>(low, high,
+                                ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+                                ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintFloat(JNIEnv *env, jclass, jlong low, jlong high, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randint<float>(low, high,
+                               ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+                               ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintLong(JNIEnv *env, jclass, jlong low, jlong high, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randint<long>(low, high,
+                              ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+                              ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintInt(JNIEnv *env, jclass, jlong low, jlong high, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::randint<int>(low, high,
+                             ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+                             ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randLike(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(torch::rand_like(ctorch::cast(tensor_handle)));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randLikeAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = torch::rand_like(ctorch::cast(tensor_handle));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randnLike(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(torch::randn_like(ctorch::cast(tensor_handle)));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randnLikeAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = torch::randn_like(ctorch::cast(tensor_handle));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_randintLike(JNIEnv *, jclass, jlong tensor_handle, jlong low, jlong high)
+{
+    return (long)new torch::Tensor(torch::randint_like(ctorch::cast(tensor_handle), low, high));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_randintLikeAssign(JNIEnv *, jclass, jlong tensor_handle, jlong low, jlong high)
+{
+    ctorch::cast(tensor_handle) = torch::randint_like(ctorch::cast(tensor_handle), low, high);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullDouble(JNIEnv *env, jclass, jdouble value, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::full<double>(
+            value,
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullFloat(JNIEnv *env, jclass, jfloat value, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::full<float>(
+            value,
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullLong(JNIEnv *env, jclass, jlong value, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::full<long>(
+            value,
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_fullInt(JNIEnv *env, jclass, jint value, jintArray shape, jint device)
+{
+    return (long)new torch::Tensor(
+        ctorch::full<int>(
+            value,
+            ctorch::to_vec_int(env->GetIntArrayElements(shape, 0), env->GetArrayLength(shape)),
+            ctorch::int_to_device(device)));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesDouble(JNIEnv *, jclass, jdouble value, jlong other)
+{
+    return (long)new torch::Tensor(value * ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesFloat(JNIEnv *, jclass, jfloat value, jlong other)
+{
+    return (long)new torch::Tensor(value * ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesLong(JNIEnv *, jclass, jlong value, jlong other)
+{
+    return (long)new torch::Tensor(value * ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesInt(JNIEnv *, jclass, jint value, jlong other)
+{
+    return (long)new torch::Tensor(value * ctorch::cast(other));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesDoubleAssign(JNIEnv *, jclass, jdouble value, jlong other)
+{
+    ctorch::cast(other) *= value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesFloatAssign(JNIEnv *, jclass, jfloat value, jlong other)
+{
+    ctorch::cast(other) *= value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesLongAssign(JNIEnv *, jclass, jlong value, jlong other)
+{
+    ctorch::cast(other) *= value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesIntAssign(JNIEnv *, jclass, jint value, jlong other)
+{
+    ctorch::cast(other) *= value;
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusDouble(JNIEnv *, jclass, jdouble value, jlong other)
+{
+    return (long)new torch::Tensor(value + ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusFloat(JNIEnv *, jclass, jfloat value, jlong other)
+{
+    return (long)new torch::Tensor(value + ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusLong(JNIEnv *, jclass, jlong value, jlong other)
+{
+    return (long)new torch::Tensor(value + ctorch::cast(other));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusInt(JNIEnv *, jclass, jint value, jlong other)
+{
+    return (long)new torch::Tensor(value + ctorch::cast(other));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusDoubleAssign(JNIEnv *, jclass, jdouble value, jlong other)
+{
+    ctorch::cast(other) += value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusFloatAssign(JNIEnv *, jclass, jfloat value, jlong other)
+{
+    ctorch::cast(other) += value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusLongAssign(JNIEnv *, jclass, jlong value, jlong other)
+{
+    ctorch::cast(other) += value;
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusIntAssign(JNIEnv *, jclass, jint value, jlong other)
+{
+    ctorch::cast(other) += value;
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_timesTensor(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    return (long)new torch::Tensor(ctorch::cast(lhs) * ctorch::cast(rhs));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_timesTensorAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(lhs) *= ctorch::cast(rhs);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_divTensor(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    return (long)new torch::Tensor(ctorch::cast(lhs) / ctorch::cast(rhs));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_divTensorAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(lhs) /= ctorch::cast(rhs);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_plusTensor(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    return (long)new torch::Tensor(ctorch::cast(lhs) + ctorch::cast(rhs));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_plusTensorAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(lhs) += ctorch::cast(rhs);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_minusTensor(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    return (long)new torch::Tensor(ctorch::cast(lhs) - ctorch::cast(rhs));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_minusTensorAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(lhs) -= ctorch::cast(rhs);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_unaryMinus(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(-ctorch::cast(tensor_handle));
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_absTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).abs());
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_absTensorAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).abs();
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_transposeTensor(JNIEnv *, jclass, jlong tensor_handle, jint i, jint j)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).transpose(i, j));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_transposeTensorAssign(JNIEnv *, jclass, jlong tensor_handle, jint i, jint j)
+{
+    ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).transpose(i, j);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_expTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).exp());
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_expTensorAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).exp();
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_logTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).log());
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_logTensorAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).log();
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_sumTensor(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).sum());
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_sumTensorAssign(JNIEnv *, jclass, jlong tensor_handle)
+{
+    ctorch::cast(tensor_handle) = ctorch::cast(tensor_handle).sum();
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_matmul(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    return (long)new torch::Tensor(torch::matmul(ctorch::cast(lhs), ctorch::cast(rhs)));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_matmulAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(lhs) = ctorch::cast(lhs).matmul(ctorch::cast(rhs));
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_matmulRightAssign(JNIEnv *, jclass, jlong lhs, jlong rhs)
+{
+    ctorch::cast(rhs) = ctorch::cast(lhs).matmul(ctorch::cast(rhs));
+}
+
+JNIEXPORT jlong JNICALL
+Java_space_kscience_kmath_torch_JTorch_diagEmbed(JNIEnv *, jclass, jlong diags_handle, jint offset, jint dim1, jint dim2)
+{
+    return (long)new torch::Tensor(torch::diag_embed(ctorch::cast(diags_handle), offset, dim1, dim2));
+}
+
+JNIEXPORT void JNICALL
+Java_space_kscience_kmath_torch_JTorch_svdTensor(JNIEnv *, jclass, jlong tensor_handle, jlong U_handle, jlong S_handle, jlong V_handle)
+{
+    auto [U, S, V] = torch::svd(ctorch::cast(tensor_handle));
+    ctorch::cast(U_handle) = U;
+    ctorch::cast(S_handle) = S;
+    ctorch::cast(V_handle) = V;
+}
+
+JNIEXPORT void JNICALL
+Java_space_kscience_kmath_torch_JTorch_symeigTensor(JNIEnv *, jclass, jlong tensor_handle, jlong S_handle, jlong V_handle, jboolean eigenvectors)
+{
+    auto [S, V] = torch::symeig(ctorch::cast(tensor_handle), eigenvectors);
+    ctorch::cast(S_handle) = S;
+    ctorch::cast(V_handle) = V;
+}
+
+JNIEXPORT jboolean JNICALL Java_space_kscience_kmath_torch_JTorch_requiresGrad(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return ctorch::cast(tensor_handle).requires_grad();
+}
+
+JNIEXPORT void JNICALL Java_space_kscience_kmath_torch_JTorch_setRequiresGrad(JNIEnv *, jclass, jlong tensor_handle, jboolean status)
+{
+    ctorch::cast(tensor_handle).requires_grad_(status);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_detachFromGraph(JNIEnv *, jclass, jlong tensor_handle)
+{
+    return (long)new torch::Tensor(ctorch::cast(tensor_handle).detach());
+}
+
+JNIEXPORT jlong JNICALL
+Java_space_kscience_kmath_torch_JTorch_autogradTensor(JNIEnv *, jclass, jlong value, jlong variable, jboolean retain_graph)
+{
+    return (long)new torch::Tensor(torch::autograd::grad({ctorch::cast(value)}, {ctorch::cast(variable)}, {}, retain_graph)[0]);
+}
+
+JNIEXPORT jlong JNICALL Java_space_kscience_kmath_torch_JTorch_autohessTensor(JNIEnv *, jclass, jlong value, jlong variable)
+{
+    return (long)new torch::Tensor(ctorch::hessian(ctorch::cast(value), ctorch::cast(variable)));
+}

kmath-torch/src/jvmMain/java/space/kscience/kmath/torch/JTorch.java

@@ -0,0 +1,208 @@
+package space.kscience.kmath.torch;
+
+class JTorch {
+
+    static {
+        System.loadLibrary("jtorch");
+    }
+
+    public static native int getNumThreads();
+
+    public static native void setNumThreads(int numThreads);
+
+    public static native boolean cudaIsAvailable();
+
+    public static native void setSeed(int seed);
+
+    public static native long emptyTensor();
+
+    public static native long fromBlobDouble(double[] data, int[] shape, int device);
+
+    public static native long fromBlobFloat(float[] data, int[] shape, int device);
+
+    public static native long fromBlobLong(long[] data, int[] shape, int device);
+
+    public static native long fromBlobInt(int[] data, int[] shape, int device);
+
+    public static native long copyTensor(long tensorHandle);
+
+    public static native long copyToDevice(long tensorHandle, int device);
+
+    public static native long copyToDouble(long tensorHandle);
+
+    public static native long copyToFloat(long tensorHandle);
+
+    public static native long copyToLong(long tensorHandle);
+
+    public static native long copyToInt(long tensorHandle);
+
+    public static native void swapTensors(long lhsHandle, long rhsHandle);
+
+    public static native long viewTensor(long tensorHandle, int[] shape);
+
+    public static native String tensorToString(long tensorHandle);
+
+    public static native void disposeTensor(long tensorHandle);
+
+    public static native int getDim(long tensorHandle);
+
+    public static native int getNumel(long tensorHandle);
+
+    public static native int getShapeAt(long tensorHandle, int d);
+
+    public static native int getStrideAt(long tensorHandle, int d);
+
+    public static native int getDevice(long tensorHandle);
+
+    public static native double getItemDouble(long tensorHandle);
+
+    public static native float getItemFloat(long tensorHandle);
+
+    public static native long getItemLong(long tensorHandle);
+
+    public static native int getItemInt(long tensorHandle);
+
+    public static native double getDouble(long tensorHandle, int[] index);
+
+    public static native float getFloat(long tensorHandle, int[] index);
+
+    public static native long getLong(long tensorHandle, int[] index);
+
+    public static native int getInt(long tensorHandle, int[] index);
+
+    public static native void setDouble(long tensorHandle, int[] index, double value);
+
+    public static native void setFloat(long tensorHandle, int[] index, float value);
+
+    public static native void setLong(long tensorHandle, int[] index, long value);
+
+    public static native void setInt(long tensorHandle, int[] index, int value);
+
+    public static native long randDouble(int[] shape, int device);
+
+    public static native long randnDouble(int[] shape, int device);
+
+    public static native long randFloat(int[] shape, int device);
+
+    public static native long randnFloat(int[] shape, int device);
+
+    public static native long randintDouble(long low, long high, int[] shape, int device);
+
+    public static native long randintFloat(long low, long high, int[] shape, int device);
+
+    public static native long randintLong(long low, long high, int[] shape, int device);
+
+    public static native long randintInt(long low, long high, int[] shape, int device);
+
+    public static native long randLike(long tensorHandle);
+
+    public static native void randLikeAssign(long tensorHandle);
+
+    public static native long randnLike(long tensorHandle);
+
+    public static native void randnLikeAssign(long tensorHandle);
+
+    public static native long randintLike(long tensorHandle, long low, long high);
+
+    public static native void randintLikeAssign(long tensorHandle, long low, long high);
+
+    public static native long fullDouble(double value, int[] shape, int device);
+
+    public static native long fullFloat(float value, int[] shape, int device);
+
+    public static native long fullLong(long value, int[] shape, int device);
+
+    public static native long fullInt(int value, int[] shape, int device);
+
+    public static native long timesDouble(double value, long other);
+
+    public static native long timesFloat(float value, long other);
+
+    public static native long timesLong(long value, long other);
+
+    public static native long timesInt(int value, long other);
+
+    public static native void timesDoubleAssign(double value, long other);
+
+    public static native void timesFloatAssign(float value, long other);
+
+    public static native void timesLongAssign(long value, long other);
+
+    public static native void timesIntAssign(int value, long other);
+
+    public static native long plusDouble(double value, long other);
+
+    public static native long plusFloat(float value, long other);
+
+    public static native long plusLong(long value, long other);
+
+    public static native long plusInt(int value, long other);
+
+    public static native void plusDoubleAssign(double value, long other);
+
+    public static native void plusFloatAssign(float value, long other);
+
+    public static native void plusLongAssign(long value, long other);
+
+    public static native void plusIntAssign(int value, long other);
+
+    public static native long timesTensor(long lhs, long rhs);
+
+    public static native void timesTensorAssign(long lhs, long rhs);
+
+    public static native long divTensor(long lhs, long rhs);
+
+    public static native void divTensorAssign(long lhs, long rhs);
+
+    public static native long plusTensor(long lhs, long rhs);
+
+    public static native void plusTensorAssign(long lhs, long rhs);
+
+    public static native long minusTensor(long lhs, long rhs);
+
+    public static native void minusTensorAssign(long lhs, long rhs);
+
+    public static native long unaryMinus(long tensorHandle);
+
+    public static native long absTensor(long tensorHandle);
+
+    public static native void absTensorAssign(long tensorHandle);
+
+    public static native long transposeTensor(long tensorHandle, int i, int j);
+
+    public static native void transposeTensorAssign(long tensorHandle, int i, int j);
+
+    public static native long expTensor(long tensorHandle);
+
+    public static native void expTensorAssign(long tensorHandle);
+
+    public static native long logTensor(long tensorHandle);
+
+    public static native void logTensorAssign(long tensorHandle);
+
+    public static native long sumTensor(long tensorHandle);
+
+    public static native void sumTensorAssign(long tensorHandle);
+
+    public static native long matmul(long lhs, long rhs);
+
+    public static native void matmulAssign(long lhs, long rhs);
+
+    public static native void matmulRightAssign(long lhs, long rhs);
+
+    public static native long diagEmbed(long diagsHandle, int offset, int dim1, int dim2);
+
+    public static native void svdTensor(long tensorHandle, long Uhandle, long Shandle, long Vhandle);
+
+    public static native void symeigTensor(long tensorHandle, long Shandle, long Vhandle, boolean eigenvectors);
+
+    public static native boolean requiresGrad(long tensorHandle);
+
+    public static native void setRequiresGrad(long tensorHandle, boolean status);
+
+    public static native long detachFromGraph(long tensorHandle);
+
+    public static native long autogradTensor(long value, long variable, boolean retainGraph);
+
+    public static native long autohessTensor(long value, long variable);
+}

kmath-torch/src/jvmMain/kotlin/space/kscience/kmath/torch/TorchTensorAlgebraJVM.kt

@@ -0,0 +1,390 @@
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.memory.DeferScope
+import space.kscience.kmath.memory.withDeferScope
+import space.kscience.kmath.tensors.*
+
+public sealed class TorchTensorAlgebraJVM<
+        T,
+        PrimitiveArrayType,
+        TorchTensorType : TorchTensorJVM<T>> constructor(
+    internal val scope: DeferScope
+) : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType> {
+    override fun getNumThreads(): Int {
+        return JTorch.getNumThreads()
+    }
+
+    override fun setNumThreads(numThreads: Int): Unit {
+        JTorch.setNumThreads(numThreads)
+    }
+
+    override fun cudaAvailable(): Boolean {
+        return JTorch.cudaIsAvailable()
+    }
+
+    override fun setSeed(seed: Int): Unit {
+        JTorch.setSeed(seed)
+    }
+
+    override var checks: Boolean = false
+
+    internal abstract fun wrap(tensorHandle: Long): TorchTensorType
+
+    override operator fun TorchTensorType.times(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(JTorch.timesTensor(this.tensorHandle, other.tensorHandle))
+    }
+
+    override operator fun TorchTensorType.timesAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        JTorch.timesTensorAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.plus(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(JTorch.plusTensor(this.tensorHandle, other.tensorHandle))
+    }
+
+    override operator fun TorchTensorType.plusAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        JTorch.plusTensorAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.minus(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(JTorch.minusTensor(this.tensorHandle, other.tensorHandle))
+    }
+
+    override operator fun TorchTensorType.minusAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        JTorch.minusTensorAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.unaryMinus(): TorchTensorType =
+        wrap(JTorch.unaryMinus(this.tensorHandle))
+
+    override infix fun TorchTensorType.dot(other: TorchTensorType): TorchTensorType {
+        if (checks) checkDotOperation(this, other)
+        return wrap(JTorch.matmul(this.tensorHandle, other.tensorHandle))
+    }
+
+    override infix fun TorchTensorType.dotAssign(other: TorchTensorType): Unit {
+        if (checks) checkDotOperation(this, other)
+        JTorch.matmulAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override infix fun TorchTensorType.dotRightAssign(other: TorchTensorType): Unit {
+        if (checks) checkDotOperation(this, other)
+        JTorch.matmulRightAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override fun diagonalEmbedding(
+        diagonalEntries: TorchTensorType, offset: Int, dim1: Int, dim2: Int
+    ): TorchTensorType =
+        wrap(JTorch.diagEmbed(diagonalEntries.tensorHandle, offset, dim1, dim2))
+
+    override fun TorchTensorType.transpose(i: Int, j: Int): TorchTensorType {
+        if (checks) checkTranspose(this.dimension, i, j)
+        return wrap(JTorch.transposeTensor(tensorHandle, i, j))
+    }
+
+    override fun TorchTensorType.transposeAssign(i: Int, j: Int): Unit {
+        if (checks) checkTranspose(this.dimension, i, j)
+        JTorch.transposeTensorAssign(tensorHandle, i, j)
+    }
+
+    override fun TorchTensorType.view(shape: IntArray): TorchTensorType {
+        if (checks) checkView(this, shape)
+        return wrap(JTorch.viewTensor(this.tensorHandle, shape))
+    }
+
+    override fun TorchTensorType.abs(): TorchTensorType = wrap(JTorch.absTensor(tensorHandle))
+    override fun TorchTensorType.absAssign(): Unit = JTorch.absTensorAssign(tensorHandle)
+
+    override fun TorchTensorType.sum(): TorchTensorType = wrap(JTorch.sumTensor(tensorHandle))
+    override fun TorchTensorType.sumAssign(): Unit = JTorch.sumTensorAssign(tensorHandle)
+
+    override fun TorchTensorType.randIntegral(low: Long, high: Long): TorchTensorType =
+        wrap(JTorch.randintLike(this.tensorHandle, low, high))
+
+    override fun TorchTensorType.randIntegralAssign(low: Long, high: Long): Unit =
+        JTorch.randintLikeAssign(this.tensorHandle, low, high)
+
+    override fun TorchTensorType.copy(): TorchTensorType =
+        wrap(JTorch.copyTensor(this.tensorHandle))
+
+    override fun TorchTensorType.copyToDevice(device: Device): TorchTensorType =
+        wrap(JTorch.copyToDevice(this.tensorHandle, device.toInt()))
+
+    override infix fun TorchTensorType.swap(other: TorchTensorType): Unit =
+        JTorch.swapTensors(this.tensorHandle, other.tensorHandle)
+}
+
+public sealed class TorchTensorPartialDivisionAlgebraJVM<T, PrimitiveArrayType,
+        TorchTensorType : TorchTensorOverFieldJVM<T>>(scope: DeferScope) :
+    TorchTensorAlgebraJVM<T, PrimitiveArrayType, TorchTensorType>(scope),
+    TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType> {
+
+    override operator fun TorchTensorType.div(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(JTorch.divTensor(this.tensorHandle, other.tensorHandle))
+    }
+
+    override operator fun TorchTensorType.divAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        JTorch.divTensorAssign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override fun TorchTensorType.randUniform(): TorchTensorType =
+        wrap(JTorch.randLike(this.tensorHandle))
+
+    override fun TorchTensorType.randUniformAssign(): Unit =
+        JTorch.randLikeAssign(this.tensorHandle)
+
+    override fun TorchTensorType.randNormal(): TorchTensorType =
+        wrap(JTorch.randnLike(this.tensorHandle))
+
+    override fun TorchTensorType.randNormalAssign(): Unit =
+        JTorch.randnLikeAssign(this.tensorHandle)
+
+    override fun TorchTensorType.exp(): TorchTensorType = wrap(JTorch.expTensor(tensorHandle))
+    override fun TorchTensorType.expAssign(): Unit = JTorch.expTensorAssign(tensorHandle)
+    override fun TorchTensorType.log(): TorchTensorType = wrap(JTorch.logTensor(tensorHandle))
+    override fun TorchTensorType.logAssign(): Unit = JTorch.logTensorAssign(tensorHandle)
+
+    override fun TorchTensorType.svd(): Triple<TorchTensorType, TorchTensorType, TorchTensorType> {
+        val U = JTorch.emptyTensor()
+        val V = JTorch.emptyTensor()
+        val S = JTorch.emptyTensor()
+        JTorch.svdTensor(this.tensorHandle, U, S, V)
+        return Triple(wrap(U), wrap(S), wrap(V))
+    }
+
+    override fun TorchTensorType.symEig(eigenvectors: Boolean): Pair<TorchTensorType, TorchTensorType> {
+        val V = JTorch.emptyTensor()
+        val S = JTorch.emptyTensor()
+        JTorch.symeigTensor(this.tensorHandle, S, V, eigenvectors)
+        return Pair(wrap(S), wrap(V))
+    }
+
+    override fun TorchTensorType.grad(variable: TorchTensorType, retainGraph: Boolean): TorchTensorType {
+        if (checks) this.checkIsValue()
+        return wrap(JTorch.autogradTensor(this.tensorHandle, variable.tensorHandle, retainGraph))
+    }
+
+    override infix fun TorchTensorType.hess(variable: TorchTensorType): TorchTensorType {
+        if (checks) this.checkIsValue()
+        return wrap(JTorch.autohessTensor(this.tensorHandle, variable.tensorHandle))
+    }
+
+    override fun TorchTensorType.detachFromGraph(): TorchTensorType =
+        wrap(JTorch.detachFromGraph(this.tensorHandle))
+
+}
+
+public class TorchTensorRealAlgebra(scope: DeferScope) :
+    TorchTensorPartialDivisionAlgebraJVM<Double, DoubleArray, TorchTensorReal>(scope) {
+    override fun wrap(tensorHandle: Long): TorchTensorReal =
+        TorchTensorReal(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorReal.copyToArray(): DoubleArray =
+        this.elements().map { it.second }.toList().toDoubleArray()
+
+    override fun copyFromArray(array: DoubleArray, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(JTorch.fromBlobDouble(array, shape, device.toInt()))
+
+    override fun randNormal(shape: IntArray, device: Device): TorchTensorReal =
+        wrap(JTorch.randnDouble(shape, device.toInt()))
+
+    override fun randUniform(shape: IntArray, device: Device): TorchTensorReal =
+        wrap(JTorch.randDouble(shape, device.toInt()))
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(JTorch.randintDouble(low, high, shape, device.toInt()))
+
+    override operator fun Double.plus(other: TorchTensorReal): TorchTensorReal =
+        wrap(JTorch.plusDouble(this, other.tensorHandle))
+
+    override fun TorchTensorReal.plus(value: Double): TorchTensorReal =
+        wrap(JTorch.plusDouble(value, this.tensorHandle))
+
+    override fun TorchTensorReal.plusAssign(value: Double): Unit =
+        JTorch.plusDoubleAssign(value, this.tensorHandle)
+
+    override operator fun Double.minus(other: TorchTensorReal): TorchTensorReal =
+        wrap(JTorch.plusDouble(-this, other.tensorHandle))
+
+    override fun TorchTensorReal.minus(value: Double): TorchTensorReal =
+        wrap(JTorch.plusDouble(-value, this.tensorHandle))
+
+    override fun TorchTensorReal.minusAssign(value: Double): Unit =
+        JTorch.plusDoubleAssign(-value, this.tensorHandle)
+
+    override operator fun Double.times(other: TorchTensorReal): TorchTensorReal =
+        wrap(JTorch.timesDouble(this, other.tensorHandle))
+
+    override fun TorchTensorReal.times(value: Double): TorchTensorReal =
+        wrap(JTorch.timesDouble(value, this.tensorHandle))
+
+    override fun TorchTensorReal.timesAssign(value: Double): Unit =
+        JTorch.timesDoubleAssign(value, this.tensorHandle)
+
+    override fun full(value: Double, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(JTorch.fullDouble(value, shape, device.toInt()))
+
+}
+
+public class TorchTensorFloatAlgebra(scope: DeferScope) :
+    TorchTensorPartialDivisionAlgebraJVM<Float, FloatArray, TorchTensorFloat>(scope) {
+    override fun wrap(tensorHandle: Long): TorchTensorFloat =
+        TorchTensorFloat(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorFloat.copyToArray(): FloatArray =
+        this.elements().map { it.second }.toList().toFloatArray()
+
+    override fun copyFromArray(array: FloatArray, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(JTorch.fromBlobFloat(array, shape, device.toInt()))
+
+    override fun randNormal(shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(JTorch.randnFloat(shape, device.toInt()))
+
+    override fun randUniform(shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(JTorch.randFloat(shape, device.toInt()))
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(JTorch.randintFloat(low, high, shape, device.toInt()))
+
+    override operator fun Float.plus(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(JTorch.plusFloat(this, other.tensorHandle))
+
+    override fun TorchTensorFloat.plus(value: Float): TorchTensorFloat =
+        wrap(JTorch.plusFloat(value, this.tensorHandle))
+
+    override fun TorchTensorFloat.plusAssign(value: Float): Unit =
+        JTorch.plusFloatAssign(value, this.tensorHandle)
+
+    override operator fun Float.minus(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(JTorch.plusFloat(-this, other.tensorHandle))
+
+    override fun TorchTensorFloat.minus(value: Float): TorchTensorFloat =
+        wrap(JTorch.plusFloat(-value, this.tensorHandle))
+
+    override fun TorchTensorFloat.minusAssign(value: Float): Unit =
+        JTorch.plusFloatAssign(-value, this.tensorHandle)
+
+    override operator fun Float.times(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(JTorch.timesFloat(this, other.tensorHandle))
+
+    override fun TorchTensorFloat.times(value: Float): TorchTensorFloat =
+        wrap(JTorch.timesFloat(value, this.tensorHandle))
+
+    override fun TorchTensorFloat.timesAssign(value: Float): Unit =
+        JTorch.timesFloatAssign(value, this.tensorHandle)
+
+    override fun full(value: Float, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(JTorch.fullFloat(value, shape, device.toInt()))
+
+}
+
+public class TorchTensorLongAlgebra(scope: DeferScope) :
+    TorchTensorAlgebraJVM<Long, LongArray, TorchTensorLong>(scope) {
+    override fun wrap(tensorHandle: Long): TorchTensorLong =
+        TorchTensorLong(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorLong.copyToArray(): LongArray =
+        this.elements().map { it.second }.toList().toLongArray()
+
+    override fun copyFromArray(array: LongArray, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(JTorch.fromBlobLong(array, shape, device.toInt()))
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(JTorch.randintLong(low, high, shape, device.toInt()))
+
+    override operator fun Long.plus(other: TorchTensorLong): TorchTensorLong =
+        wrap(JTorch.plusLong(this, other.tensorHandle))
+
+    override fun TorchTensorLong.plus(value: Long): TorchTensorLong =
+        wrap(JTorch.plusLong(value, this.tensorHandle))
+
+    override fun TorchTensorLong.plusAssign(value: Long): Unit =
+        JTorch.plusLongAssign(value, this.tensorHandle)
+
+    override operator fun Long.minus(other: TorchTensorLong): TorchTensorLong =
+        wrap(JTorch.plusLong(-this, other.tensorHandle))
+
+    override fun TorchTensorLong.minus(value: Long): TorchTensorLong =
+        wrap(JTorch.plusLong(-value, this.tensorHandle))
+
+    override fun TorchTensorLong.minusAssign(value: Long): Unit =
+        JTorch.plusLongAssign(-value, this.tensorHandle)
+
+    override operator fun Long.times(other: TorchTensorLong): TorchTensorLong =
+        wrap(JTorch.timesLong(this, other.tensorHandle))
+
+    override fun TorchTensorLong.times(value: Long): TorchTensorLong =
+        wrap(JTorch.timesLong(value, this.tensorHandle))
+
+    override fun TorchTensorLong.timesAssign(value: Long): Unit =
+        JTorch.timesLongAssign(value, this.tensorHandle)
+
+    override fun full(value: Long, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(JTorch.fullLong(value, shape, device.toInt()))
+}
+
+public class TorchTensorIntAlgebra(scope: DeferScope) :
+    TorchTensorAlgebraJVM<Int, IntArray, TorchTensorInt>(scope) {
+    override fun wrap(tensorHandle: Long): TorchTensorInt =
+        TorchTensorInt(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorInt.copyToArray(): IntArray =
+        this.elements().map { it.second }.toList().toIntArray()
+
+    override fun copyFromArray(array: IntArray, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(JTorch.fromBlobInt(array, shape, device.toInt()))
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(JTorch.randintInt(low, high, shape, device.toInt()))
+
+    override operator fun Int.plus(other: TorchTensorInt): TorchTensorInt =
+        wrap(JTorch.plusInt(this, other.tensorHandle))
+
+    override fun TorchTensorInt.plus(value: Int): TorchTensorInt =
+        wrap(JTorch.plusInt(value, this.tensorHandle))
+
+    override fun TorchTensorInt.plusAssign(value: Int): Unit =
+        JTorch.plusIntAssign(value, this.tensorHandle)
+
+    override operator fun Int.minus(other: TorchTensorInt): TorchTensorInt =
+        wrap(JTorch.plusInt(-this, other.tensorHandle))
+
+    override fun TorchTensorInt.minus(value: Int): TorchTensorInt =
+        wrap(JTorch.plusInt(-value, this.tensorHandle))
+
+    override fun TorchTensorInt.minusAssign(value: Int): Unit =
+        JTorch.plusIntAssign(-value, this.tensorHandle)
+
+    override operator fun Int.times(other: TorchTensorInt): TorchTensorInt =
+        wrap(JTorch.timesInt(this, other.tensorHandle))
+
+    override fun TorchTensorInt.times(value: Int): TorchTensorInt =
+        wrap(JTorch.timesInt(value, this.tensorHandle))
+
+    override fun TorchTensorInt.timesAssign(value: Int): Unit =
+        JTorch.timesIntAssign(value, this.tensorHandle)
+
+    override fun full(value: Int, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(JTorch.fullInt(value, shape, device.toInt()))
+
+}
+
+public inline fun <R> TorchTensorRealAlgebra(block: TorchTensorRealAlgebra.() -> R): R =
+    withDeferScope { TorchTensorRealAlgebra(this).block() }
+
+public inline fun <R> TorchTensorFloatAlgebra(block: TorchTensorFloatAlgebra.() -> R): R =
+    withDeferScope { TorchTensorFloatAlgebra(this).block() }
+
+public inline fun <R> TorchTensorLongAlgebra(block: TorchTensorLongAlgebra.() -> R): R =
+    withDeferScope { TorchTensorLongAlgebra(this).block() }
+
+public inline fun <R> TorchTensorIntAlgebra(block: TorchTensorIntAlgebra.() -> R): R =
+    withDeferScope { TorchTensorIntAlgebra(this).block() }

kmath-torch/src/jvmMain/kotlin/space/kscience/kmath/torch/TorchTensorJVM.kt

@@ -0,0 +1,94 @@
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.memory.DeferScope
+
+public sealed class TorchTensorJVM<T> constructor(
+    scope: DeferScope,
+    internal val tensorHandle: Long
+) : TorchTensor<T>, TorchTensorMemoryHolder(scope)
+{
+    override fun close(): Unit = JTorch.disposeTensor(tensorHandle)
+
+    override val dimension: Int get() = JTorch.getDim(tensorHandle)
+    override val shape: IntArray
+        get() = (1..dimension).map { JTorch.getShapeAt(tensorHandle, it - 1) }.toIntArray()
+    override val strides: IntArray
+        get() = (1..dimension).map { JTorch.getStrideAt(tensorHandle, it - 1) }.toIntArray()
+    override val size: Int get() = JTorch.getNumel(tensorHandle)
+    override val device: Device get() = Device.fromInt(JTorch.getDevice(tensorHandle))
+
+    override fun toString(): String = JTorch.tensorToString(tensorHandle)
+
+    public fun copyToDouble(): TorchTensorReal = TorchTensorReal(
+        scope = scope,
+        tensorHandle = JTorch.copyToDouble(this.tensorHandle)
+    )
+
+    public fun copyToFloat(): TorchTensorFloat = TorchTensorFloat(
+        scope = scope,
+        tensorHandle = JTorch.copyToFloat(this.tensorHandle)
+    )
+
+    public fun copyToLong(): TorchTensorLong = TorchTensorLong(
+        scope = scope,
+        tensorHandle = JTorch.copyToLong(this.tensorHandle)
+    )
+
+    public fun copyToInt(): TorchTensorInt = TorchTensorInt(
+        scope = scope,
+        tensorHandle = JTorch.copyToInt(this.tensorHandle)
+    )
+}
+
+public sealed class TorchTensorOverFieldJVM<T> constructor(
+    scope: DeferScope,
+    tensorHandle: Long
+) : TorchTensorJVM<T>(scope, tensorHandle), TorchTensorOverField<T> {
+    override var requiresGrad: Boolean
+        get() = JTorch.requiresGrad(tensorHandle)
+        set(value) = JTorch.setRequiresGrad(tensorHandle, value)
+}
+
+public class TorchTensorReal internal constructor(
+    scope: DeferScope,
+    tensorHandle: Long
+) : TorchTensorOverFieldJVM<Double>(scope, tensorHandle) {
+    override fun item(): Double = JTorch.getItemDouble(tensorHandle)
+    override fun get(index: IntArray): Double = JTorch.getDouble(tensorHandle, index)
+    override fun set(index: IntArray, value: Double) {
+        JTorch.setDouble(tensorHandle, index, value)
+    }
+}
+
+public class TorchTensorFloat internal constructor(
+    scope: DeferScope,
+    tensorHandle: Long
+) : TorchTensorOverFieldJVM<Float>(scope, tensorHandle) {
+    override fun item(): Float = JTorch.getItemFloat(tensorHandle)
+    override fun get(index: IntArray): Float = JTorch.getFloat(tensorHandle, index)
+    override fun set(index: IntArray, value: Float) {
+        JTorch.setFloat(tensorHandle, index, value)
+    }
+}
+
+public class TorchTensorLong internal constructor(
+    scope: DeferScope,
+    tensorHandle: Long
+) : TorchTensorOverFieldJVM<Long>(scope, tensorHandle) {
+    override fun item(): Long = JTorch.getItemLong(tensorHandle)
+    override fun get(index: IntArray): Long = JTorch.getLong(tensorHandle, index)
+    override fun set(index: IntArray, value: Long) {
+        JTorch.setLong(tensorHandle, index, value)
+    }
+}
+
+public class TorchTensorInt internal constructor(
+    scope: DeferScope,
+    tensorHandle: Long
+) : TorchTensorOverFieldJVM<Int>(scope, tensorHandle) {
+    override fun item(): Int = JTorch.getItemInt(tensorHandle)
+    override fun get(index: IntArray): Int = JTorch.getInt(tensorHandle, index)
+    override fun set(index: IntArray, value: Int) {
+        JTorch.setInt(tensorHandle, index, value)
+    }
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/BenchmarkMatMul.kt

@@ -0,0 +1,26 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+class BenchmarkMatMul {
+
+    @Test
+    fun benchmarkMatMulDouble() = TorchTensorRealAlgebra {
+        benchmarkMatMul(20, 10, 100000, "Real")
+        benchmarkMatMul(200, 10, 10000, "Real")
+        benchmarkMatMul(2000, 3, 20, "Real")
+    }
+
+    @Test
+    fun benchmarkMatMulFloat() = TorchTensorFloatAlgebra {
+        benchmarkMatMul(20, 10, 100000, "Float")
+        benchmarkMatMul(200, 10, 10000, "Float")
+        benchmarkMatMul(2000, 3, 20, "Float")
+        if (cudaAvailable()) {
+            benchmarkMatMul(20, 10, 100000, "Float", Device.CUDA(0))
+            benchmarkMatMul(200, 10, 10000, "Float", Device.CUDA(0))
+            benchmarkMatMul(2000, 10, 1000, "Float", Device.CUDA(0))
+        }
+    }
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/BenchmarkRandomGenerators.kt

@@ -0,0 +1,27 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+class BenchmarkRandomGenerators {
+    @Test
+    fun benchmarkRand1() = TorchTensorFloatAlgebra{
+        benchmarkingRand1()
+    }
+
+    @Test
+    fun benchmarkRand3() = TorchTensorFloatAlgebra{
+        benchmarkingRand3()
+    }
+
+    @Test
+    fun benchmarkRand5() = TorchTensorFloatAlgebra{
+        benchmarkingRand5()
+    }
+
+    @Test
+    fun benchmarkRand7() = TorchTensorFloatAlgebra{
+        benchmarkingRand7()
+    }
+
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/TestAutograd.kt

@@ -0,0 +1,24 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+class TestAutograd {
+    @Test
+    fun testAutoGrad() = TorchTensorFloatAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingAutoGrad(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedAutoGrad() = TorchTensorFloatAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedAutoGrad(device)
+            }
+        }
+    }
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/TestTorchTensor.kt

@@ -0,0 +1,39 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.*
+
+
+class TestTorchTensor {
+
+    @Test
+    fun testCopying() = TorchTensorFloatAlgebra {
+        withCuda { device ->
+            testingCopying(device)
+        }
+    }
+
+    @Test
+    fun testRequiresGrad() = TorchTensorRealAlgebra {
+        testingRequiresGrad()
+    }
+
+    @Test
+    fun testTypeMoving() = TorchTensorFloatAlgebra {
+        val tensorInt = copyFromArray(floatArrayOf(1f, 2f, 3f), intArrayOf(3)).copyToInt()
+        TorchTensorIntAlgebra {
+            val temporalTensor = copyFromArray(intArrayOf(4, 5, 6), intArrayOf(3))
+            tensorInt swap temporalTensor
+            assertTrue(temporalTensor.copyToArray() contentEquals intArrayOf(1, 2, 3))
+        }
+        assertTrue(tensorInt.copyToFloat().copyToArray() contentEquals floatArrayOf(4f, 5f, 6f))
+    }
+
+    @Test
+    fun testViewWithNoCopy() = TorchTensorIntAlgebra {
+        withChecks {
+            withCuda {
+                    device ->  testingViewWithNoCopy(device)
+            }
+        }
+    }
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/TestTorchTensorAlgebra.kt

@@ -0,0 +1,63 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+class TestTorchTensorAlgebra {
+
+    @Test
+    fun testScalarProduct() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingScalarProduct(device)
+            }
+        }
+    }
+
+    @Test
+    fun testMatrixMultiplication() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingMatrixMultiplication(device)
+            }
+        }
+    }
+
+    @Test
+    fun testLinearStructure() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingLinearStructure(device)
+            }
+        }
+    }
+
+    @Test
+    fun testTensorTransformations() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingTensorTransformations(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedSVD() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedSVD(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedSymEig() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedSymEig(device)
+            }
+        }
+    }
+
+
+}

kmath-torch/src/jvmTest/kotlin/space/kscience/kmath/torch/TestUtils.kt

@@ -0,0 +1,21 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.*
+
+
+class TestUtils {
+
+    @Test
+    fun testSetNumThreads() {
+        TorchTensorLongAlgebra {
+            testingSetNumThreads()
+        }
+    }
+
+    @Test
+    fun testSeedSetting() = TorchTensorFloatAlgebra {
+        withCuda { device ->
+            testingSetSeed(device)
+        }
+    }
+}

kmath-torch/src/nativeInterop/cinterop/libctorch.def

@@ -0,0 +1,2 @@
+package=space.kscience.kmath.torch.ctorch
+headers=ctorch.h

kmath-torch/src/nativeMain/kotlin/space/kscience/kmath/torch/TorchTensorAlgebraNative.kt

@@ -0,0 +1,444 @@
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.memory.DeferScope
+import space.kscience.kmath.memory.withDeferScope
+
+import kotlinx.cinterop.*
+import space.kscience.kmath.tensors.*
+import space.kscience.kmath.torch.ctorch.*
+
+public sealed class TorchTensorAlgebraNative<
+        T,
+        TVar : CPrimitiveVar,
+        PrimitiveArrayType,
+        TorchTensorType : TorchTensorNative<T>> constructor(
+    internal val scope: DeferScope
+) : TorchTensorAlgebra<T, PrimitiveArrayType, TorchTensorType> {
+
+    override fun getNumThreads(): Int {
+        return get_num_threads()
+    }
+
+    override fun setNumThreads(numThreads: Int): Unit {
+        set_num_threads(numThreads)
+    }
+
+    override fun cudaAvailable(): Boolean {
+        return cuda_is_available()
+    }
+
+    override fun setSeed(seed: Int): Unit {
+        set_seed(seed)
+    }
+
+    override var checks: Boolean = false
+
+    internal abstract fun wrap(tensorHandle: COpaquePointer): TorchTensorType
+
+    public abstract fun fromBlob(arrayBlob: CPointer<TVar>, shape: IntArray): TorchTensorType
+    public abstract fun TorchTensorType.getData(): CPointer<TVar>
+
+    override operator fun TorchTensorType.times(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(times_tensor(this.tensorHandle, other.tensorHandle)!!)
+    }
+
+    override operator fun TorchTensorType.timesAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        times_tensor_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.plus(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(plus_tensor(this.tensorHandle, other.tensorHandle)!!)
+    }
+
+    override operator fun TorchTensorType.plusAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        plus_tensor_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.minus(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(minus_tensor(this.tensorHandle, other.tensorHandle)!!)
+    }
+
+    override operator fun TorchTensorType.minusAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        minus_tensor_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override operator fun TorchTensorType.unaryMinus(): TorchTensorType =
+        wrap(unary_minus(this.tensorHandle)!!)
+
+    override infix fun TorchTensorType.dot(other: TorchTensorType): TorchTensorType {
+        if (checks) checkDotOperation(this, other)
+        return wrap(matmul(this.tensorHandle, other.tensorHandle)!!)
+    }
+
+    override infix fun TorchTensorType.dotAssign(other: TorchTensorType): Unit {
+        if (checks) checkDotOperation(this, other)
+        matmul_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override infix fun TorchTensorType.dotRightAssign(other: TorchTensorType): Unit {
+        if (checks) checkDotOperation(this, other)
+        matmul_right_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override fun diagonalEmbedding(
+        diagonalEntries: TorchTensorType, offset: Int, dim1: Int, dim2: Int
+    ): TorchTensorType =
+        wrap(diag_embed(diagonalEntries.tensorHandle, offset, dim1, dim2)!!)
+
+    override fun TorchTensorType.transpose(i: Int, j: Int): TorchTensorType {
+        if (checks) checkTranspose(this.dimension, i, j)
+        return wrap(transpose_tensor(tensorHandle, i, j)!!)
+    }
+
+    override fun TorchTensorType.transposeAssign(i: Int, j: Int): Unit {
+        if (checks) checkTranspose(this.dimension, i, j)
+        transpose_tensor_assign(tensorHandle, i, j)
+    }
+
+    override fun TorchTensorType.view(shape: IntArray): TorchTensorType {
+        if (checks) checkView(this, shape)
+        return wrap(view_tensor(this.tensorHandle, shape.toCValues(), shape.size)!!)
+    }
+
+    override fun TorchTensorType.abs(): TorchTensorType = wrap(abs_tensor(tensorHandle)!!)
+    override fun TorchTensorType.absAssign(): Unit = abs_tensor_assign(tensorHandle)
+
+    override fun TorchTensorType.sum(): TorchTensorType = wrap(sum_tensor(tensorHandle)!!)
+    override fun TorchTensorType.sumAssign(): Unit = sum_tensor_assign(tensorHandle)
+
+    override fun TorchTensorType.randIntegral(low: Long, high: Long): TorchTensorType =
+        wrap(randint_like(this.tensorHandle, low, high)!!)
+
+    override fun TorchTensorType.randIntegralAssign(low: Long, high: Long): Unit =
+        randint_like_assign(this.tensorHandle, low, high)
+
+    override fun TorchTensorType.copy(): TorchTensorType =
+        wrap(copy_tensor(this.tensorHandle)!!)
+
+    override fun TorchTensorType.copyToDevice(device: Device): TorchTensorType =
+        wrap(copy_to_device(this.tensorHandle, device.toInt())!!)
+
+    override infix fun TorchTensorType.swap(other: TorchTensorType): Unit =
+        swap_tensors(this.tensorHandle, other.tensorHandle)
+}
+
+public sealed class TorchTensorPartialDivisionAlgebraNative<T, TVar : CPrimitiveVar,
+        PrimitiveArrayType, TorchTensorType : TorchTensorOverFieldNative<T>>(scope: DeferScope) :
+    TorchTensorAlgebraNative<T, TVar, PrimitiveArrayType, TorchTensorType>(scope),
+    TorchTensorPartialDivisionAlgebra<T, PrimitiveArrayType, TorchTensorType> {
+
+    override operator fun TorchTensorType.div(other: TorchTensorType): TorchTensorType {
+        if (checks) checkLinearOperation(this, other)
+        return wrap(div_tensor(this.tensorHandle, other.tensorHandle)!!)
+    }
+
+    override operator fun TorchTensorType.divAssign(other: TorchTensorType): Unit {
+        if (checks) checkLinearOperation(this, other)
+        div_tensor_assign(this.tensorHandle, other.tensorHandle)
+    }
+
+    override fun TorchTensorType.randUniform(): TorchTensorType =
+        wrap(rand_like(this.tensorHandle)!!)
+
+    override fun TorchTensorType.randUniformAssign(): Unit =
+        rand_like_assign(this.tensorHandle)
+
+
+    override fun TorchTensorType.randNormal(): TorchTensorType =
+        wrap(randn_like(this.tensorHandle)!!)
+
+    override fun TorchTensorType.randNormalAssign(): Unit =
+        randn_like_assign(this.tensorHandle)
+
+
+    override fun TorchTensorType.exp(): TorchTensorType = wrap(exp_tensor(tensorHandle)!!)
+    override fun TorchTensorType.expAssign(): Unit = exp_tensor_assign(tensorHandle)
+    override fun TorchTensorType.log(): TorchTensorType = wrap(log_tensor(tensorHandle)!!)
+    override fun TorchTensorType.logAssign(): Unit = log_tensor_assign(tensorHandle)
+
+    override fun TorchTensorType.svd(): Triple<TorchTensorType, TorchTensorType, TorchTensorType> {
+        val U = empty_tensor()!!
+        val V = empty_tensor()!!
+        val S = empty_tensor()!!
+        svd_tensor(this.tensorHandle, U, S, V)
+        return Triple(wrap(U), wrap(S), wrap(V))
+    }
+
+    override fun TorchTensorType.symEig(eigenvectors: Boolean): Pair<TorchTensorType, TorchTensorType> {
+        val V = empty_tensor()!!
+        val S = empty_tensor()!!
+        symeig_tensor(this.tensorHandle, S, V, eigenvectors)
+        return Pair(wrap(S), wrap(V))
+    }
+
+    override fun TorchTensorType.grad(variable: TorchTensorType, retainGraph: Boolean): TorchTensorType {
+        if (checks) this.checkIsValue()
+        return wrap(autograd_tensor(this.tensorHandle, variable.tensorHandle, retainGraph)!!)
+    }
+
+    override infix fun TorchTensorType.hess(variable: TorchTensorType): TorchTensorType {
+        if (checks) this.checkIsValue()
+        return wrap(autohess_tensor(this.tensorHandle, variable.tensorHandle)!!)
+    }
+
+    override fun TorchTensorType.detachFromGraph(): TorchTensorType =
+        wrap(detach_from_graph(this.tensorHandle)!!)
+
+}
+
+public class TorchTensorRealAlgebra(scope: DeferScope) :
+    TorchTensorPartialDivisionAlgebraNative<Double, DoubleVar, DoubleArray, TorchTensorReal>(scope) {
+    override fun wrap(tensorHandle: COpaquePointer): TorchTensorReal =
+        TorchTensorReal(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorReal.copyToArray(): DoubleArray =
+        this.elements().map { it.second }.toList().toDoubleArray()
+
+    override fun copyFromArray(array: DoubleArray, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(from_blob_double(array.toCValues(), shape.toCValues(), shape.size, device.toInt(), true)!!)
+
+    override fun fromBlob(arrayBlob: CPointer<DoubleVar>, shape: IntArray): TorchTensorReal =
+        wrap(from_blob_double(arrayBlob, shape.toCValues(), shape.size, Device.CPU.toInt(), false)!!)
+
+    override fun TorchTensorReal.getData(): CPointer<DoubleVar> {
+        require(this.device is Device.CPU) {
+            "This tensor is not on available on CPU"
+        }
+        return get_data_double(this.tensorHandle)!!
+    }
+
+    override fun randNormal(shape: IntArray, device: Device): TorchTensorReal =
+        wrap(randn_double(shape.toCValues(), shape.size, device.toInt())!!)
+
+    override fun randUniform(shape: IntArray, device: Device): TorchTensorReal =
+        wrap(rand_double(shape.toCValues(), shape.size, device.toInt())!!)
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(randint_double(low, high, shape.toCValues(), shape.size, device.toInt())!!)
+
+    override operator fun Double.plus(other: TorchTensorReal): TorchTensorReal =
+        wrap(plus_double(this, other.tensorHandle)!!)
+
+    override fun TorchTensorReal.plus(value: Double): TorchTensorReal =
+        wrap(plus_double(value, this.tensorHandle)!!)
+
+    override fun TorchTensorReal.plusAssign(value: Double): Unit {
+        plus_double_assign(value, this.tensorHandle)
+    }
+
+    override operator fun Double.minus(other: TorchTensorReal): TorchTensorReal =
+        wrap(plus_double(-this, other.tensorHandle)!!)
+
+    override fun TorchTensorReal.minus(value: Double): TorchTensorReal =
+        wrap(plus_double(-value, this.tensorHandle)!!)
+
+    override fun TorchTensorReal.minusAssign(value: Double): Unit {
+        plus_double_assign(-value, this.tensorHandle)
+    }
+
+    override operator fun Double.times(other: TorchTensorReal): TorchTensorReal =
+        wrap(times_double(this, other.tensorHandle)!!)
+
+    override fun TorchTensorReal.times(value: Double): TorchTensorReal =
+        wrap(times_double(value, this.tensorHandle)!!)
+
+    override fun TorchTensorReal.timesAssign(value: Double): Unit {
+        times_double_assign(value, this.tensorHandle)
+    }
+
+    override fun full(value: Double, shape: IntArray, device: Device): TorchTensorReal =
+        wrap(full_double(value, shape.toCValues(), shape.size, device.toInt())!!)
+}
+
+
+public class TorchTensorFloatAlgebra(scope: DeferScope) :
+    TorchTensorPartialDivisionAlgebraNative<Float, FloatVar, FloatArray, TorchTensorFloat>(scope) {
+    override fun wrap(tensorHandle: COpaquePointer): TorchTensorFloat =
+        TorchTensorFloat(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorFloat.copyToArray(): FloatArray =
+        this.elements().map { it.second }.toList().toFloatArray()
+
+    override fun copyFromArray(array: FloatArray, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(from_blob_float(array.toCValues(), shape.toCValues(), shape.size, device.toInt(), true)!!)
+
+    override fun fromBlob(arrayBlob: CPointer<FloatVar>, shape: IntArray): TorchTensorFloat =
+        wrap(from_blob_float(arrayBlob, shape.toCValues(), shape.size, Device.CPU.toInt(), false)!!)
+
+    override fun TorchTensorFloat.getData(): CPointer<FloatVar> {
+        require(this.device is Device.CPU) {
+            "This tensor is not on available on CPU"
+        }
+        return get_data_float(this.tensorHandle)!!
+    }
+
+    override fun randNormal(shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(randn_float(shape.toCValues(), shape.size, device.toInt())!!)
+
+    override fun randUniform(shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(rand_float(shape.toCValues(), shape.size, device.toInt())!!)
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(randint_float(low, high, shape.toCValues(), shape.size, device.toInt())!!)
+
+    override operator fun Float.plus(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(plus_float(this, other.tensorHandle)!!)
+
+    override fun TorchTensorFloat.plus(value: Float): TorchTensorFloat =
+        wrap(plus_float(value, this.tensorHandle)!!)
+
+    override fun TorchTensorFloat.plusAssign(value: Float): Unit =
+        plus_float_assign(value, this.tensorHandle)
+
+    override operator fun Float.minus(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(plus_float(-this, other.tensorHandle)!!)
+
+    override fun TorchTensorFloat.minus(value: Float): TorchTensorFloat =
+        wrap(plus_float(-value, this.tensorHandle)!!)
+
+    override fun TorchTensorFloat.minusAssign(value: Float): Unit =
+        plus_float_assign(-value, this.tensorHandle)
+
+    override operator fun Float.times(other: TorchTensorFloat): TorchTensorFloat =
+        wrap(times_float(this, other.tensorHandle)!!)
+
+    override fun TorchTensorFloat.times(value: Float): TorchTensorFloat =
+        wrap(times_float(value, this.tensorHandle)!!)
+
+    override fun TorchTensorFloat.timesAssign(value: Float): Unit =
+        times_float_assign(value, this.tensorHandle)
+
+    override fun full(value: Float, shape: IntArray, device: Device): TorchTensorFloat =
+        wrap(full_float(value, shape.toCValues(), shape.size, device.toInt())!!)
+
+}
+
+public class TorchTensorLongAlgebra(scope: DeferScope) :
+    TorchTensorAlgebraNative<Long, LongVar, LongArray, TorchTensorLong>(scope) {
+    override fun wrap(tensorHandle: COpaquePointer): TorchTensorLong =
+        TorchTensorLong(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorLong.copyToArray(): LongArray =
+        this.elements().map { it.second }.toList().toLongArray()
+
+    override fun copyFromArray(array: LongArray, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(from_blob_long(array.toCValues(), shape.toCValues(), shape.size, device.toInt(), true)!!)
+
+    override fun fromBlob(arrayBlob: CPointer<LongVar>, shape: IntArray): TorchTensorLong =
+        wrap(from_blob_long(arrayBlob, shape.toCValues(), shape.size, Device.CPU.toInt(), false)!!)
+
+    override fun TorchTensorLong.getData(): CPointer<LongVar> {
+        check(this.device is Device.CPU) {
+            "This tensor is not on available on CPU"
+        }
+        return get_data_long(this.tensorHandle)!!
+    }
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(randint_long(low, high, shape.toCValues(), shape.size, device.toInt())!!)
+
+    override operator fun Long.plus(other: TorchTensorLong): TorchTensorLong =
+        wrap(plus_long(this, other.tensorHandle)!!)
+
+    override fun TorchTensorLong.plus(value: Long): TorchTensorLong =
+        wrap(plus_long(value, this.tensorHandle)!!)
+
+    override fun TorchTensorLong.plusAssign(value: Long): Unit =
+        plus_long_assign(value, this.tensorHandle)
+
+    override operator fun Long.minus(other: TorchTensorLong): TorchTensorLong =
+        wrap(plus_long(-this, other.tensorHandle)!!)
+
+    override fun TorchTensorLong.minus(value: Long): TorchTensorLong =
+        wrap(plus_long(-value, this.tensorHandle)!!)
+
+    override fun TorchTensorLong.minusAssign(value: Long): Unit =
+        plus_long_assign(-value, this.tensorHandle)
+
+    override operator fun Long.times(other: TorchTensorLong): TorchTensorLong =
+        wrap(times_long(this, other.tensorHandle)!!)
+
+    override fun TorchTensorLong.times(value: Long): TorchTensorLong =
+        wrap(times_long(value, this.tensorHandle)!!)
+
+    override fun TorchTensorLong.timesAssign(value: Long): Unit =
+        times_long_assign(value, this.tensorHandle)
+
+    override fun full(value: Long, shape: IntArray, device: Device): TorchTensorLong =
+        wrap(full_long(value, shape.toCValues(), shape.size, device.toInt())!!)
+}
+
+public class TorchTensorIntAlgebra(scope: DeferScope) :
+    TorchTensorAlgebraNative<Int, IntVar, IntArray, TorchTensorInt>(scope) {
+    override fun wrap(tensorHandle: COpaquePointer): TorchTensorInt =
+        TorchTensorInt(scope = scope, tensorHandle = tensorHandle)
+
+    override fun TorchTensorInt.copyToArray(): IntArray =
+        this.elements().map { it.second }.toList().toIntArray()
+
+    override fun copyFromArray(array: IntArray, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(from_blob_int(array.toCValues(), shape.toCValues(), shape.size, device.toInt(), true)!!)
+
+    override fun fromBlob(arrayBlob: CPointer<IntVar>, shape: IntArray): TorchTensorInt =
+        wrap(from_blob_int(arrayBlob, shape.toCValues(), shape.size, Device.CPU.toInt(), false)!!)
+
+    override fun TorchTensorInt.getData(): CPointer<IntVar> {
+        require(this.device is Device.CPU) {
+            "This tensor is not on available on CPU"
+        }
+        return get_data_int(this.tensorHandle)!!
+    }
+
+    override fun randIntegral(low: Long, high: Long, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(randint_int(low, high, shape.toCValues(), shape.size, device.toInt())!!)
+
+    override operator fun Int.plus(other: TorchTensorInt): TorchTensorInt =
+        wrap(plus_int(this, other.tensorHandle)!!)
+
+    override fun TorchTensorInt.plus(value: Int): TorchTensorInt =
+        wrap(plus_int(value, this.tensorHandle)!!)
+
+    override fun TorchTensorInt.plusAssign(value: Int): Unit =
+        plus_int_assign(value, this.tensorHandle)
+
+    override operator fun Int.minus(other: TorchTensorInt): TorchTensorInt =
+        wrap(plus_int(-this, other.tensorHandle)!!)
+
+    override fun TorchTensorInt.minus(value: Int): TorchTensorInt =
+        wrap(plus_int(-value, this.tensorHandle)!!)
+
+    override fun TorchTensorInt.minusAssign(value: Int): Unit =
+        plus_int_assign(-value, this.tensorHandle)
+
+    override operator fun Int.times(other: TorchTensorInt): TorchTensorInt =
+        wrap(times_int(this, other.tensorHandle)!!)
+
+    override fun TorchTensorInt.times(value: Int): TorchTensorInt =
+        wrap(times_int(value, this.tensorHandle)!!)
+
+    override fun TorchTensorInt.timesAssign(value: Int): Unit =
+        times_int_assign(value, this.tensorHandle)
+
+    override fun full(value: Int, shape: IntArray, device: Device): TorchTensorInt =
+        wrap(full_int(value, shape.toCValues(), shape.size, device.toInt())!!)
+}
+
+
+public inline fun <R> TorchTensorRealAlgebra(block: TorchTensorRealAlgebra.() -> R): R =
+    withDeferScope { TorchTensorRealAlgebra(this).block() }
+
+public inline fun <R> TorchTensorFloatAlgebra(block: TorchTensorFloatAlgebra.() -> R): R =
+    withDeferScope { TorchTensorFloatAlgebra(this).block() }
+
+public inline fun <R> TorchTensorLongAlgebra(block: TorchTensorLongAlgebra.() -> R): R =
+    withDeferScope { TorchTensorLongAlgebra(this).block() }
+
+public inline fun <R> TorchTensorIntAlgebra(block: TorchTensorIntAlgebra.() -> R): R =
+    withDeferScope { TorchTensorIntAlgebra(this).block() }
+

kmath-torch/src/nativeMain/kotlin/space/kscience/kmath/torch/TorchTensorNative.kt

@@ -0,0 +1,104 @@
+package space.kscience.kmath.torch
+
+import space.kscience.kmath.memory.DeferScope
+
+import kotlinx.cinterop.*
+import space.kscience.kmath.torch.ctorch.*
+
+
+public sealed class TorchTensorNative<T> constructor(
+    scope: DeferScope,
+    internal val tensorHandle: COpaquePointer
+) : TorchTensor<T>, TorchTensorMemoryHolder(scope) {
+
+    override fun close(): Unit = dispose_tensor(tensorHandle)
+
+    override val dimension: Int get() = get_dim(tensorHandle)
+    override val shape: IntArray
+        get() = (1..dimension).map { get_shape_at(tensorHandle, it - 1) }.toIntArray()
+    override val strides: IntArray
+        get() = (1..dimension).map { get_stride_at(tensorHandle, it - 1) }.toIntArray()
+    override val size: Int get() = get_numel(tensorHandle)
+    override val device: Device get() = Device.fromInt(get_device(tensorHandle))
+
+    override fun toString(): String {
+        val nativeStringRepresentation: CPointer<ByteVar> = tensor_to_string(tensorHandle)!!
+        val stringRepresentation = nativeStringRepresentation.toKString()
+        dispose_char(nativeStringRepresentation)
+        return stringRepresentation
+    }
+
+    public fun copyToDouble(): TorchTensorReal = TorchTensorReal(
+        scope = scope,
+        tensorHandle = copy_to_double(this.tensorHandle)!!
+    )
+
+    public fun copyToFloat(): TorchTensorFloat = TorchTensorFloat(
+        scope = scope,
+        tensorHandle = copy_to_float(this.tensorHandle)!!
+    )
+
+    public fun copyToLong(): TorchTensorLong = TorchTensorLong(
+        scope = scope,
+        tensorHandle = copy_to_long(this.tensorHandle)!!
+    )
+
+    public fun copyToInt(): TorchTensorInt = TorchTensorInt(
+        scope = scope,
+        tensorHandle = copy_to_int(this.tensorHandle)!!
+    )
+}
+
+public sealed class TorchTensorOverFieldNative<T> constructor(
+    scope: DeferScope,
+    tensorHandle: COpaquePointer
+) : TorchTensorNative<T>(scope, tensorHandle), TorchTensorOverField<T> {
+    override var requiresGrad: Boolean
+        get() = requires_grad(tensorHandle)
+        set(value) = requires_grad_(tensorHandle, value)
+}
+
+
+public class TorchTensorReal internal constructor(
+    scope: DeferScope,
+    tensorHandle: COpaquePointer
+) : TorchTensorOverFieldNative<Double>(scope, tensorHandle) {
+    override fun item(): Double = get_item_double(tensorHandle)
+    override fun get(index: IntArray): Double = get_double(tensorHandle, index.toCValues())
+    override fun set(index: IntArray, value: Double) {
+        set_double(tensorHandle, index.toCValues(), value)
+    }
+}
+
+public class TorchTensorFloat internal constructor(
+    scope: DeferScope,
+    tensorHandle: COpaquePointer
+) : TorchTensorOverFieldNative<Float>(scope, tensorHandle) {
+    override fun item(): Float = get_item_float(tensorHandle)
+    override fun get(index: IntArray): Float = get_float(tensorHandle, index.toCValues())
+    override fun set(index: IntArray, value: Float) {
+        set_float(tensorHandle, index.toCValues(), value)
+    }
+}
+
+public class TorchTensorLong internal constructor(
+    scope: DeferScope,
+    tensorHandle: COpaquePointer
+) : TorchTensorNative<Long>(scope, tensorHandle) {
+    override fun item(): Long = get_item_long(tensorHandle)
+    override fun get(index: IntArray): Long = get_long(tensorHandle, index.toCValues())
+    override fun set(index: IntArray, value: Long) {
+        set_long(tensorHandle, index.toCValues(), value)
+    }
+}
+
+public class TorchTensorInt internal constructor(
+    scope: DeferScope,
+    tensorHandle: COpaquePointer
+) : TorchTensorNative<Int>(scope, tensorHandle) {
+    override fun item(): Int = get_item_int(tensorHandle)
+    override fun get(index: IntArray): Int = get_int(tensorHandle, index.toCValues())
+    override fun set(index: IntArray, value: Int) {
+        set_int(tensorHandle, index.toCValues(), value)
+    }
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/BenchmarkMatMul.kt

@@ -0,0 +1,26 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+internal class BenchmarkMatMul {
+
+    @Test
+    fun benchmarkMatMulDouble() = TorchTensorRealAlgebra {
+        benchmarkMatMul(20, 10, 100000, "Real")
+        benchmarkMatMul(200, 10, 10000, "Real")
+        benchmarkMatMul(2000, 3, 20, "Real")
+    }
+
+    @Test
+    fun benchmarkMatMulFloat() = TorchTensorFloatAlgebra {
+        benchmarkMatMul(20, 10, 100000, "Float")
+        benchmarkMatMul(200, 10, 10000, "Float")
+        benchmarkMatMul(2000, 3, 20, "Float")
+        if (cudaAvailable()) {
+            benchmarkMatMul(20, 10, 100000, "Float", Device.CUDA(0))
+            benchmarkMatMul(200, 10, 10000, "Float", Device.CUDA(0))
+            benchmarkMatMul(2000, 10, 1000, "Float", Device.CUDA(0))
+        }
+    }
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/BenchmarkRandomGenerators.kt

@@ -0,0 +1,28 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+internal class BenchmarkRandomGenerators {
+
+    @Test
+    fun benchmarkRand1() = TorchTensorFloatAlgebra{
+        benchmarkingRand1()
+    }
+
+    @Test
+    fun benchmarkRand3() = TorchTensorFloatAlgebra{
+        benchmarkingRand3()
+    }
+
+    @Test
+    fun benchmarkRand5() = TorchTensorFloatAlgebra{
+        benchmarkingRand5()
+    }
+
+    @Test
+    fun benchmarkRand7() = TorchTensorFloatAlgebra{
+        benchmarkingRand7()
+    }
+
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/TestAutograd.kt

@@ -0,0 +1,24 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+internal class TestAutograd {
+    @Test
+    fun testAutoGrad() = TorchTensorFloatAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingAutoGrad(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedAutoGrad() = TorchTensorFloatAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedAutoGrad(device)
+            }
+        }
+    }
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/TestTorchTensor.kt

@@ -0,0 +1,57 @@
+package space.kscience.kmath.torch
+
+import kotlinx.cinterop.*
+import kotlin.test.*
+
+
+class TestTorchTensor {
+
+    @Test
+    fun testCopying() = TorchTensorFloatAlgebra {
+        withCuda {
+            device -> testingCopying(device)
+        }
+    }
+
+    @Test
+    fun testNativeNoCopyDataTransferOnCPU() = memScoped {
+        val data = allocArray<DoubleVar>(1)
+        data[0] = 1.0
+        TorchTensorRealAlgebra {
+            val tensor = fromBlob(data, intArrayOf(1))
+            assertEquals(tensor[intArrayOf(0)], 1.0)
+            data[0] = 2.0
+            assertEquals(tensor[intArrayOf(0)], 2.0)
+            val tensorData = tensor.getData()
+            tensorData[0] = 3.0
+            assertEquals(tensor[intArrayOf(0)], 3.0)
+        }
+        assertEquals(data[0], 3.0)
+    }
+
+    @Test
+    fun testRequiresGrad() = TorchTensorRealAlgebra {
+        testingRequiresGrad()
+    }
+
+    @Test
+    fun testTypeMoving() = TorchTensorFloatAlgebra {
+        val tensorInt = copyFromArray(floatArrayOf(1f, 2f, 3f), intArrayOf(3)).copyToInt()
+        TorchTensorIntAlgebra {
+            val temporalTensor = copyFromArray(intArrayOf(4, 5, 6), intArrayOf(3))
+            tensorInt swap temporalTensor
+            assertTrue(temporalTensor.copyToArray() contentEquals intArrayOf(1, 2, 3))
+        }
+        assertTrue(tensorInt.copyToFloat().copyToArray() contentEquals floatArrayOf(4f, 5f, 6f))
+    }
+
+    @Test
+    fun testViewWithNoCopy() = TorchTensorIntAlgebra {
+        withChecks {
+            withCuda {
+                    device ->  testingViewWithNoCopy(device)
+            }
+        }
+    }
+
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/TestTorchTensorAlgebra.kt

@@ -0,0 +1,62 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.Test
+
+
+internal class TestTorchTensorAlgebra {
+
+    @Test
+    fun testScalarProduct() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingScalarProduct(device)
+            }
+        }
+    }
+
+    @Test
+    fun testMatrixMultiplication() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingMatrixMultiplication(device)
+            }
+        }
+    }
+
+    @Test
+    fun testLinearStructure() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingLinearStructure(device)
+            }
+        }
+    }
+
+    @Test
+    fun testTensorTransformations() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingTensorTransformations(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedSVD() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedSVD(device)
+            }
+        }
+    }
+
+    @Test
+    fun testBatchedSymEig() = TorchTensorRealAlgebra {
+        withChecks {
+            withCuda { device ->
+                testingBatchedSymEig(device)
+            }
+        }
+    }
+
+}

kmath-torch/src/nativeTest/kotlin/space/kscience/kmath/torch/TestUtils.kt

@@ -0,0 +1,20 @@
+package space.kscience.kmath.torch
+
+import kotlin.test.*
+
+
+internal class TestUtils {
+    @Test
+    fun testSetNumThreads() {
+        TorchTensorLongAlgebra {
+            testingSetNumThreads()
+        }
+    }
+
+    @Test
+    fun testSeedSetting() = TorchTensorFloatAlgebra {
+        withCuda {
+            device -> testingSetSeed(device)
+        }
+    }
+}

kmath-torchscript/plugin/build.gradle.kts

@@ -0,0 +1,12 @@
+plugins {
+    id("ru.mipt.npm.gradle.project")
+}
+
+allprojects {
+    repositories {
+        gradlePluginPortal()
+        mavenCentral()
+        maven("https://maven.google.com")
+        maven("https://plugins.gradle.org/m2/")
+    }
+}

kmath-torchscript/plugin/compiler-plugin/build.gradle.kts

@@ -0,0 +1,8 @@
+plugins {
+    kotlin("jvm")
+    id("ru.mipt.npm.gradle.common")
+}
+
+dependencies {
+    compileOnly(kotlin("compiler-embeddable"))
+}

kmath-torchscript/plugin/compiler-plugin/src/main/kotlin/space/kscience/kmath/torchscript/compiler/TorchscriptPlugin.kt

@@ -0,0 +1,30 @@
+/*
+ * Copyright 2018-2021 KMath contributors.
+ * Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+ */
+
+package space.kscience.kmath.torchscript.compiler
+
+import org.jetbrains.kotlin.backend.common.extensions.IrGenerationExtension
+import org.jetbrains.kotlin.backend.common.extensions.IrPluginContext
+import org.jetbrains.kotlin.com.intellij.mock.MockProject
+import org.jetbrains.kotlin.compiler.plugin.ComponentRegistrar
+import org.jetbrains.kotlin.config.CompilerConfiguration
+import org.jetbrains.kotlin.ir.declarations.IrModuleFragment
+
+public class TorchscriptPluginComponentRegistrar : ComponentRegistrar {
+    public override fun registerProjectComponents(project: MockProject, configuration: CompilerConfiguration): Unit =
+        registerExtensions(project)
+
+    private companion object {
+        private fun registerExtensions(project: MockProject) {
+            IrGenerationExtension.registerExtension(project, TorchscriptExtension)
+        }
+    }
+}
+
+public object TorchscriptExtension : IrGenerationExtension {
+    override fun generate(moduleFragment: IrModuleFragment, pluginContext: IrPluginContext) {
+        TODO()
+    }
+}

kmath-torchscript/plugin/compiler-plugin/src/main/kotlin/space/kscience/kmath/torchscript/compiler/resolve/namingConventions.kt

@@ -0,0 +1,13 @@
+/*
+ * Copyright 2018-2021 KMath contributors.
+ * Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+ */
+
+package space.kscience.kmath.torchscript.compiler.resolve
+
+import org.jetbrains.kotlin.name.FqName
+
+public object TorchscriptAnnotations {
+    public val moduleAnnotationFqName: FqName = FqName("space.kscience.kmath.torchscript.Module")
+    public val intrinsicAnnotationFqName: FqName = FqName("space.kscience.kmath.torchscript.Intrinsic")
+}

kmath-torchscript/plugin/compiler-plugin/src/main/resources/META-INF/services/org.jetbrains.kotlin.compiler.plugin.ComponentRegistrar

@@ -0,0 +1,6 @@
+#
+# Copyright 2018-2021 KMath contributors.
+# Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+#
+
+space.kscience.kmath.torchscript.compiler.TorchscriptPluginComponentRegistrar

kmath-torchscript/plugin/gradle-plugin/build.gradle.kts

@@ -0,0 +1,19 @@
+@file:Suppress("UNUSED_VARIABLE")
+
+plugins {
+    `kotlin-dsl`
+    kotlin("kapt")
+}
+
+dependencies {
+    compileOnly("com.google.auto.service:auto-service-annotations:1.0")
+    compileOnly(kotlin("gradle-plugin"))
+    kapt("com.google.auto.service:auto-service:1.0")
+}
+
+gradlePlugin {
+    val kmathTorchScriptPlugin by plugins.registering {
+        id = "space.kscience.kmath.torchscript.torchscript-compiler-plugin"
+        implementationClass = "space.kscience.kmath.torchscript.gradle.TorchScriptKotlinGradleSubplugin"
+    }
+}

kmath-torchscript/plugin/gradle-plugin/src/main/kotlin/space/kscience/kmath/torchscript/gradle/TorchScriptKotlinGradleSubplugin.kt

@@ -0,0 +1,9 @@
+/*
+ * Copyright 2018-2021 KMath contributors.
+ * Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+ */
+
+package space.kscience.kmath.torchscript.gradle
+
+class TorchScriptKotlinGradleSubplugin {
+}

kmath-torchscript/plugin/settings.gradle.kts

@@ -0,0 +1,19 @@
+rootProject.name = "plugin"
+
+pluginManagement {
+    repositories {
+        maven("https://repo.kotlin.link")
+        mavenCentral()
+        gradlePluginPortal()
+    }
+
+    val toolsVersion = "0.10.0"
+
+    plugins {
+        id("ru.mipt.npm.gradle.project") version toolsVersion
+        id("ru.mipt.npm.gradle.mpp") version toolsVersion
+        id("ru.mipt.npm.gradle.jvm") version toolsVersion
+    }
+}
+
+include(":compiler-plugin", ":gradle-plugin")

kmath-torchscript/runtime/build.gradle.kts

@@ -0,0 +1,5 @@
+plugins {
+    kotlin("multiplatform")
+    id("ru.mipt.npm.gradle.common")
+}
+

kmath-torchscript/runtime/src/commonMain/kotlin/space.kscience.kmath.torchscript/annotations.kt

@@ -0,0 +1,10 @@
+/*
+ * Copyright 2018-2021 KMath contributors.
+ * Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+ */
+
+package space.kscience.kmath.torchscript
+
+public annotation class Module
+
+public annotation class Intrinsic

settings.gradle.kts

@@ -46,3 +46,11 @@ include(
     ":examples",
     ":benchmarks",
 )
+
+if (System.getProperty("os.name") == "Linux") {
+    include(":kmath-torch")
+    include(":kmath-torchscript-library")
+    include(":kmath-torchscript:example")
+    include(":kmath-torchscript:runtime")
+    includeBuild("kmath-torchscript/plugin")
+}