Fixing 2D and 1D casts
This commit is contained in:
Roland Grinis
parent
22b68e5ca4
commit
92710097f0
@ -2731,8 +2731,6 @@ public class space/kscience/kmath/tensors/core/BufferedTensor : space/kscience/k
|
||||
public fun <init> ([ILspace/kscience/kmath/structures/MutableBuffer;I)V
|
||||
public fun elements ()Lkotlin/sequences/Sequence;
|
||||
public fun equals (Ljava/lang/Object;)Z
|
||||
public final fun forEachMatrix (Lkotlin/jvm/functions/Function1;)V
|
||||
public final fun forEachVector (Lkotlin/jvm/functions/Function1;)V
|
||||
public fun get ([I)Ljava/lang/Object;
|
||||
public final fun getBuffer ()Lspace/kscience/kmath/structures/MutableBuffer;
|
||||
public fun getDimension ()I
|
||||
@ -2740,37 +2738,7 @@ public class space/kscience/kmath/tensors/core/BufferedTensor : space/kscience/k
|
||||
public final fun getNumel ()I
|
||||
public fun getShape ()[I
|
||||
public fun hashCode ()I
|
||||
public final fun matrixSequence ()Lkotlin/sequences/Sequence;
|
||||
public fun set ([ILjava/lang/Object;)V
|
||||
public final fun vectorSequence ()Lkotlin/sequences/Sequence;
|
||||
}
|
||||
|
||||
public final class space/kscience/kmath/tensors/core/BufferedTensor1D : space/kscience/kmath/tensors/core/BufferedTensor, space/kscience/kmath/nd/MutableStructure1D {
|
||||
public fun copy ()Lspace/kscience/kmath/structures/MutableBuffer;
|
||||
public fun get (I)Ljava/lang/Object;
|
||||
public fun get ([I)Ljava/lang/Object;
|
||||
public fun getDimension ()I
|
||||
public fun getSize ()I
|
||||
public fun iterator ()Ljava/util/Iterator;
|
||||
public fun set (ILjava/lang/Object;)V
|
||||
public fun set ([ILjava/lang/Object;)V
|
||||
}
|
||||
|
||||
public final class space/kscience/kmath/tensors/core/BufferedTensor2D : space/kscience/kmath/tensors/core/BufferedTensor, space/kscience/kmath/nd/MutableStructure2D {
|
||||
public fun elements ()Lkotlin/sequences/Sequence;
|
||||
public fun get (II)Ljava/lang/Object;
|
||||
public fun get ([I)Ljava/lang/Object;
|
||||
public fun getColNum ()I
|
||||
public fun getColumns ()Ljava/util/List;
|
||||
public fun getRowNum ()I
|
||||
public fun getRows ()Ljava/util/List;
|
||||
public fun getShape ()[I
|
||||
public fun set (IILjava/lang/Object;)V
|
||||
}
|
||||
|
||||
public final class space/kscience/kmath/tensors/core/BufferedTensorKt {
|
||||
public static final fun as1D (Lspace/kscience/kmath/tensors/core/BufferedTensor;)Lspace/kscience/kmath/tensors/core/BufferedTensor1D;
|
||||
public static final fun as2D (Lspace/kscience/kmath/tensors/core/BufferedTensor;)Lspace/kscience/kmath/tensors/core/BufferedTensor2D;
|
||||
}
|
||||
|
||||
public final class space/kscience/kmath/tensors/core/DoubleAnalyticTensorAlgebra : space/kscience/kmath/tensors/core/DoubleTensorAlgebra, space/kscience/kmath/tensors/AnalyticTensorAlgebra {
|
||||
@ -2876,7 +2844,6 @@ public class space/kscience/kmath/tensors/core/DoubleTensorAlgebra : space/kscie
|
||||
public synthetic fun eq (Lspace/kscience/kmath/nd/MutableStructureND;Lspace/kscience/kmath/nd/MutableStructureND;Ljava/lang/Object;)Z
|
||||
public final fun eq (Lspace/kscience/kmath/tensors/core/DoubleTensor;Lspace/kscience/kmath/tensors/core/DoubleTensor;)Z
|
||||
public fun eq (Lspace/kscience/kmath/tensors/core/DoubleTensor;Lspace/kscience/kmath/tensors/core/DoubleTensor;D)Z
|
||||
public final fun eq (Lspace/kscience/kmath/tensors/core/DoubleTensor;Lspace/kscience/kmath/tensors/core/DoubleTensor;Lkotlin/jvm/functions/Function2;)Z
|
||||
public synthetic fun eye (I)Lspace/kscience/kmath/nd/MutableStructureND;
|
||||
public fun eye (I)Lspace/kscience/kmath/tensors/core/DoubleTensor;
|
||||
public final fun fromArray ([I[D)Lspace/kscience/kmath/tensors/core/DoubleTensor;
|
||||
|
||||
@ -35,35 +35,34 @@ public open class BufferedTensor<T>(
|
||||
|
||||
override fun hashCode(): Int = 0
|
||||
|
||||
public fun vectorSequence(): Sequence<BufferedTensor1D<T>> = sequence {
|
||||
check(shape.size >= 1) { "todo" }
|
||||
internal fun vectorSequence(): Sequence<BufferedTensor<T>> = sequence {
|
||||
val n = shape.size
|
||||
val vectorOffset = shape[n - 1]
|
||||
val vectorShape = intArrayOf(shape.last())
|
||||
for (offset in 0 until numel step vectorOffset) {
|
||||
val vector = BufferedTensor<T>(vectorShape, buffer, offset).as1D()
|
||||
val vector = BufferedTensor(vectorShape, buffer, offset)
|
||||
yield(vector)
|
||||
}
|
||||
}
|
||||
|
||||
public fun matrixSequence(): Sequence<BufferedTensor2D<T>> = sequence {
|
||||
internal fun matrixSequence(): Sequence<BufferedTensor<T>> = sequence {
|
||||
check(shape.size >= 2) { "todo" }
|
||||
val n = shape.size
|
||||
val matrixOffset = shape[n - 1] * shape[n - 2]
|
||||
val matrixShape = intArrayOf(shape[n - 2], shape[n - 1]) //todo better way?
|
||||
val matrixShape = intArrayOf(shape[n - 2], shape[n - 1])
|
||||
for (offset in 0 until numel step matrixOffset) {
|
||||
val matrix = BufferedTensor<T>(matrixShape, buffer, offset).as2D()
|
||||
val matrix = BufferedTensor(matrixShape, buffer, offset)
|
||||
yield(matrix)
|
||||
}
|
||||
}
|
||||
|
||||
public inline fun forEachVector(vectorAction: (BufferedTensor1D<T>) -> Unit): Unit {
|
||||
internal inline fun forEachVector(vectorAction: (BufferedTensor<T>) -> Unit): Unit {
|
||||
for (vector in vectorSequence()) {
|
||||
vectorAction(vector)
|
||||
}
|
||||
}
|
||||
|
||||
public inline fun forEachMatrix(matrixAction: (BufferedTensor2D<T>) -> Unit): Unit {
|
||||
internal inline fun forEachMatrix(matrixAction: (BufferedTensor<T>) -> Unit): Unit {
|
||||
for (matrix in matrixSequence()) {
|
||||
matrixAction(matrix)
|
||||
}
|
||||
@ -71,7 +70,6 @@ public open class BufferedTensor<T>(
|
||||
|
||||
}
|
||||
|
||||
|
||||
public class IntTensor internal constructor(
|
||||
shape: IntArray,
|
||||
buffer: IntArray,
|
||||
@ -112,90 +110,7 @@ public class DoubleTensor internal constructor(
|
||||
this(bufferedTensor.shape, bufferedTensor.buffer.array(), bufferedTensor.bufferStart)
|
||||
}
|
||||
|
||||
|
||||
public class BufferedTensor2D<T> internal constructor(
|
||||
private val tensor: BufferedTensor<T>,
|
||||
) : BufferedTensor<T>(tensor), MutableStructure2D<T> {
|
||||
init {
|
||||
check(shape.size == 2) {
|
||||
"Shape ${shape.toList()} not compatible with DoubleTensor2D"
|
||||
}
|
||||
}
|
||||
|
||||
override val shape: IntArray
|
||||
get() = tensor.shape
|
||||
|
||||
override val rowNum: Int
|
||||
get() = shape[0]
|
||||
override val colNum: Int
|
||||
get() = shape[1]
|
||||
|
||||
override fun get(i: Int, j: Int): T = tensor[intArrayOf(i, j)]
|
||||
|
||||
override fun get(index: IntArray): T = tensor[index]
|
||||
|
||||
override fun elements(): Sequence<Pair<IntArray, T>> = tensor.elements()
|
||||
|
||||
override fun set(i: Int, j: Int, value: T) {
|
||||
tensor[intArrayOf(i, j)] = value
|
||||
}
|
||||
|
||||
override val rows: List<BufferedTensor1D<T>>
|
||||
get() = List(rowNum) { i ->
|
||||
BufferedTensor1D(
|
||||
BufferedTensor(
|
||||
shape = intArrayOf(colNum),
|
||||
buffer = VirtualMutableBuffer(colNum) { j -> get(i, j) },
|
||||
bufferStart = 0
|
||||
)
|
||||
)
|
||||
}
|
||||
|
||||
override val columns: List<BufferedTensor1D<T>>
|
||||
get() = List(colNum) { j ->
|
||||
BufferedTensor1D(
|
||||
BufferedTensor(
|
||||
shape = intArrayOf(rowNum),
|
||||
buffer = VirtualMutableBuffer(rowNum) { i -> get(i, j) },
|
||||
bufferStart = 0
|
||||
)
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
public class BufferedTensor1D<T> internal constructor(
|
||||
private val tensor: BufferedTensor<T>
|
||||
) : BufferedTensor<T>(tensor), MutableStructure1D<T> {
|
||||
init {
|
||||
check(shape.size == 1) {
|
||||
"Shape ${shape.toList()} not compatible with DoubleTensor1D"
|
||||
}
|
||||
}
|
||||
|
||||
override fun get(index: IntArray): T = tensor[index]
|
||||
|
||||
override fun set(index: IntArray, value: T) {
|
||||
tensor[index] = value
|
||||
}
|
||||
|
||||
override val size: Int
|
||||
get() = tensor.linearStructure.size
|
||||
|
||||
override fun get(index: Int): T = tensor[intArrayOf(index)]
|
||||
|
||||
override fun set(index: Int, value: T) {
|
||||
tensor[intArrayOf(index)] = value
|
||||
}
|
||||
|
||||
override fun copy(): MutableBuffer<T> = tensor.buffer.copy()
|
||||
|
||||
}
|
||||
|
||||
internal fun BufferedTensor<Int>.asIntTensor(): IntTensor = IntTensor(this)
|
||||
internal fun BufferedTensor<Long>.asLongTensor(): LongTensor = LongTensor(this)
|
||||
internal fun BufferedTensor<Float>.asFloatTensor(): FloatTensor = FloatTensor(this)
|
||||
internal fun BufferedTensor<Double>.asDoubleTensor(): DoubleTensor = DoubleTensor(this)
|
||||
|
||||
|
||||
public fun <T> BufferedTensor<T>.as2D(): BufferedTensor2D<T> = BufferedTensor2D(this)
|
||||
public fun <T> BufferedTensor<T>.as1D(): BufferedTensor1D<T> = BufferedTensor1D(this)
|
||||
internal fun BufferedTensor<Int>.asTensor(): IntTensor = IntTensor(this)
|
||||
internal fun BufferedTensor<Long>.asTensor(): LongTensor = LongTensor(this)
|
||||
internal fun BufferedTensor<Float>.asTensor(): FloatTensor = FloatTensor(this)
|
||||
internal fun BufferedTensor<Double>.asTensor(): DoubleTensor = DoubleTensor(this)
|
||||
|
||||
@ -1,5 +1,9 @@
|
||||
package space.kscience.kmath.tensors.core
|
||||
|
||||
import space.kscience.kmath.nd.MutableStructure1D
|
||||
import space.kscience.kmath.nd.MutableStructure2D
|
||||
import space.kscience.kmath.nd.as1D
|
||||
import space.kscience.kmath.nd.as2D
|
||||
import space.kscience.kmath.tensors.LinearOpsTensorAlgebra
|
||||
import kotlin.math.sqrt
|
||||
|
||||
@ -11,6 +15,48 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
|
||||
override fun DoubleTensor.det(): DoubleTensor = detLU()
|
||||
|
||||
private inline fun luHelper(lu: MutableStructure2D<Double>, pivots: MutableStructure1D<Int>, m: Int) {
|
||||
for (row in 0 until m) pivots[row] = row
|
||||
|
||||
for (i in 0 until m) {
|
||||
var maxVal = -1.0
|
||||
var maxInd = i
|
||||
|
||||
for (k in i until m) {
|
||||
val absA = kotlin.math.abs(lu[k, i])
|
||||
if (absA > maxVal) {
|
||||
maxVal = absA
|
||||
maxInd = k
|
||||
}
|
||||
}
|
||||
|
||||
//todo check singularity
|
||||
|
||||
if (maxInd != i) {
|
||||
|
||||
val j = pivots[i]
|
||||
pivots[i] = pivots[maxInd]
|
||||
pivots[maxInd] = j
|
||||
|
||||
for (k in 0 until m) {
|
||||
val tmp = lu[i, k]
|
||||
lu[i, k] = lu[maxInd, k]
|
||||
lu[maxInd, k] = tmp
|
||||
}
|
||||
|
||||
pivots[m] += 1
|
||||
|
||||
}
|
||||
|
||||
for (j in i + 1 until m) {
|
||||
lu[j, i] /= lu[i, i]
|
||||
for (k in i + 1 until m) {
|
||||
lu[j, k] -= lu[j, i] * lu[i, k]
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
override fun DoubleTensor.lu(): Pair<DoubleTensor, IntTensor> {
|
||||
|
||||
checkSquareMatrix(shape)
|
||||
@ -27,90 +73,93 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
IntArray(pivotsShape.reduce(Int::times)) { 0 }
|
||||
)
|
||||
|
||||
for ((lu, pivots) in luTensor.matrixSequence().zip(pivotsTensor.vectorSequence())){
|
||||
for (row in 0 until m) pivots[row] = row
|
||||
|
||||
for (i in 0 until m) {
|
||||
var maxVal = -1.0
|
||||
var maxInd = i
|
||||
|
||||
for (k in i until m) {
|
||||
val absA = kotlin.math.abs(lu[k, i])
|
||||
if (absA > maxVal) {
|
||||
maxVal = absA
|
||||
maxInd = k
|
||||
}
|
||||
}
|
||||
|
||||
//todo check singularity
|
||||
|
||||
if (maxInd != i) {
|
||||
|
||||
val j = pivots[i]
|
||||
pivots[i] = pivots[maxInd]
|
||||
pivots[maxInd] = j
|
||||
|
||||
for (k in 0 until m) {
|
||||
val tmp = lu[i, k]
|
||||
lu[i, k] = lu[maxInd, k]
|
||||
lu[maxInd, k] = tmp
|
||||
}
|
||||
|
||||
pivots[m] += 1
|
||||
|
||||
}
|
||||
|
||||
for (j in i + 1 until m) {
|
||||
lu[j, i] /= lu[i, i]
|
||||
for (k in i + 1 until m) {
|
||||
lu[j, k] -= lu[j, i] * lu[i, k]
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
for ((lu, pivots) in luTensor.matrixSequence().zip(pivotsTensor.vectorSequence()))
|
||||
luHelper(lu.as2D(), pivots.as1D(), m)
|
||||
|
||||
return Pair(luTensor, pivotsTensor)
|
||||
|
||||
}
|
||||
|
||||
override fun luPivot(luTensor: DoubleTensor, pivotsTensor: IntTensor): Triple<DoubleTensor, DoubleTensor, DoubleTensor> {
|
||||
private inline fun pivInit(
|
||||
p: MutableStructure2D<Double>,
|
||||
pivot: MutableStructure1D<Int>,
|
||||
n: Int
|
||||
) {
|
||||
for (i in 0 until n) {
|
||||
p[i, pivot[i]] = 1.0
|
||||
}
|
||||
}
|
||||
|
||||
private inline fun luPivotHelper(
|
||||
l: MutableStructure2D<Double>,
|
||||
u: MutableStructure2D<Double>,
|
||||
lu: MutableStructure2D<Double>,
|
||||
n: Int
|
||||
) {
|
||||
for (i in 0 until n) {
|
||||
for (j in 0 until n) {
|
||||
if (i == j) {
|
||||
l[i, j] = 1.0
|
||||
}
|
||||
if (j < i) {
|
||||
l[i, j] = lu[i, j]
|
||||
}
|
||||
if (j >= i) {
|
||||
u[i, j] = lu[i, j]
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
override fun luPivot(
|
||||
luTensor: DoubleTensor,
|
||||
pivotsTensor: IntTensor
|
||||
): Triple<DoubleTensor, DoubleTensor, DoubleTensor> {
|
||||
//todo checks
|
||||
checkSquareMatrix(luTensor.shape)
|
||||
check(luTensor.shape.dropLast(1).toIntArray() contentEquals pivotsTensor.shape) { "Bed shapes (("} //todo rewrite
|
||||
check(
|
||||
luTensor.shape.dropLast(1).toIntArray() contentEquals pivotsTensor.shape
|
||||
) { "Bed shapes ((" } //todo rewrite
|
||||
|
||||
val n = luTensor.shape.last()
|
||||
val pTensor = luTensor.zeroesLike()
|
||||
for ((p, pivot) in pTensor.matrixSequence().zip(pivotsTensor.vectorSequence())){
|
||||
for (i in 0 until n){
|
||||
p[i, pivot[i]] = 1.0
|
||||
}
|
||||
}
|
||||
for ((p, pivot) in pTensor.matrixSequence().zip(pivotsTensor.vectorSequence()))
|
||||
pivInit(p.as2D(), pivot.as1D(), n)
|
||||
|
||||
val lTensor = luTensor.zeroesLike()
|
||||
val uTensor = luTensor.zeroesLike()
|
||||
|
||||
for ((pairLU, lu) in lTensor.matrixSequence().zip(uTensor.matrixSequence()).zip(luTensor.matrixSequence())){
|
||||
for ((pairLU, lu) in lTensor.matrixSequence().zip(uTensor.matrixSequence())
|
||||
.zip(luTensor.matrixSequence())) {
|
||||
val (l, u) = pairLU
|
||||
for (i in 0 until n) {
|
||||
for (j in 0 until n) {
|
||||
if (i == j) {
|
||||
l[i, j] = 1.0
|
||||
}
|
||||
if (j < i) {
|
||||
l[i, j] = lu[i, j]
|
||||
}
|
||||
if (j >= i) {
|
||||
u[i, j] = lu[i, j]
|
||||
}
|
||||
}
|
||||
}
|
||||
luPivotHelper(l.as2D(), u.as2D(), lu.as2D(), n)
|
||||
}
|
||||
|
||||
return Triple(pTensor, lTensor, uTensor)
|
||||
|
||||
}
|
||||
|
||||
private inline fun choleskyHelper(
|
||||
a: MutableStructure2D<Double>,
|
||||
l: MutableStructure2D<Double>,
|
||||
n: Int
|
||||
) {
|
||||
for (i in 0 until n) {
|
||||
for (j in 0 until i) {
|
||||
var h = a[i, j]
|
||||
for (k in 0 until j) {
|
||||
h -= l[i, k] * l[j, k]
|
||||
}
|
||||
l[i, j] = h / l[j, j]
|
||||
}
|
||||
var h = a[i, i]
|
||||
for (j in 0 until i) {
|
||||
h -= l[i, j] * l[i, j]
|
||||
}
|
||||
l[i, i] = sqrt(h)
|
||||
}
|
||||
}
|
||||
|
||||
override fun DoubleTensor.cholesky(): DoubleTensor {
|
||||
// todo checks
|
||||
checkSquareMatrix(shape)
|
||||
@ -118,22 +167,8 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
val n = shape.last()
|
||||
val lTensor = zeroesLike()
|
||||
|
||||
for ((a, l) in this.matrixSequence().zip(lTensor.matrixSequence())) {
|
||||
for (i in 0 until n) {
|
||||
for (j in 0 until i) {
|
||||
var h = a[i, j]
|
||||
for (k in 0 until j) {
|
||||
h -= l[i, k] * l[j, k]
|
||||
}
|
||||
l[i, j] = h / l[j, j]
|
||||
}
|
||||
var h = a[i, i]
|
||||
for (j in 0 until i) {
|
||||
h -= l[i, j] * l[i, j]
|
||||
}
|
||||
l[i, i] = sqrt(h)
|
||||
}
|
||||
}
|
||||
for ((a, l) in this.matrixSequence().zip(lTensor.matrixSequence()))
|
||||
for (i in 0 until n) choleskyHelper(a.as2D(), l.as2D(), n)
|
||||
|
||||
return lTensor
|
||||
}
|
||||
@ -150,9 +185,11 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
TODO("ANDREI")
|
||||
}
|
||||
|
||||
private fun luMatrixDet(lu: BufferedTensor2D<Double>, pivots: BufferedTensor1D<Int>): Double {
|
||||
private fun luMatrixDet(luTensor: MutableStructure2D<Double>, pivotsTensor: MutableStructure1D<Int>): Double {
|
||||
val lu = luTensor.as2D()
|
||||
val pivots = pivotsTensor.as1D()
|
||||
val m = lu.shape[0]
|
||||
val sign = if((pivots[m] - m) % 2 == 0) 1.0 else -1.0
|
||||
val sign = if ((pivots[m] - m) % 2 == 0) 1.0 else -1.0
|
||||
return (0 until m).asSequence().map { lu[it, it] }.fold(sign) { left, right -> left * right }
|
||||
}
|
||||
|
||||
@ -162,34 +199,34 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
|
||||
val detTensorShape = IntArray(n - 1) { i -> shape[i] }
|
||||
detTensorShape[n - 2] = 1
|
||||
val resBuffer = DoubleArray(detTensorShape.reduce(Int::times)) { 0.0 }
|
||||
val resBuffer = DoubleArray(detTensorShape.reduce(Int::times)) { 0.0 }
|
||||
|
||||
val detTensor = DoubleTensor(
|
||||
detTensorShape,
|
||||
resBuffer
|
||||
)
|
||||
|
||||
luTensor.matrixSequence().zip(pivotsTensor.vectorSequence()).forEachIndexed { index, (luMatrix, pivots) ->
|
||||
resBuffer[index] = luMatrixDet(luMatrix, pivots)
|
||||
luTensor.matrixSequence().zip(pivotsTensor.vectorSequence()).forEachIndexed { index, (lu, pivots) ->
|
||||
resBuffer[index] = luMatrixDet(lu.as2D(), pivots.as1D())
|
||||
}
|
||||
|
||||
return detTensor
|
||||
}
|
||||
|
||||
private fun luMatrixInv(
|
||||
lu: BufferedTensor2D<Double>,
|
||||
pivots: BufferedTensor1D<Int>,
|
||||
invMatrix : BufferedTensor2D<Double>
|
||||
): Unit {
|
||||
lu: MutableStructure2D<Double>,
|
||||
pivots: MutableStructure1D<Int>,
|
||||
invMatrix: MutableStructure2D<Double>
|
||||
) {
|
||||
val m = lu.shape[0]
|
||||
|
||||
for (j in 0 until m) {
|
||||
for (i in 0 until m) {
|
||||
if (pivots[i] == j){
|
||||
if (pivots[i] == j) {
|
||||
invMatrix[i, j] = 1.0
|
||||
}
|
||||
|
||||
for (k in 0 until i){
|
||||
for (k in 0 until i) {
|
||||
invMatrix[i, j] -= lu[i, k] * invMatrix[k, j]
|
||||
}
|
||||
}
|
||||
@ -205,13 +242,12 @@ public class DoubleLinearOpsTensorAlgebra :
|
||||
|
||||
public fun DoubleTensor.invLU(): DoubleTensor {
|
||||
val (luTensor, pivotsTensor) = lu()
|
||||
val n = shape.size
|
||||
val invTensor = luTensor.zeroesLike()
|
||||
|
||||
val seq = luTensor.matrixSequence().zip(pivotsTensor.vectorSequence()).zip(invTensor.matrixSequence())
|
||||
for ((luP, invMatrix) in seq) {
|
||||
val (lu, pivots) = luP
|
||||
luMatrixInv(lu, pivots, invMatrix)
|
||||
luMatrixInv(lu.as2D(), pivots.as1D(), invMatrix.as2D())
|
||||
}
|
||||
|
||||
return invTensor
|
||||
|
||||
@ -1,8 +1,7 @@
|
||||
package space.kscience.kmath.tensors.core
|
||||
|
||||
import space.kscience.kmath.linear.Matrix
|
||||
import space.kscience.kmath.nd.MutableStructure2D
|
||||
import space.kscience.kmath.nd.Structure2D
|
||||
import space.kscience.kmath.nd.as2D
|
||||
import space.kscience.kmath.tensors.TensorPartialDivisionAlgebra
|
||||
import kotlin.math.abs
|
||||
|
||||
@ -224,6 +223,23 @@ public open class DoubleTensorAlgebra : TensorPartialDivisionAlgebra<Double, Dou
|
||||
return this.view(other.shape)
|
||||
}
|
||||
|
||||
private inline fun dotHelper(
|
||||
a: MutableStructure2D<Double>,
|
||||
b: MutableStructure2D<Double>,
|
||||
res: MutableStructure2D<Double>,
|
||||
l: Int, m: Int, n: Int
|
||||
) {
|
||||
for (i in 0 until l) {
|
||||
for (j in 0 until n) {
|
||||
var curr = 0.0
|
||||
for (k in 0 until m) {
|
||||
curr += a[i, k] * b[k, j]
|
||||
}
|
||||
res[i, j] = curr
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
override fun DoubleTensor.dot(other: DoubleTensor): DoubleTensor {
|
||||
if (this.shape.size == 1 && other.shape.size == 1) {
|
||||
return DoubleTensor(intArrayOf(1), doubleArrayOf(this.times(other).buffer.array().sum()))
|
||||
@ -240,7 +256,7 @@ public open class DoubleTensorAlgebra : TensorPartialDivisionAlgebra<Double, Dou
|
||||
}
|
||||
if (other.shape.size == 1) {
|
||||
lastDim = true
|
||||
newOther = other.view(other.shape + intArrayOf(1) )
|
||||
newOther = other.view(other.shape + intArrayOf(1))
|
||||
}
|
||||
|
||||
val broadcastTensors = broadcastOuterTensors(newThis, newOther)
|
||||
@ -248,7 +264,7 @@ public open class DoubleTensorAlgebra : TensorPartialDivisionAlgebra<Double, Dou
|
||||
newOther = broadcastTensors[1]
|
||||
|
||||
val l = newThis.shape[newThis.shape.size - 2]
|
||||
val m1= newThis.shape[newThis.shape.size - 1]
|
||||
val m1 = newThis.shape[newThis.shape.size - 1]
|
||||
val m2 = newOther.shape[newOther.shape.size - 2]
|
||||
val n = newOther.shape[newOther.shape.size - 1]
|
||||
if (m1 != m2) {
|
||||
@ -262,21 +278,14 @@ public open class DoubleTensorAlgebra : TensorPartialDivisionAlgebra<Double, Dou
|
||||
|
||||
for ((res, ab) in resTensor.matrixSequence().zip(newThis.matrixSequence().zip(newOther.matrixSequence()))) {
|
||||
val (a, b) = ab
|
||||
|
||||
for (i in 0 until l) {
|
||||
for (j in 0 until n) {
|
||||
var curr = 0.0
|
||||
for (k in 0 until m) {
|
||||
curr += a[i, k] * b[k, j]
|
||||
}
|
||||
res[i, j] = curr
|
||||
}
|
||||
}
|
||||
dotHelper(a.as2D(), b.as2D(), res.as2D(), l, m, n)
|
||||
}
|
||||
|
||||
if (penultimateDim) {
|
||||
return resTensor.view(resTensor.shape.dropLast(2).toIntArray() +
|
||||
intArrayOf(resTensor.shape.last()))
|
||||
return resTensor.view(
|
||||
resTensor.shape.dropLast(2).toIntArray() +
|
||||
intArrayOf(resTensor.shape.last())
|
||||
)
|
||||
}
|
||||
if (lastDim) {
|
||||
return resTensor.view(resTensor.shape.dropLast(1).toIntArray())
|
||||
@ -307,15 +316,11 @@ public open class DoubleTensorAlgebra : TensorPartialDivisionAlgebra<Double, Dou
|
||||
|
||||
public fun DoubleTensor.eq(other: DoubleTensor): Boolean = this.eq(other, 1e-5)
|
||||
|
||||
public fun DoubleTensor.contentEquals(other: DoubleTensor, eqFunction: (Double, Double) -> Boolean): Boolean {
|
||||
if (!(this.shape contentEquals other.shape)) {
|
||||
return false
|
||||
}
|
||||
return this.eq(other, eqFunction)
|
||||
}
|
||||
public fun DoubleTensor.contentEquals(other: DoubleTensor, eqFunction: (Double, Double) -> Boolean): Boolean =
|
||||
this.eq(other, eqFunction)
|
||||
|
||||
public fun DoubleTensor.eq(other: DoubleTensor, eqFunction: (Double, Double) -> Boolean): Boolean {
|
||||
// todo broadcasting checking
|
||||
private fun DoubleTensor.eq(other: DoubleTensor, eqFunction: (Double, Double) -> Boolean): Boolean {
|
||||
checkShapesCompatible(this, other)
|
||||
val n = this.linearStructure.size
|
||||
if (n != other.linearStructure.size) {
|
||||
return false
|
||||
|
||||
@ -11,14 +11,14 @@ internal inline fun <T, TensorType : TensorStructure<T>,
|
||||
"Illegal empty shape provided"
|
||||
}
|
||||
|
||||
internal inline fun < TensorType : TensorStructure<Double>,
|
||||
internal inline fun <TensorType : TensorStructure<Double>,
|
||||
TorchTensorAlgebraType : TensorAlgebra<Double, TensorType>>
|
||||
TorchTensorAlgebraType.checkEmptyDoubleBuffer(buffer: DoubleArray): Unit =
|
||||
check(buffer.isNotEmpty()) {
|
||||
"Illegal empty buffer provided"
|
||||
}
|
||||
|
||||
internal inline fun < TensorType : TensorStructure<Double>,
|
||||
internal inline fun <TensorType : TensorStructure<Double>,
|
||||
TorchTensorAlgebraType : TensorAlgebra<Double, TensorType>>
|
||||
TorchTensorAlgebraType.checkBufferShapeConsistency(shape: IntArray, buffer: DoubleArray): Unit =
|
||||
check(buffer.size == shape.reduce(Int::times)) {
|
||||
@ -56,4 +56,4 @@ internal inline fun <T, TensorType : TensorStructure<T>,
|
||||
check(shape[n - 1] == shape[n - 2]) {
|
||||
"Tensor must be batches of square matrices, but they are ${shape[n - 1]} by ${shape[n - 1]} matrices"
|
||||
}
|
||||
}
|
||||
}
|
||||
@ -1,5 +1,7 @@
|
||||
package space.kscience.kmath.tensors.core
|
||||
|
||||
import space.kscience.kmath.nd.as1D
|
||||
import space.kscience.kmath.nd.as2D
|
||||
import space.kscience.kmath.structures.toDoubleArray
|
||||
import kotlin.test.Test
|
||||
import kotlin.test.assertEquals
|
||||
|
||||
Loading…
Reference in New Issue
Block a user