teldufalsari/kmath
2
0
Fork 0
forked from kscience/kmath
Code Pull Requests 1 Activity

Merge remote-tracking branch 'upstream/zelenyy' into zelenyy

Browse Source 
# Conflicts:
#	kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/CreationRoutines.kt
This commit is contained in:
Mikhail Zelenyy 2019-01-23 14:13:37 +03:00
commit ba230d0de1
5 changed files with 88 additions and 69 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/NDAlgebra.kt
View File

@ -14,7 +14,7 @@ class ShapeMismatchException(val expected: IntArray, val actual: IntArray) : Run
/**
* The base interface for all nd-algebra implementations
* @param T the type of nd-structure element
* @param C the type of the context
* @param C the type of the element context
* @param N the type of the structure
*/
interface NDAlgebra<T, C, N : NDStructure<T>> {
@ -112,10 +112,13 @@ interface NDField<T, F : Field<T>, N : NDStructure<T>> : Field<N>, NDRing<T, F,
operator fun T.div(arg: N) = map(arg) { divide(it, this@div) }
companion object {
private val realNDFieldCache = HashMap<IntArray, RealNDField>()
/**
* Create a nd-field for [Double] values
* Create a nd-field for [Double] values or pull it from cache if it was created previously
*/
fun real(shape: IntArray) = RealNDField(shape)
fun real(shape: IntArray) = realNDFieldCache.getOrPut(shape){RealNDField(shape)}
/**
* Create a nd-field with boxing generic buffer

23
kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/NDElement.kt
View File

@ -7,6 +7,9 @@ import scientifik.kmath.operations.Space
/**
* The root for all [NDStructure] based algebra elements. Does not implement algebra element root because of problems with recursive self-types
* @param T the type of the element of the structure
* @param C the type of the context for the element
* @param N the type of the underlying [NDStructure]
*/
interface NDElement<T, C, N : NDStructure<T>> : NDStructure<T> {
@ -16,9 +19,6 @@ interface NDElement<T, C, N : NDStructure<T>> : NDStructure<T> {
fun N.wrap(): NDElement<T, C, N>
fun mapIndexed(transform: C.(index: IntArray, T) -> T) = context.mapIndexed(unwrap(), transform).wrap()
fun map(transform: C.(T) -> T) = context.map(unwrap(), transform).wrap()
companion object {
/**
* Create a optimized NDArray of doubles
@ -61,10 +61,17 @@ interface NDElement<T, C, N : NDStructure<T>> : NDStructure<T> {
}
}
fun <T, C, N : NDStructure<T>> NDElement<T, C, N>.mapIndexed(transform: C.(index: IntArray, T) -> T) =
context.mapIndexed(unwrap(), transform).wrap()
fun <T, C, N : NDStructure<T>> NDElement<T, C, N>.map(transform: C.(T) -> T) = context.map(unwrap(), transform).wrap()
/**
* Element by element application of any operation on elements to the whole [NDElement]
*/
operator fun <T, C> Function1<T, T>.invoke(ndElement: NDElement<T, C, *>) =
operator fun <T, C, N : NDStructure<T>> Function1<T, T>.invoke(ndElement: NDElement<T, C, N>) =
ndElement.map { value -> this@invoke(value) }
/* plus and minus */
@ -72,13 +79,13 @@ operator fun <T, C> Function1<T, T>.invoke(ndElement: NDElement<T, C, *>) =
/**
* Summation operation for [NDElement] and single element
*/
operator fun <T, S : Space<T>> NDElement<T, S, *>.plus(arg: T) =
operator fun <T, S : Space<T>, N : NDStructure<T>> NDElement<T, S, N>.plus(arg: T) =
map { value -> arg + value }
/**
* Subtraction operation between [NDElement] and single element
*/
operator fun <T, S : Space<T>> NDElement<T, S, *>.minus(arg: T) =
operator fun <T, S : Space<T>, N : NDStructure<T>> NDElement<T, S, N>.minus(arg: T) =
map { value -> arg - value }
/* prod and div */
@ -86,13 +93,13 @@ operator fun <T, S : Space<T>> NDElement<T, S, *>.minus(arg: T) =
/**
* Product operation for [NDElement] and single element
*/
operator fun <T, R : Ring<T>> NDElement<T, R, *>.times(arg: T) =
operator fun <T, R : Ring<T>, N : NDStructure<T>> NDElement<T, R, N>.times(arg: T) =
map { value -> arg * value }
/**
* Division operation between [NDElement] and single element
*/
operator fun <T, F : Field<T>> NDElement<T, F, *>.div(arg: T) =
operator fun <T, F : Field<T>, N : NDStructure<T>> NDElement<T, F, N>.div(arg: T) =
map { value -> arg / value }

102
kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/CreationRoutines.kt → kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/NDFactories.kt
View File

@ -1,14 +1,19 @@
package scientifik.kmath.structures
import scientifik.kmath.operations.RealField.power
import kotlin.math.*
import kotlin.math.ceil
import kotlin.math.log
import kotlin.math.min
import kotlin.math.sign
object RealFactory {
/**
* Numpy-like factories for [RealNDElement]
*/
object RealNDFactory {
/**
* Create a NDArray filled with ones
* Get a [RealNDElement] filled with [RealNDField.one]. Due to caching all instances with the same shape point to the same object
*/
fun ones(vararg shape: Int) = NDElement.real(shape) { 1.0 }
fun ones(vararg shape: Int) = NDField.real(shape).one
/**
* Create a 2D NDArray, with ones on the diagonal and zeros elsewhere.
@ -31,40 +36,37 @@ object RealFactory {
* Return evenly spaced values within a given interval.
*
* Values are generated within the half-open interval [start, stop) (in other words, the interval including start but excluding stop).
* @param range use it like:
* (start..stop) to step
*/
fun range(range: Pair<ClosedFloatingPointRange<Double>, Double>) =
NDElement.real1D(ceil((range.first.endInclusive - range.first.start) / range.second).toInt()) { i -> range.first.start + i * range.second }
fun range(range: ClosedFloatingPointRange<Double>, step: Double = 1.0) =
NDElement.real1D(ceil((range.endInclusive - range.start) / step).toInt()) { i ->
range.start + i * step
}
/**
* Return evenly spaced numbers over a specified interval.
* @param range use it like:
* (start..stop) to number
* start is starting value, finaly value depend from endPoint parameter
* @param range start is starting value, final value depend from endPoint parameter
* @param endPoint If True, right boundary of range is the last sample. Otherwise, it is not included.
*/
fun linSpace(
range: Pair<ClosedFloatingPointRange<Double>, Int>,
fun linspace(
range: ClosedFloatingPointRange<Double>,
num: Int = 100,
endPoint: Boolean = true
): Pair<RealNDElement, Double> {
val div = if (endPoint) (range.second - 1) else range.second
val delta = range.first.start - range.first.endInclusive
if (range.second > 1) {
): RealNDElement {
val div = if (endPoint) (num - 1) else num
val delta = range.start - range.endInclusive
return if (num > 1) {
val step = delta / div
if (step == 0.0) {
error("Bad ranges: step = $step")
}
val result = NDElement.real1D(range.second) {
if (endPoint and (it == range.second - 1)) {
range.first.endInclusive
NDElement.real1D(num) {
if (endPoint and (it == num - 1)) {
range.endInclusive
}
range.first.start + it * step
range.start + it * step
}
return result to step
} else {
val step = Double.NaN
return NDElement.real1D(1) { range.first.start } to step
NDElement.real1D(1) { range.start }
}
}
@ -77,29 +79,25 @@ object RealFactory {
* @param endPoint If True, power(base,stop) is the last sample. Otherwise, it is not included.
* @param base - The base of the log space.
*/
fun logSpace(
range: Pair<ClosedFloatingPointRange<Double>, Int>,
fun logspace(
range: ClosedFloatingPointRange<Double>,
num: Int = 100,
endPoint: Boolean = true,
base: Double = 10.0
): RealNDElement {
val lin = linSpace(range, endPoint).first
val tempFun = { x: Double -> power(base, x) }
return tempFun(lin) // FIXME: RealNDElement.map return not suitable type ( `linSpace(range, endPoint).first.map{power(base, it}`)
}
) = linspace(range, num, endPoint).map { power(base, it) }
/**
* Return numbers spaced evenly on a log scale (a geometric progression).
*
* This is similar to [logSpace], but with endpoints specified directly. Each output sample is a constant multiple of the previous.
* This is similar to [logspace], but with endpoints specified directly. Each output sample is a constant multiple of the previous.
* @param range use it like:
* (start..stop) to number
* start is starting value, finaly value depend from endPoint parameter
* @param endPoint If True, right boundary of range is the last sample. Otherwise, it is not included.
*/
fun geomSpace(range: Pair<ClosedFloatingPointRange<Double>, Int>, endPoint: Boolean = true): RealNDElement {
var start = range.first.start
var stop = range.first.endInclusive
val num = range.second
fun geomspace(range: ClosedFloatingPointRange<Double>, num : Int = 100, endPoint: Boolean = true): RealNDElement {
var start = range.start
var stop = range.endInclusive
if (start == 0.0 || stop == 0.0) {
error("Geometric sequence cannot include zero")
}
@ -110,10 +108,9 @@ object RealFactory {
outSign = -outSign
}
val logRange = logSpace((log(start, 10.0)..log(stop, 10.0) to num), endPoint = endPoint)
val function = { x: Double -> outSign * x }
return function(logRange) // FIXME: `outSign*log_` --- don't define times operator
return logspace(log(start, 10.0)..log(stop, 10.0), num, endPoint = endPoint).map {
outSign * it
}
}
/**
@ -126,12 +123,11 @@ object RealFactory {
error("Input must be 2D NDArray")
}
val size = min(array.shape[0], array.shape[0])
if (offset >= 0) {
return NDElement.real1D(size) { i -> array[i, i + offset] }
return if (offset >= 0) {
NDElement.real1D(size) { i -> array[i, i + offset] }
} else {
return NDElement.real1D(size) { i -> array[i - offset, i] }
NDElement.real1D(size) { i -> array[i - offset, i] }
}
}
/**
@ -144,13 +140,13 @@ object RealFactory {
error("Input must be 1D NDArray")
}
val size = array.shape[0]
if (offset >= 0) {
return NDElement.real2D(size, size + offset) { i, j ->
if (i == j + offset) array[i] else 0.0
return if (offset < 0) {
NDElement.real2D(size - offset, size) { i, j ->
if (i - offset == j) array[j] else 0.0
}
} else {
return NDElement.real2D(size - offset, size) { i, j ->
if (i - offset == j) array[j] else 0.0
NDElement.real2D(size, size + offset) { i, j ->
if (i == j + offset) array[i] else 0.0
}
}
}
@ -166,12 +162,12 @@ object RealFactory {
error("Input must be 1D NDArray")
}
val size = if (nCols == 0) array.shape[0] else nCols
if (increasing) {
return NDElement.real2D(array.shape[0], size) { i, j ->
return if (increasing) {
NDElement.real2D(array.shape[0], size) { i, j ->
power(array[i], j)
}
} else {
return NDElement.real2D(array.shape[0], size) { i, j ->
NDElement.real2D(array.shape[0], size) { i, j ->
power(array[i], size - j - 1)
}
}

20
kmath-core/src/commonMain/kotlin/scientifik/kmath/structures/RealNDField.kt
View File

@ -86,11 +86,25 @@ inline fun BufferedNDField<Double, RealField>.produceInline(crossinline initiali
return BufferedNDFieldElement(this, DoubleBuffer(array))
}
/**
* Map one [RealNDElement] using function with indexes
*/
inline fun RealNDElement.mapIndexed(crossinline transform: RealField.(index: IntArray, Double) -> Double) =
context.produceInline { offset -> transform(strides.index(offset), buffer[offset]) }
/**
* Map one [RealNDElement] using function without indexes
*/
inline fun RealNDElement.map(crossinline transform: RealField.(Double) -> Double): RealNDElement {
val array = DoubleArray(strides.linearSize) { offset -> RealField.transform(buffer[offset]) }
return BufferedNDFieldElement(context, DoubleBuffer(array))
}
/**
* Element by element application of any operation on elements to the whole array. Just like in numpy
*/
operator fun Function1<Double, Double>.invoke(ndElement: RealNDElement) =
ndElement.context.produceInline { i -> invoke(ndElement.buffer[i]) }
ndElement.map { this@invoke(it) }
/* plus and minus */
@ -99,10 +113,10 @@ operator fun Function1<Double, Double>.invoke(ndElement: RealNDElement) =
* Summation operation for [BufferedNDElement] and single element
*/
operator fun RealNDElement.plus(arg: Double) =
context.produceInline { i -> buffer[i] + arg }
map { it + arg }
/**
* Subtraction operation between [BufferedNDElement] and single element
*/
operator fun RealNDElement.minus(arg: Double) =
context.produceInline { i -> buffer[i] - arg }
map { it - arg }

3
settings.gradle.kts
View File

@ -2,8 +2,7 @@ pluginManagement {
repositories {
mavenCentral()
maven("https://plugins.gradle.org/m2/")
maven { setUrl("https://dl.bintray.com/kotlin/kotlin-eap") }
maven { setUrl("https://plugins.gradle.org/m2/") }
maven ("https://dl.bintray.com/kotlin/kotlin-eap")
}
}