Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance.
Project price only 1 $
You can buy this project and download/modify it how often you want.
breeze.linalg.functions.split.scala Maven / Gradle / Ivy
package breeze.linalg
import breeze.generic.UFunc
import scala.reflect.ClassTag
import spire.implicits._
import breeze.storage.Zero
/**
* split the array
*
* @author stucchio
*/
object split extends UFunc {
implicit def implIntVec[T: ClassTag]: Impl2[DenseVector[T], Int, IndexedSeq[DenseVector[T]]] = {
new Impl2[DenseVector[T], Int, IndexedSeq[DenseVector[T]]] {
def apply(v: DenseVector[T], n: Int): IndexedSeq[DenseVector[T]] = {
require(n >= 0)
require(n <= v.size)
require(v.size % n == 0)
val individualVectorSize = v.size / n
val result = new collection.mutable.ArrayBuffer[DenseVector[T]]()
cforRange(0 until n) { k =>
val offsetInOriginalVector = k*individualVectorSize
val chunk = new Array[T](individualVectorSize)
cforRange(0 until individualVectorSize){i =>
chunk(i) = v(offsetInOriginalVector+i)
}
result += DenseVector[T](chunk)
}
result
}
}
}
implicit def implSeqVec[T: ClassTag]: Impl2[DenseVector[T], Seq[Int], IndexedSeq[DenseVector[T]]] = new Impl2[DenseVector[T], Seq[Int], IndexedSeq[DenseVector[T]]] {
def apply(v: DenseVector[T], nSeq: Seq[Int]): IndexedSeq[DenseVector[T]] = {
require(nSeq.size < v.size)
val result = new collection.mutable.ArrayBuffer[DenseVector[T]]()
var lastN: Int = 0
nSeq.foreach{ n =>
val chunk = new Array[T](n - lastN)
cforRange(lastN until n) { i =>
chunk(i-lastN) = v(i)
}
result += DenseVector[T](chunk)
lastN = n
}
if (lastN < v.size) { //If we did not already add last chunk to result, do it now.
val chunk = new Array[T](v.size - lastN)
cforRange(lastN until v.size) { i =>
chunk(i-lastN) = v(i)
}
result += DenseVector[T](chunk)
}
result
}
}
implicit def implIntMatrix[T: ClassTag](implicit zero: Zero[T]): Impl3[DenseMatrix[T], Int, Int, IndexedSeq[DenseMatrix[T]]] = new Impl3[DenseMatrix[T], Int, Int, IndexedSeq[DenseMatrix[T]]] {
def apply(v: DenseMatrix[T], n: Int, axis: Int): IndexedSeq[DenseMatrix[T]] = axis match {
case 0 => vsplit(v,n)
case 1 => hsplit(v,n)
case _ => throw new IllegalArgumentException("Matrices have only two axes.")
}
}
}
object hsplit extends UFunc {
implicit def implIntVec[T: ClassTag]: Impl2[DenseVector[T], Int, IndexedSeq[DenseVector[T]]] = new Impl2[DenseVector[T], Int, IndexedSeq[DenseVector[T]]] { //For vectors just an alias
def apply(v: DenseVector[T], n: Int): IndexedSeq[DenseVector[T]] = hsplit(v,n)
}
implicit def implSeqVec[T: ClassTag]: Impl2[DenseVector[T], Seq[Int], IndexedSeq[DenseVector[T]]] = new Impl2[DenseVector[T], Seq[Int], IndexedSeq[DenseVector[T]]] { //For vectors just an alias
def apply(v: DenseVector[T], n: Seq[Int]): IndexedSeq[DenseVector[T]] = hsplit(v,n)
}
implicit def implIntMat[T: ClassTag](implicit zero: Zero[T]): Impl2[DenseMatrix[T], Int, IndexedSeq[DenseMatrix[T]]] = new Impl2[DenseMatrix[T],Int, IndexedSeq[DenseMatrix[T]]] { //for matrices
def apply(v: DenseMatrix[T], n: Int): IndexedSeq[DenseMatrix[T]] = {
require(n >= 0)
require(n <= v.cols)
require(v.cols % n == 0)
val result = new collection.mutable.ArrayBuffer[DenseMatrix[T]]()
val newCols = v.cols / n
val newSize = v.rows * newCols
cforRange(0 until n) { k =>
val offsetInOriginalMatrix = k*newCols
val chunk = DenseMatrix.create(v.rows, newCols, new Array[T](newSize))
cforRange2(0 until v.rows, 0 until newCols) { (i, j) =>
chunk(i,j) = v(i,j+offsetInOriginalMatrix)
}
result += chunk
}
result
}
}
}
object vsplit extends UFunc {
implicit def implIntMat[T: ClassTag](implicit zero: Zero[T]): Impl2[DenseMatrix[T], Int, IndexedSeq[DenseMatrix[T]]] = new Impl2[DenseMatrix[T],Int, IndexedSeq[DenseMatrix[T]]] { //for matrices
def apply(v: DenseMatrix[T], n: Int): IndexedSeq[DenseMatrix[T]] = {
require(n >= 0)
require(n <= v.cols)
require(v.cols % n == 0)
val result = new collection.mutable.ArrayBuffer[DenseMatrix[T]]()
val newRows = v.rows / n
cforRange(0 until n) { k =>
val offsetInOriginalMatrix = k*newRows
val chunk = DenseMatrix.create(newRows, v.cols, new Array[T](v.cols * newRows))
cforRange2(0 until newRows, 0 until v.cols) { (i, j) =>
chunk(i,j) = v(i+offsetInOriginalMatrix,j)
}
result += chunk
}
result
}
}
}
