diff --git a/mllib/src/main/scala/org/apache/spark/mllib/linalg/Matrices.scala b/mllib/src/main/scala/org/apache/spark/mllib/linalg/Matrices.scala index 2cc52e94282ba36c610f0403a84299aef3994e79..327366a1a3a825799c4d38a73c18d018fac2f3a0 100644 --- a/mllib/src/main/scala/org/apache/spark/mllib/linalg/Matrices.scala +++ b/mllib/src/main/scala/org/apache/spark/mllib/linalg/Matrices.scala @@ -17,12 +17,10 @@ package org.apache.spark.mllib.linalg -import java.util.Arrays +import java.util.{Random, Arrays} import breeze.linalg.{Matrix => BM, DenseMatrix => BDM, CSCMatrix => BSM} -import org.apache.spark.util.random.XORShiftRandom - /** * Trait for a local matrix. */ @@ -67,14 +65,14 @@ sealed trait Matrix extends Serializable { } /** Convenience method for `Matrix`^T^-`DenseMatrix` multiplication. */ - def transposeMultiply(y: DenseMatrix): DenseMatrix = { + private[mllib] def transposeMultiply(y: DenseMatrix): DenseMatrix = { val C: DenseMatrix = Matrices.zeros(numCols, y.numCols).asInstanceOf[DenseMatrix] BLAS.gemm(true, false, 1.0, this, y, 0.0, C) C } /** Convenience method for `Matrix`^T^-`DenseVector` multiplication. */ - def transposeMultiply(y: DenseVector): DenseVector = { + private[mllib] def transposeMultiply(y: DenseVector): DenseVector = { val output = new DenseVector(new Array[Double](numCols)) BLAS.gemv(true, 1.0, this, y, 0.0, output) output @@ -291,22 +289,22 @@ object Matrices { * Generate a `DenseMatrix` consisting of i.i.d. uniform random numbers. * @param numRows number of rows of the matrix * @param numCols number of columns of the matrix + * @param rng a random number generator * @return `DenseMatrix` with size `numRows` x `numCols` and values in U(0, 1) */ - def rand(numRows: Int, numCols: Int): Matrix = { - val rand = new XORShiftRandom - new DenseMatrix(numRows, numCols, Array.fill(numRows * numCols)(rand.nextDouble())) + def rand(numRows: Int, numCols: Int, rng: Random): Matrix = { + new DenseMatrix(numRows, numCols, Array.fill(numRows * numCols)(rng.nextDouble())) } /** * Generate a `DenseMatrix` consisting of i.i.d. gaussian random numbers. * @param numRows number of rows of the matrix * @param numCols number of columns of the matrix + * @param rng a random number generator * @return `DenseMatrix` with size `numRows` x `numCols` and values in N(0, 1) */ - def randn(numRows: Int, numCols: Int): Matrix = { - val rand = new XORShiftRandom - new DenseMatrix(numRows, numCols, Array.fill(numRows * numCols)(rand.nextGaussian())) + def randn(numRows: Int, numCols: Int, rng: Random): Matrix = { + new DenseMatrix(numRows, numCols, Array.fill(numRows * numCols)(rng.nextGaussian())) } /** diff --git a/mllib/src/test/scala/org/apache/spark/mllib/linalg/MatricesSuite.scala b/mllib/src/test/scala/org/apache/spark/mllib/linalg/MatricesSuite.scala index 5f8b8c4b72697a41c5a1f5089b672076a1188908..322a0e92429184faae29d107bb9665fb2cca0ad5 100644 --- a/mllib/src/test/scala/org/apache/spark/mllib/linalg/MatricesSuite.scala +++ b/mllib/src/test/scala/org/apache/spark/mllib/linalg/MatricesSuite.scala @@ -17,7 +17,11 @@ package org.apache.spark.mllib.linalg +import java.util.Random + +import org.mockito.Mockito.when import org.scalatest.FunSuite +import org.scalatest.mock.MockitoSugar._ class MatricesSuite extends FunSuite { test("dense matrix construction") { @@ -112,4 +116,50 @@ class MatricesSuite extends FunSuite { assert(sparseMat(0, 1) === 10.0) assert(sparseMat.values(2) === 10.0) } + + test("zeros") { + val mat = Matrices.zeros(2, 3).asInstanceOf[DenseMatrix] + assert(mat.numRows === 2) + assert(mat.numCols === 3) + assert(mat.values.forall(_ == 0.0)) + } + + test("ones") { + val mat = Matrices.ones(2, 3).asInstanceOf[DenseMatrix] + assert(mat.numRows === 2) + assert(mat.numCols === 3) + assert(mat.values.forall(_ == 1.0)) + } + + test("eye") { + val mat = Matrices.eye(2).asInstanceOf[DenseMatrix] + assert(mat.numCols === 2) + assert(mat.numCols === 2) + assert(mat.values.toSeq === Seq(1.0, 0.0, 0.0, 1.0)) + } + + test("rand") { + val rng = mock[Random] + when(rng.nextDouble()).thenReturn(1.0, 2.0, 3.0, 4.0) + val mat = Matrices.rand(2, 2, rng).asInstanceOf[DenseMatrix] + assert(mat.numRows === 2) + assert(mat.numCols === 2) + assert(mat.values.toSeq === Seq(1.0, 2.0, 3.0, 4.0)) + } + + test("randn") { + val rng = mock[Random] + when(rng.nextGaussian()).thenReturn(1.0, 2.0, 3.0, 4.0) + val mat = Matrices.randn(2, 2, rng).asInstanceOf[DenseMatrix] + assert(mat.numRows === 2) + assert(mat.numCols === 2) + assert(mat.values.toSeq === Seq(1.0, 2.0, 3.0, 4.0)) + } + + test("diag") { + val mat = Matrices.diag(Vectors.dense(1.0, 2.0)).asInstanceOf[DenseMatrix] + assert(mat.numRows === 2) + assert(mat.numCols === 2) + assert(mat.values.toSeq === Seq(1.0, 0.0, 0.0, 2.0)) + } } diff --git a/project/MimaExcludes.scala b/project/MimaExcludes.scala index 94de14ddbd2bbc260dcd7e29cb1a5dff2c5bfa40..230239aa40500d563cd67760a7269d212421ef01 100644 --- a/project/MimaExcludes.scala +++ b/project/MimaExcludes.scala @@ -47,6 +47,12 @@ object MimaExcludes { "org.apache.spark.SparkStageInfoImpl.this"), ProblemFilters.exclude[MissingMethodProblem]( "org.apache.spark.SparkStageInfo.submissionTime") + ) ++ Seq( + // SPARK-4614 + ProblemFilters.exclude[MissingMethodProblem]( + "org.apache.spark.mllib.linalg.Matrices.randn"), + ProblemFilters.exclude[MissingMethodProblem]( + "org.apache.spark.mllib.linalg.Matrices.rand") ) case v if v.startsWith("1.2") =>