library(gpuR)
context("vclMatrix algebra")
if(detectGPUs() >= 1){
current_context <- set_device_context("gpu")
}else{
current_context <- currentContext()
}
# set seed
set.seed(123)
ORDER <- 4
ORDER_PAD <- 129
# Base R objects
Aint <- matrix(sample(seq(10), ORDER^2, replace=TRUE), nrow=ORDER, ncol=ORDER)
Bint <- matrix(sample(seq(10), ORDER^2, replace=TRUE), nrow=ORDER, ncol=ORDER)
AintPad <- matrix(sample(seq(10), ORDER*ORDER_PAD, replace=TRUE), nrow=ORDER, ncol=ORDER_PAD)
BintPad <- matrix(sample(seq(10), ORDER*ORDER_PAD, replace=TRUE), nrow=ORDER_PAD, ncol = ORDER)
A <- matrix(rnorm(ORDER^2), nrow=ORDER, ncol=ORDER)
B <- matrix(rnorm(ORDER^2), nrow=ORDER, ncol=ORDER)
E <- matrix(rnorm(15), nrow=5)
v <- rnorm(ORDER^2)
v2 <- rnorm(ORDER)
# Single Precision tests
test_that("vclMatrix Single Precision Matrix multiplication", {
has_gpu_skip()
C <- A %*% B
fgpuA <- vclMatrix(A, type="float")
fgpuB <- vclMatrix(B, type="float")
fgpuE <- vclMatrix(E, type = "float")
fgpuC <- fgpuA %*% fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
expect_error(fgpuA %*% fgpuE,
info = "error not thrown for non-conformant matrices")
fgpuC <- A %*% fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
fgpuC <- fgpuA %*% B
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
})
test_that("vclMatrix Single Precision Matrix-Vector multiplication", {
has_gpu_skip()
C <- A %*% v2
C2 <- v2 %*% B
fgpuA <- vclMatrix(A, type="float")
fgpuB <- vclMatrix(B, type="float")
fgpuV <- vclVector(v2, type = "float")
fgpuC <- fgpuA %*% fgpuV
fgpuC2 <- fgpuV %*% fgpuB
expect_equal(fgpuC[,], c(C), tolerance=1e-07,
info="float matrix elements not equivalent")
expect_equal(fgpuC2[,], c(C2), tolerance=1e-07,
info="float matrix elements not equivalent")
})
test_that("vclMatrix Single Precision Matrix Subtraction", {
has_gpu_skip()
C <- A - B
fgpuA <- vclMatrix(A, type="float")
fgpuB <- vclMatrix(B, type="float")
fgpuE <- vclMatrix(E, type="float")
fgpuC <- fgpuA - fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
expect_error(fgpuA - fgpuE)
fgpuC <- fgpuA - B
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
fgpuC <- A - fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
})
#
# test_that("vclMatrix Single Precision Matrix/Vector Subtraction", {
#
# has_gpu_skip()
#
# C <- A - c(B)
# C2 <- c(A) - B
#
# fgpuA <- vclMatrix(A, type="float")
# fgpuB <- vclVector(c(B), type="float")
#
# fgpuC <- fgpuA - fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuA <- vclVector(c(A), type="float")
# fgpuB <- vclMatrix(B, type="float")
#
# fgpuC <- fgpuA - fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C2, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# })
test_that("vclMatrix Single Precision Scalar Matrix Subtraction", {
has_gpu_skip()
C <- A - 1
C2 <- 1 - A
fgpuA <- vclMatrix(A, type="float")
fgpuC <- fgpuA - 1
fgpuC2 <- 1 - fgpuA
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
expect_is(fgpuC2, "fvclMatrix")
expect_equal(fgpuC2[,], C2, tolerance=1e-07,
info="float matrix elements not equivalent")
})
test_that("vclMatrix Single Precision Unary Scalar Matrix Subtraction", {
has_gpu_skip()
C <- -A
fgpuA <- vclMatrix(A, type="float")
fgpuC <- -fgpuA
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
})
test_that("vclMatrix Single Precision Matrix Addition", {
has_gpu_skip()
C <- A + B
fgpuA <- vclMatrix(A, type="float")
fgpuB <- vclMatrix(B, type="float")
fgpuE <- vclMatrix(E, type="float")
fgpuC <- fgpuA + fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
expect_error(fgpuA + fgpuE)
fgpuC <- A + fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
fgpuC <- fgpuA + B
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
})
test_that("vclMatrix Single Precision Matrix/Vector Addition", {
has_gpu_skip()
C <- A + c(B)
fgpuA <- vclMatrix(A, type="float")
fgpuB <- vclVector(c(B), type="float")
fgpuC <- fgpuA + fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
fgpuA <- vclVector(c(A), type="float")
fgpuB <- vclMatrix(B, type="float")
fgpuC <- fgpuA + fgpuB
expect_is(fgpuC, "fvclMatrix")
expect_equal(fgpuC[,], C, tolerance=1e-07,
info="float matrix elements not equivalent")
})
# test_that("vclMatrix Single Precision Scalar Matrix Addition", {
#
# has_gpu_skip()
#
# C <- A + 1
# C2 <- 1 + A
#
# fgpuA <- vclMatrix(A, type="float")
#
# fgpuC <- fgpuA + 1
# fgpuC2 <- 1 + fgpuA
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_is(fgpuC2, "fvclMatrix")
# expect_equal(fgpuC2[,], C2, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Matrix Element-Wise Multiplication", {
#
# has_gpu_skip()
#
# C <- A * B
#
# fgpuA <- vclMatrix(A, type="float")
# fgpuB <- vclMatrix(B, type="float")
# fgpuE <- vclMatrix(E, type="float")
#
# fgpuC <- fgpuA * fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_error(fgpuA * fgpuE)
#
# fgpuC <- A * fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuC <- fgpuA * B
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Scalar Matrix Multiplication", {
#
# has_gpu_skip()
#
# C <- A * 2
# C2 <- 2 * A
#
# dgpuA <- vclMatrix(A, type="float")
#
# dgpuC <- dgpuA * 2
# dgpuC2 <- 2 * dgpuA
#
# expect_is(dgpuC, "fvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_is(dgpuC2, "fvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Matrix Element-Wise Division", {
#
# has_gpu_skip()
#
# C <- A / B
#
# fgpuA <- vclMatrix(A, type="float")
# fgpuB <- vclMatrix(B, type="float")
# fgpuE <- vclMatrix(E, type="float")
#
# fgpuC <- fgpuA / fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_error(fgpuA / fgpuE)
#
# fgpuC <- A / fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuC <- fgpuA / B
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Scalar Matrix Division", {
#
# has_gpu_skip()
#
# C <- A/2
# C2 <- 2/A
#
# dgpuA <- vclMatrix(A, type="float")
#
# dgpuC <- dgpuA/2
# dgpuC2 <- 2/dgpuA
#
# expect_is(dgpuC, "fvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_is(dgpuC2, "fvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Matrix Element-Wise Power", {
#
# has_gpu_skip()
# pocl_check()
#
# C <- A ^ B
#
# fgpuA <- vclMatrix(A, type="float")
# fgpuB <- vclMatrix(B, type="float")
# fgpuE <- vclMatrix(E, type="float")
#
# fgpuC <- fgpuA ^ fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_error(fgpuA ^ fgpuE)
#
# fgpuC <- A ^ fgpuB
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuC <- fgpuA ^ B
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision Scalar Matrix Power", {
#
# has_gpu_skip()
#
# C <- A^2
# C2 <- 2^A
#
# dgpuA <- vclMatrix(A, type="float")
#
# dgpuC <- dgpuA^2
# dgpuC2 <- 2^dgpuA
#
# expect_is(dgpuC, "fvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_equal(dgpuC2[,], C2, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision crossprod", {
#
# has_gpu_skip()
#
# X <- matrix(rnorm(10), nrow=2)
# Y <- matrix(rnorm(10), nrow=2)
# Z <- matrix(rnorm(10), nrow=5)
#
# C <- crossprod(X,Y)
# Cs <- crossprod(X)
#
# fgpuX <- vclMatrix(X, type="float")
# fgpuY <- vclMatrix(Y, type="float")
# fgpuZ <- vclMatrix(Z, type="float")
#
# fgpuC <- crossprod(fgpuX, fgpuY)
# fgpuCs <- crossprod(fgpuX)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_equal(fgpuCs[,], Cs, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_error(crossprod(fgpuX, fgpuZ))
#
# fgpuC <- crossprod(fgpuX, Y)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuC <- crossprod(X, fgpuY)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision tcrossprod", {
#
# has_gpu_skip()
#
# X <- matrix(rnorm(10), nrow=2)
# Y <- matrix(rnorm(10), nrow=2)
# Z <- matrix(rnorm(12), nrow=2)
#
# C <- tcrossprod(X,Y)
# Cs <- tcrossprod(X)
#
# fgpuX <- vclMatrix(X, type="float")
# fgpuY <- vclMatrix(Y, type="float")
# fgpuZ <- vclMatrix(Z, type="float")
#
# fgpuC <- tcrossprod(fgpuX, fgpuY)
# fgpuCs <- tcrossprod(fgpuX)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_equal(fgpuCs[,], Cs, tolerance=1e-07,
# info="float matrix elements not equivalent")
# expect_error(tcrossprod(fgpuX, fgpuZ))
#
# fgpuC <- tcrossprod(fgpuX, Y)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
#
# fgpuC <- tcrossprod(X, fgpuY)
#
# expect_is(fgpuC, "fvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=1e-07,
# info="float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision transpose", {
#
# has_gpu_skip()
#
# At <- t(A)
#
# fgpuA <- vclMatrix(A, type="float")
# fgpuAt <- t(fgpuA)
#
# expect_is(fgpuAt, "fvclMatrix")
# expect_equal(fgpuAt[,], At, tolerance=1e-07,
# info="transposed float matrix elements not equivalent")
# })
#
# test_that("vclMatrix Single Precision determinant", {
#
# has_gpu_skip()
#
# d <- det(A)
#
# fgpuA <- vclMatrix(A, type="float")
# fgpud <- det(fgpuA)
#
# expect_is(fgpud, "numeric")
# expect_equal(fgpud, d, tolerance=1e-07,
# info="float determinants not equivalent")
# })
# Integer tests
test_that("vclMatrix Integer Matrix multiplication", {
has_gpu_skip()
Cint <- Aint %*% Bint
CintPad <- AintPad %*% BintPad
igpuA <- vclMatrix(Aint, type="integer")
igpuB <- vclMatrix(Bint, type="integer")
igpuApad <- vclMatrix(AintPad, type="integer")
igpuBpad <- vclMatrix(BintPad, type="integer")
igpuC <- igpuA %*% igpuB
expect_equivalent(igpuC[,], Cint,
info="integer matrix elements not equivalent")
igpuC <- Aint %*% igpuB
expect_equivalent(igpuC[,], Cint,
info="integer matrix elements not equivalent")
igpuC <- igpuA %*% Bint
expect_equivalent(igpuC[,], Cint,
info="integer matrix elements not equivalent")
igpuCpad <- igpuApad %*% igpuBpad
expect_equivalent(igpuCpad[], CintPad,
info = "padded rectangular matrix elements not equivalent")
})
# test_that("vclMatrix Integer Matrix Subtraction", {
#
# has_gpu_skip()
#
# Cint <- Aint - Bint
#
# igpuA <- vclMatrix(Aint, type="integer")
# igpuB <- vclMatrix(Bint, type="integer")
#
# igpuC <- igpuA - igpuB
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
#
#
# igpuC <- igpuA - Bint
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
#
#
# igpuC <- Aint - igpuB
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Scalar Matrix Subtraction", {
#
# has_gpu_skip()
#
# C <- Aint - 1L
# C2 <- 1L - Aint
#
# fgpuA <- vclMatrix(Aint, type="integer")
#
# fgpuC <- fgpuA - 1L
# fgpuC2 <- 1L - fgpuA
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# expect_is(fgpuC2, "ivclMatrix")
# expect_equal(fgpuC2[,], C2,
# info="intger matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Unary Scalar Matrix Subtraction", {
#
# has_gpu_skip()
#
# C <- -Aint
#
# fgpuA <- vclMatrix(Aint, type="integer")
#
# fgpuC <- -fgpuA
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Matrix Addition", {
#
# has_gpu_skip()
#
# Cint <- Aint + Bint
#
# igpuA <- vclMatrix(Aint, type="integer")
# igpuB <- vclMatrix(Bint, type="integer")
#
# igpuC <- igpuA + igpuB
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
#
# igpuC <- Aint + igpuB
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
#
# igpuC <- igpuA + Bint
#
# expect_is(igpuC, "ivclMatrix")
# expect_equal(igpuC[,], Cint,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Scalar Matrix Addition", {
#
# has_gpu_skip()
#
# C <- Aint + 1L
# C2 <- 1L + Aint
#
# fgpuA <- vclMatrix(Aint, type="integer")
#
# fgpuC <- fgpuA + 1L
# fgpuC2 <- 1L + fgpuA
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# expect_is(fgpuC2, "ivclMatrix")
# expect_equal(fgpuC2[,], C2,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Matrix Element-Wise Multiplication", {
#
# has_gpu_skip()
#
# C <- Aint * Bint
#
# fgpuA <- vclMatrix(Aint, type="integer")
# fgpuB <- vclMatrix(Bint, type="integer")
#
# fgpuC <- fgpuA * fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- Aint * fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- fgpuA * Bint
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Scalar Matrix Multiplication", {
#
# has_gpu_skip()
#
# C <- Aint * 2L
# C2 <- 2L * Aint
#
# dgpuA <- vclMatrix(Aint, type="integer")
#
# dgpuC <- dgpuA * 2L
# dgpuC2 <- 2L * dgpuA
#
# expect_is(dgpuC, "ivclMatrix")
# expect_equal(dgpuC[,], C,
# info="integer matrix elements not equivalent")
# expect_is(dgpuC2, "ivclMatrix")
# expect_equal(dgpuC2[,], C2,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Matrix Element-Wise Division", {
#
# has_gpu_skip()
#
# C <- Aint / Bint
# C <- apply(C, 2, as.integer)
#
# fgpuA <- vclMatrix(Aint, type="integer")
# fgpuB <- vclMatrix(Bint, type="integer")
#
# fgpuC <- fgpuA / fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- Aint / fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- fgpuA / Bint
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Scalar Matrix Division", {
#
# has_gpu_skip()
#
# C <- Aint/2L
# C2 <- 2L/Aint
#
# C <- apply(C, 2, as.integer)
# C2 <- apply(C2, 2, as.integer)
#
# dgpuA <- vclMatrix(Aint, type="integer")
#
# dgpuC <- dgpuA/2L
# dgpuC2 <- 2L/dgpuA
#
# expect_is(dgpuC, "ivclMatrix")
# expect_equal(dgpuC[,], C,
# info="integer matrix elements not equivalent")
# expect_is(dgpuC2, "ivclMatrix")
# expect_equal(dgpuC2[,], C2,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Matrix Element-Wise Power", {
#
# has_gpu_skip()
# pocl_check()
#
# Apow <- matrix(seq.int(9), ncol=3, nrow=3)
# Bpow <- matrix(2, ncol = 3, nrow = 3)
# C <- Apow ^ Bpow
#
# fgpuA <- vclMatrix(Apow, type="integer")
# fgpuB <- vclMatrix(Bpow, type="integer")
#
# fgpuC <- fgpuA ^ fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- Apow ^ fgpuB
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
#
# fgpuC <- fgpuA ^ Bpow
#
# expect_is(fgpuC, "ivclMatrix")
# expect_equal(fgpuC[,], C,
# info="integer matrix elements not equivalent")
# })
#
# test_that("vclMatrix Integer Precision Scalar Matrix Power", {
#
# has_gpu_skip()
#
# C <- Aint^2L
# C2 <- 2L^Aint
#
# C <- apply(C, 2, as.integer)
# C2 <- apply(C2, 2, as.integer)
#
# dgpuA <- vclMatrix(Aint, type="integer")
#
# dgpuC <- dgpuA^2L
# dgpuC2 <- 2L^dgpuA
#
# expect_is(dgpuC, "ivclMatrix")
# expect_equal(dgpuC[,], C,
# info="integer matrix elements not equivalent")
# expect_equal(dgpuC2[,], C2,
# info="integer matrix elements not equivalent")
# })
#
# # Double Precision tests
#
# test_that("vclMatrix Double Precision Matrix multiplication", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A %*% B
#
# dgpuA <- vclMatrix(A, type="double")
# dgpuB <- vclMatrix(B, type="double")
#
# dgpuC <- dgpuA %*% dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- dgpuA %*% B
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- A %*% dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix-Vector multiplication", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A %*% v2
# C2 <- v2 %*% B
#
# fgpuA <- vclMatrix(A, type="double")
# fgpuB <- vclMatrix(B, type="double")
# fgpuV <- vclVector(v2, type = "double")
#
# fgpuC <- fgpuA %*% fgpuV
# fgpuC2 <- fgpuV %*% fgpuB
#
# expect_equal(fgpuC[,], c(C), tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
# expect_equal(fgpuC2[,], c(C2), tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix Subtraction", {
#
# has_gpu_skip()
# has_double_skip()
#
#
# C <- A - B
#
# dgpuA <- vclMatrix(A, type="double")
# dgpuB <- vclMatrix(B, type="double")
# dgpuE <- vclMatrix(E, type="double")
#
# dgpuC <- dgpuA - dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(dgpuA - dgpuE)
#
# dgpuC <- A - dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- dgpuA - B
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix/Vector Subtraction", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A - c(B)
# C2 <- c(A) - B
#
# fgpuA <- vclMatrix(A, type="double")
# fgpuB <- vclVector(c(B), type="double")
#
# fgpuC <- fgpuA - fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
#
# fgpuA <- vclVector(c(A), type="double")
# fgpuB <- vclMatrix(B, type="double")
#
# fgpuC <- fgpuA - fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C2, tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
#
# })
#
# test_that("vclMatrix Double Precision Matrix Addition", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A + B
#
# dgpuA <- vclMatrix(A, type="double")
# dgpuB <- vclMatrix(B, type="double")
# dgpuE <- vclMatrix(E, type="double")
#
# dgpuC <- dgpuA + dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(dgpuA + dgpuE)
#
# dgpuC <- A + dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- dgpuA + B
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix/Vector Addition", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A + c(B)
#
# fgpuA <- vclMatrix(A, type="double")
# fgpuB <- vclVector(c(B), type="double")
#
# fgpuC <- fgpuA + fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
#
# fgpuA <- vclVector(c(A), type="double")
# fgpuB <- vclMatrix(B, type="double")
#
# fgpuC <- fgpuA + fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps^0.5,
# info="double matrix elements not equivalent")
#
# })
#
# test_that("vclMatrix Double Precision Scalar Matrix Addition", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A + 1
# C2 <- 1 + A
#
# dgpuA <- vclMatrix(A, type="double")
#
# dgpuC <- dgpuA + 1
# dgpuC2 <- 1 + dgpuA
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_is(dgpuC2, "dvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Scalar Matrix Subtraction", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A - 1
# C2 <- 1 - A
#
# dgpuA <- vclMatrix(A, type="double")
#
# dgpuC <- dgpuA - 1
# dgpuC2 <- 1 - dgpuA
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_is(dgpuC2, "dvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Unary Matrix Subtraction", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- -A
#
# fgpuA <- vclMatrix(A, type="double")
#
# fgpuC <- -fgpuA
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix Element-Wise Multiplication", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A * B
#
# dgpuA <- vclMatrix(A, type="double")
# dgpuB <- vclMatrix(B, type="double")
# dgpuE <- vclMatrix(E, type="double")
#
# dgpuC <- dgpuA * dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(dgpuA * dgpuE)
#
# dgpuC <- A * dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- dgpuA * B
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Scalar Matrix Multiplication", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A * 2
# C2 <- 2 * A
#
# dgpuA <- vclMatrix(A, type="double")
#
# dgpuC <- dgpuA * 2
# dgpuC2 <- 2 * dgpuA
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_is(dgpuC2, "dvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix Element-Wise Division", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A / B
#
# dgpuA <- vclMatrix(A, type="double")
# dgpuB <- vclMatrix(B, type="double")
# dgpuE <- vclMatrix(E, type="double")
#
# dgpuC <- dgpuA / dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(dgpuA * dgpuE)
#
# dgpuC <- A / dgpuB
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# dgpuC <- dgpuA / B
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Scalar Matrix Division", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A/2
# C2 <- 2/A
#
# dgpuA <- vclMatrix(A, type="double")
#
# dgpuC <- dgpuA/2
# dgpuC2 <- 2/dgpuA
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_is(dgpuC2, "dvclMatrix")
# expect_equal(dgpuC2[,], C2, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Matrix Element-Wise Power", {
#
# has_gpu_skip()
# has_double_skip()
# pocl_check()
#
# C <- A ^ B
#
# fgpuA <- vclMatrix(A, type="double")
# fgpuB <- vclMatrix(B, type="double")
# fgpuE <- vclMatrix(E, type="double")
#
# fgpuC <- fgpuA ^ fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(fgpuA ^ fgpuE)
#
# fgpuC <- A ^ fgpuB
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# fgpuC <- fgpuA ^ B
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision Scalar Matrix Power", {
#
# has_gpu_skip()
# has_double_skip()
#
# C <- A^2
#
# dgpuA <- vclMatrix(A, type="double")
#
# dgpuC <- dgpuA^2
#
# expect_is(dgpuC, "dvclMatrix")
# expect_equal(dgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision crossprod", {
#
# has_gpu_skip()
# has_double_skip()
#
# X <- matrix(rnorm(10), nrow=2)
# Y <- matrix(rnorm(10), nrow=2)
# Z <- matrix(rnorm(10), nrow=5)
#
# C <- crossprod(X,Y)
# Cs <- crossprod(X)
#
# fgpuX <- vclMatrix(X, type="double")
# fgpuY <- vclMatrix(Y, type="double")
# fgpuZ <- vclMatrix(Z, type="double")
#
# fgpuC <- crossprod(fgpuX, fgpuY)
# fgpuCs <- crossprod(fgpuX)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_equal(fgpuCs[,], Cs, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(crossprod(fgpuX, fgpuZ))
#
# fgpuC <- crossprod(fgpuX, Y)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# fgpuC <- crossprod(X, fgpuY)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision tcrossprod", {
#
# has_gpu_skip()
# has_double_skip()
#
# X <- matrix(rnorm(10), nrow=2)
# Y <- matrix(rnorm(10), nrow=2)
# Z <- matrix(rnorm(12), nrow=2)
#
# C <- tcrossprod(X,Y)
# Cs <- tcrossprod(X)
#
# fgpuX <- vclMatrix(X, type="double")
# fgpuY <- vclMatrix(Y, type="double")
# fgpuZ <- vclMatrix(Z, type="double")
#
# fgpuC <- tcrossprod(fgpuX, fgpuY)
# fgpuCs <- tcrossprod(fgpuX)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_equal(fgpuCs[,], Cs, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# expect_error(tcrossprod(fgpuX, fgpuZ))
#
# fgpuC <- tcrossprod(fgpuX, Y)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
#
# fgpuC <- tcrossprod(X, fgpuY)
#
# expect_is(fgpuC, "dvclMatrix")
# expect_equal(fgpuC[,], C, tolerance=.Machine$double.eps ^ 0.5,
# info="double matrix elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision transpose", {
#
# has_gpu_skip()
# has_double_skip()
#
# At <- t(A)
#
# fgpuA <- vclMatrix(A, type="double")
# fgpuAt <- t(fgpuA)
#
# expect_is(fgpuAt, "dvclMatrix")
# expect_equal(fgpuAt[,], At, tolerance=.Machine$double.eps^0.5,
# info="transposed double matrix elements not equivalent")
# })
#
#
# # test_that("vclMatrix Integer Matrix multiplication", {
# #
# # has_gpu_skip()
# #
# # Cint <- Aint %*% Bint
# #
# # igpuA <- vclMatrix(Aint, type="integer")
# # igpuB <- vclMatrix(Bint, type="integer")
# #
# # igpuC <- igpuA %*% igpuB
# #
# # expect_equivalent(igpuC[,], Cint,
# # info="float matrix elements not equivalent")
# # })
# #
# # test_that("vclMatrix Integer Matrix Subtraction", {
# #
# # has_gpu_skip()
# #
# # Cint <- Aint - Bint
# #
# # igpuA <- vclMatrix(Aint, type="integer")
# # igpuB <- vclMatrix(Bint, type="integer")
# #
# # igpuC <- igpuA - igpuB
# #
# # expect_is(igpuC, "ivclMatrix")
# # expect_equal(igpuC[,], Cint,
# # info="integer matrix elements not equivalent")
# # })
# #
# # test_that("vclMatrix Integer Matrix Addition", {
# #
# # has_gpu_skip()
# #
# # Cint <- Aint + Bint
# #
# # igpuA <- vclMatrix(Aint, type="integer")
# # igpuB <- vclMatrix(Bint, type="integer")
# #
# # igpuC <- igpuA + igpuB
# #
# # expect_is(igpuC, "ivclMatrix")
# # expect_equal(igpuC[,], Cint,
# # info="integer matrix elements not equivalent")
# # })
#
#
#
# test_that("CPU vclMatrix Diagonal access", {
#
# has_gpu_skip()
#
# fgpuA <- vclMatrix(A, type="float")
#
# D <- diag(A)
# gpuD <- diag(fgpuA)
#
# expect_is(gpuD, "fvclVector")
# expect_equal(gpuD[,], D, tolerance=1e-07,
# info="float matrix diagonal elements not equivalent")
#
# vec <- rnorm(ORDER)
# diag(fgpuA) <- vclVector(vec, type = "float")
# diag(A) <- vec
#
# expect_equal(fgpuA[,], A, tolerance=1e-07,
# info="set float matrix diagonal elements not equivalent")
#
# has_double_skip()
#
# fgpuA <- vclMatrix(A, type="double")
#
# D <- diag(A)
# gpuD <- diag(fgpuA)
#
# expect_is(gpuD, "dvclVector")
# expect_equal(gpuD[,], D, tolerance=.Machine$double.eps^0.5,
# info="double matrix diagonal elements not equivalent")
#
# vec <- rnorm(ORDER)
# diag(fgpuA) <- vclVector(vec, type = "double")
# diag(A) <- vec
#
# expect_equal(fgpuA[,], A, tolerance=.Machine$double.eps^0.5,
# info="set double matrix diagonal elements not equivalent")
# })
#
# test_that("vclMatrix Double Precision determinant", {
#
# has_gpu_skip()
# has_double_skip()
#
# d <- det(A)
#
# fgpuA <- vclMatrix(A, type="double")
# fgpud <- det(fgpuA)
#
# expect_is(fgpud, "numeric")
# expect_equal(fgpud, d, tolerance=.Machine$double.eps^0.5,
# info="double determinants not equivalent")
# })
setContext(current_context)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.