irlba
irlba copied to clipboard
bwlewis
Changes in Matrix 1.6-2 breaking irlba: `function 'as_cholmod_sparse' not provided by package 'Matrix'`
Hi @bwlewis,
Changes in Matrix 1.6-2 seem to have broken irlba::irlba() for at least some types of sparse matrices.
I'm afraid I don't understand enough of Matrix and irlba to provide much help at this stage.
Cheers, Pete
With Matrix 1.6-2
# Example taken from help("sparseMatrix", package = "Matrix").
i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7)
A <- Matrix::sparseMatrix(i, j, x = x)
irlba::irlba(A)
#> Warning in irlba::irlba(A): You're computing too large a percentage of total
#> singular values, use a standard svd instead.
#> Error in irlba::irlba(A): function 'as_cholmod_sparse' not provided by package 'Matrix'
Session info
sessionInfo()
#> R version 4.3.1 (2023-06-16)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: CentOS Linux 7 (Core)
#>
#> Matrix products: default
#> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so
#> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
#> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
#> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
#> [9] LC_ADDRESS=C LC_TELEPHONE=C
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#>
#> time zone: Australia/Melbourne
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> loaded via a namespace (and not attached):
#> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-2 xfun_0.41
#> [5] lattice_0.21-8 glue_1.6.2 knitr_1.45 htmltools_0.5.7
#> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1
#> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0
#> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7
#> [21] rlang_1.1.2 fs_1.6.3
With Matrix 1.6-1
# Example taken from help("sparseMatrix", package = "Matrix").
i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7)
A <- Matrix::sparseMatrix(i, j, x = x)
irlba::irlba(A)
#> Warning in irlba::irlba(A): You're computing too large a percentage of total
#> singular values, use a standard svd instead.
#> $d
#> [1] 49.00000 44.27189 35.00000 28.00000 21.00000
#>
#> $u
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] 1.179612e-16 -6.418477e-17 -1.942890e-16 6.938894e-17 1.110223e-16
#> [2,] 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
#> [3,] 3.105155e-16 3.162278e-01 -6.245005e-17 1.110223e-16 -3.816392e-17
#> [4,] -1.734723e-16 -2.428613e-17 -3.885781e-16 5.273559e-16 1.000000e+00
#> [5,] -1.942890e-16 -2.602085e-16 9.436896e-16 1.000000e+00 -3.608225e-16
#> [6,] -2.498002e-16 -2.012279e-16 -1.000000e+00 1.054712e-15 2.775558e-17
#> [7,] 9.419548e-16 9.486833e-01 -1.318390e-16 2.914335e-16 -1.734723e-17
#> [8,] 1.000000e+00 -1.334002e-15 -3.885781e-16 -3.053113e-16 8.326673e-17
#>
#> $v
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] -6.938894e-17 -1.474515e-16 -1.665335e-16 -1.387779e-16 0.000000e+00
#> [2,] -9.714451e-17 -9.020562e-17 2.220446e-16 5.551115e-17 1.665335e-16
#> [3,] 6.071532e-18 1.192622e-17 2.081668e-17 1.040834e-17 1.734723e-18
#> [4,] 1.127570e-17 1.474515e-17 3.122502e-17 2.775558e-17 -1.040834e-17
#> [5,] -1.994932e-17 -4.770490e-17 -4.857226e-17 -1.734723e-17 -1.040834e-17
#> [6,] -7.632783e-17 5.898060e-17 -3.330669e-16 6.938894e-16 1.000000e+00
#> [7,] -8.326673e-17 -5.898060e-17 1.249001e-15 1.000000e+00 -3.053113e-16
#> [8,] -3.816392e-16 -1.734723e-16 -1.000000e+00 9.714451e-16 2.775558e-17
#> [9,] 1.065120e-15 1.000000e+00 -1.474515e-16 2.029626e-16 -1.908196e-16
#> [10,] 1.000000e+00 -1.234689e-15 -5.134781e-16 2.012279e-16 2.463307e-16
#>
#> $iter
#> [1] 1
#>
#> $mprod
#> [1] 16
Session info
sessionInfo()
#> R version 4.3.1 (2023-06-16)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: CentOS Linux 7 (Core)
#>
#> Matrix products: default
#> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so
#> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
#> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
#> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
#> [9] LC_ADDRESS=C LC_TELEPHONE=C
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#>
#> time zone: Australia/Melbourne
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> loaded via a namespace (and not attached):
#> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-1 xfun_0.41
#> [5] lattice_0.21-8 glue_1.6.2 knitr_1.45 htmltools_0.5.7
#> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1
#> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0
#> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7
#> [21] rlang_1.1.2 fs_1.6.3
Thanks for reporting this!
This will force me to revisit a project we started a while back at re-implementing the sparse parts of the code in a more generic way.
I'll try and get that ready as soon as I can.
On 11/8/23, Peter Hickey @.***> wrote:
Hi @bwlewis,
Changes in Matrix 1.6-2 seem to have broken
irlba::irlba()for at least some types of sparse matrices. I'm afraid I don't understand enough of Matrix and irlba to provide much help at this stage.Cheers, Pete
With Matrix 1.6-2
# Example taken from help("sparseMatrix", package = "Matrix"). i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7) A <- Matrix::sparseMatrix(i, j, x = x) irlba::irlba(A) #> Warning in irlba::irlba(A): You're computing too large a percentage of total #> singular values, use a standard svd instead. #> Error in irlba::irlba(A): function 'as_cholmod_sparse' not provided by package 'Matrix'Session info
sessionInfo() #> R version 4.3.1 (2023-06-16) #> Platform: x86_64-pc-linux-gnu (64-bit) #> Running under: CentOS Linux 7 (Core) #> #> Matrix products: default #> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so #> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0 #> #> locale: #> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C #> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 #> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 #> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C #> [9] LC_ADDRESS=C LC_TELEPHONE=C #> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C #> #> time zone: Australia/Melbourne #> tzcode source: system (glibc) #> #> attached base packages: #> [1] stats graphics grDevices utils datasets methods base #> #> loaded via a namespace (and not attached): #> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-2 xfun_0.41 #> [5] lattice_0.21-8 glue_1.6.2 knitr_1.45 htmltools_0.5.7 #> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1 #> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0 #> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7 #> [21] rlang_1.1.2 fs_1.6.3
With Matrix 1.6-1
# Example taken from help("sparseMatrix", package = "Matrix"). i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7) A <- Matrix::sparseMatrix(i, j, x = x) irlba::irlba(A) #> Warning in irlba::irlba(A): You're computing too large a percentage of total #> singular values, use a standard svd instead. #> $d #> [1] 49.00000 44.27189 35.00000 28.00000 21.00000 #> #> $u #> [,1] [,2] [,3] [,4] [,5] #> [1,] 1.179612e-16 -6.418477e-17 -1.942890e-16 6.938894e-17 1.110223e-16 #> [2,] 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 #> [3,] 3.105155e-16 3.162278e-01 -6.245005e-17 1.110223e-16 -3.816392e-17 #> [4,] -1.734723e-16 -2.428613e-17 -3.885781e-16 5.273559e-16 1.000000e+00 #> [5,] -1.942890e-16 -2.602085e-16 9.436896e-16 1.000000e+00 -3.608225e-16 #> [6,] -2.498002e-16 -2.012279e-16 -1.000000e+00 1.054712e-15 2.775558e-17 #> [7,] 9.419548e-16 9.486833e-01 -1.318390e-16 2.914335e-16 -1.734723e-17 #> [8,] 1.000000e+00 -1.334002e-15 -3.885781e-16 -3.053113e-16 8.326673e-17 #> #> $v #> [,1] [,2] [,3] [,4] [,5] #> [1,] -6.938894e-17 -1.474515e-16 -1.665335e-16 -1.387779e-16 0.000000e+00 #> [2,] -9.714451e-17 -9.020562e-17 2.220446e-16 5.551115e-17 1.665335e-16 #> [3,] 6.071532e-18 1.192622e-17 2.081668e-17 1.040834e-17 1.734723e-18 #> [4,] 1.127570e-17 1.474515e-17 3.122502e-17 2.775558e-17 -1.040834e-17 #> [5,] -1.994932e-17 -4.770490e-17 -4.857226e-17 -1.734723e-17 -1.040834e-17 #> [6,] -7.632783e-17 5.898060e-17 -3.330669e-16 6.938894e-16 1.000000e+00 #> [7,] -8.326673e-17 -5.898060e-17 1.249001e-15 1.000000e+00 -3.053113e-16 #> [8,] -3.816392e-16 -1.734723e-16 -1.000000e+00 9.714451e-16 2.775558e-17 #> [9,] 1.065120e-15 1.000000e+00 -1.474515e-16 2.029626e-16 -1.908196e-16 #> [10,] 1.000000e+00 -1.234689e-15 -5.134781e-16 2.012279e-16 2.463307e-16 #> #> $iter #> [1] 1 #> #> $mprod #> [1] 16Session info
sessionInfo() #> R version 4.3.1 (2023-06-16) #> Platform: x86_64-pc-linux-gnu (64-bit) #> Running under: CentOS Linux 7 (Core) #> #> Matrix products: default #> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so #> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0 #> #> locale: #> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C #> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 #> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 #> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C #> [9] LC_ADDRESS=C LC_TELEPHONE=C #> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C #> #> time zone: Australia/Melbourne #> tzcode source: system (glibc) #> #> attached base packages: #> [1] stats graphics grDevices utils datasets methods base #> #> loaded via a namespace (and not attached): #> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-1 xfun_0.41 #> [5] lattice_0.21-8 glue_1.6.2 knitr_1.45 htmltools_0.5.7 #> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1 #> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0 #> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7 #> [21] rlang_1.1.2 fs_1.6.3-- Reply to this email directly or view it on GitHub: https://github.com/bwlewis/irlba/issues/70 You are receiving this because you were mentioned.
Message ID: @.***>
Re-installing irlba from source after upgrading to Matrix v1.6-2 seems to have resolved the error.
# Example taken from help("sparseMatrix", package = "Matrix").
i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7)
A <- Matrix::sparseMatrix(i, j, x = x)
irlba::irlba(A)
#> Warning in irlba::irlba(A): You're computing too large a percentage of total
#> singular values, use a standard svd instead.
#> $d
#> [1] 49.00000 44.27189 35.00000 28.00000 21.00000
#>
#> $u
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] -4.076600e-17 -1.873501e-16 -1.387779e-17 -2.775558e-17 1.387779e-16
#> [2,] 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00
#> [3,] 8.023096e-18 -3.162278e-01 -1.942890e-16 -1.873501e-16 1.040834e-16
#> [4,] 3.469447e-18 -1.665335e-16 7.216450e-16 1.426637e-14 -1.000000e+00
#> [5,] -3.122502e-17 -1.249001e-16 -4.440892e-16 1.000000e+00 1.459943e-14
#> [6,] 1.301043e-16 5.689893e-16 -1.000000e+00 -4.440892e-16 -8.881784e-16
#> [7,] -6.938894e-18 -9.486833e-01 -4.857226e-16 -5.273559e-16 8.326673e-17
#> [8,] 1.000000e+00 -1.656661e-16 1.578598e-16 -2.081668e-17 -5.551115e-17
#>
#> $v
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] -7.047314e-18 1.214306e-17 3.469447e-18 6.938894e-18 6.938894e-18
#> [2,] 2.081668e-17 -9.020562e-17 3.053113e-16 1.110223e-16 1.110223e-16
#> [3,] 8.890458e-18 -1.214306e-17 1.387779e-17 -3.469447e-18 -5.551115e-17
#> [4,] -4.250073e-17 5.204170e-17 -3.469447e-17 6.938894e-18 0.000000e+00
#> [5,] -2.222614e-17 3.122502e-17 -1.734723e-17 -3.469447e-18 1.387779e-17
#> [6,] -3.729655e-17 -5.551115e-17 1.054712e-15 1.915135e-14 -1.000000e+00
#> [7,] -1.175275e-16 -9.714451e-17 -6.522560e-16 1.000000e+00 1.948441e-14
#> [8,] -2.211772e-17 3.920475e-16 -1.000000e+00 -7.563394e-16 -1.415534e-15
#> [9,] 3.642919e-17 -1.000000e+00 -8.187895e-16 -3.677614e-16 2.012279e-16
#> [10,] 1.000000e+00 -5.204170e-18 1.600282e-16 5.854692e-17 -8.847090e-17
#>
#> $iter
#> [1] 1
#>
#> $mprod
#> [1] 16
Session info
sessionInfo()
#> R version 4.3.1 (2023-06-16)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: CentOS Linux 7 (Core)
#>
#> Matrix products: default
#> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so
#> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
#> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
#> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
#> [9] LC_ADDRESS=C LC_TELEPHONE=C
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#>
#> time zone: Australia/Melbourne
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> loaded via a namespace (and not attached):
#> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-2 xfun_0.41
#> [5] lattice_0.22-5 glue_1.6.2 knitr_1.45 htmltools_0.5.7
#> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1
#> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0
#> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7
#> [21] rlang_1.1.2 fs_1.6.3
I never fully understand when packages have to be re-built/re-installed when dependencies like Matrix update. But if I had to guess, perhaps these remaps (https://github.com/cran/Matrix/blob/323f4d9b0722ba6f61682874a8d92ebdbf93d8fa/inst/include/Matrix/remap.h) are picked up when irlba is re-installed from source.
Also good to know.
There is a re-write of the sparse/Matrix-dependency code that has been planned for a long time that will simplify installation and handle more special structured matrix cases. I will work on this this fall.
On 11/9/23, Peter Hickey @.***> wrote:
Re-installing irlba from source after upgrading to Matrix v1.6-2 seems to have resolved the error.
# Example taken from help("sparseMatrix", package = "Matrix"). i <- c(1,3:8); j <- c(2,9,6:10); x <- 7 * (1:7) A <- Matrix::sparseMatrix(i, j, x = x) irlba::irlba(A) #> Warning in irlba::irlba(A): You're computing too large a percentage of total #> singular values, use a standard svd instead. #> $d #> [1] 49.00000 44.27189 35.00000 28.00000 21.00000 #> #> $u #> [,1] [,2] [,3] [,4] [,5] #> [1,] -4.076600e-17 -1.873501e-16 -1.387779e-17 -2.775558e-17 1.387779e-16 #> [2,] 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 #> [3,] 8.023096e-18 -3.162278e-01 -1.942890e-16 -1.873501e-16 1.040834e-16 #> [4,] 3.469447e-18 -1.665335e-16 7.216450e-16 1.426637e-14 -1.000000e+00 #> [5,] -3.122502e-17 -1.249001e-16 -4.440892e-16 1.000000e+00 1.459943e-14 #> [6,] 1.301043e-16 5.689893e-16 -1.000000e+00 -4.440892e-16 -8.881784e-16 #> [7,] -6.938894e-18 -9.486833e-01 -4.857226e-16 -5.273559e-16 8.326673e-17 #> [8,] 1.000000e+00 -1.656661e-16 1.578598e-16 -2.081668e-17 -5.551115e-17 #> #> $v #> [,1] [,2] [,3] [,4] [,5] #> [1,] -7.047314e-18 1.214306e-17 3.469447e-18 6.938894e-18 6.938894e-18 #> [2,] 2.081668e-17 -9.020562e-17 3.053113e-16 1.110223e-16 1.110223e-16 #> [3,] 8.890458e-18 -1.214306e-17 1.387779e-17 -3.469447e-18 -5.551115e-17 #> [4,] -4.250073e-17 5.204170e-17 -3.469447e-17 6.938894e-18 0.000000e+00 #> [5,] -2.222614e-17 3.122502e-17 -1.734723e-17 -3.469447e-18 1.387779e-17 #> [6,] -3.729655e-17 -5.551115e-17 1.054712e-15 1.915135e-14 -1.000000e+00 #> [7,] -1.175275e-16 -9.714451e-17 -6.522560e-16 1.000000e+00 1.948441e-14 #> [8,] -2.211772e-17 3.920475e-16 -1.000000e+00 -7.563394e-16 -1.415534e-15 #> [9,] 3.642919e-17 -1.000000e+00 -8.187895e-16 -3.677614e-16 2.012279e-16 #> [10,] 1.000000e+00 -5.204170e-18 1.600282e-16 5.854692e-17 -8.847090e-17 #> #> $iter #> [1] 1 #> #> $mprod #> [1] 16Session info
sessionInfo() #> R version 4.3.1 (2023-06-16) #> Platform: x86_64-pc-linux-gnu (64-bit) #> Running under: CentOS Linux 7 (Core) #> #> Matrix products: default #> BLAS: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRblas.so #> LAPACK: /stornext/System/data/apps/R/R-4.3.1/lib64/R/lib/libRlapack.so; LAPACK version 3.11.0 #> #> locale: #> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C #> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 #> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 #> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C #> [9] LC_ADDRESS=C LC_TELEPHONE=C #> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C #> #> time zone: Australia/Melbourne #> tzcode source: system (glibc) #> #> attached base packages: #> [1] stats graphics grDevices utils datasets methods base #> #> loaded via a namespace (and not attached): #> [1] digest_0.6.33 fastmap_1.1.1 Matrix_1.6-2 xfun_0.41 #> [5] lattice_0.22-5 glue_1.6.2 knitr_1.45 htmltools_0.5.7 #> [9] rmarkdown_2.25 lifecycle_1.0.4 cli_3.6.1 grid_4.3.1 #> [13] reprex_2.0.2 withr_2.5.2 compiler_4.3.1 rstudioapi_0.15.0 #> [17] tools_4.3.1 irlba_2.3.5.1 evaluate_0.23 yaml_2.3.7 #> [21] rlang_1.1.2 fs_1.6.3I never fully understand when packages have to be re-built/re-installed when dependencies like Matrix update. But if I had to guess, perhaps these remaps (https://github.com/cran/Matrix/blob/323f4d9b0722ba6f61682874a8d92ebdbf93d8fa/inst/include/Matrix/remap.h) are picked up when irlba is re-installed from source.
-- Reply to this email directly or view it on GitHub: https://github.com/bwlewis/irlba/issues/70#issuecomment-1804640397 You are receiving this because you were mentioned.
Message ID: @.***>
And a bit more info now posted by the Matrix developers: https://stat.ethz.ch/pipermail/r-package-devel/2023q4/010051.html
Thanks @PeteHaitch.
@bwlewis Note that it's enough for you to bump irlba version and resubmit to CRAN. The new version will then propagate to the user machine which will trigger a reinstallation.
The problem affects any user who installed irlba before Matrix 1.6-2 got into the wild and landed on their machine. Unfortunately the info posted by the Matrix developers won't help them, and it's not realistic to expect that the thousands of users in that situation reinstall irlba manually. Most users will see the error in a context involving many packages and won't have a clue what package is actually the culprit. This is the case for Bioconductor users where 40+ Bioconductor packages are affected indirectly via irlba.
However, if a new version of irlba shows up on CRAN, then the problem will be solved for all the users and system admins that run update.packages() on a regular basis.
Thanks, H.
I tried reinstalling irlba, but the issue persists with Matrix 1.6-3. Major issue for anyone running UMAP.
@sfplab Did you re-install both from source? This is what worked for me:
install.packages("Matrix", type = "source")
install.packages("irlba", type = "source")
Hi @PeteHaitch - I hadn't installed from source, because it threw an error on my M1:
ld: warning: search path '/opt/R/arm64/gfortran/lib/gcc/aarch64-apple-darwin20.2.0/11.0.0' not found ld: warning: search path '/opt/R/arm64/gfortran/lib' not found ld: library 'gfortran' not found
After going down some rabbit holes, I finally found this page on CRAN - downloading the gnu fortran package at the top of the page solved my issues installing Matrix and irlba from source: https://cran.r-project.org/bin/macosx/tools/
Leaving this info here in case anyone else has issues with fortran-dependent source install in R. And yes, RunUMAP calls to Matrix & irlba now work as expected. Thank you, @PeteHaitch!
I use an M1 and I know that there's a lot of conflicting advice online re macOS, arm64 and R. I try to follow the advice of Simon Urbanek (R/CRAN's macOS maintainer) because that's what's officially supported and works well for me.
I have an R package that depends on irlba working - is there any solution on the horizon that does not require all my users to install from source?
Yes, I am preparing an update for CRAN presently.
On Tue, Nov 28, 2023 at 6:33 AM Kristoffer Vitting-Seerup < @.***> wrote:
I have an R package that depends on irlba working - is there any solution on the horizon that does not require all my users to install from source?
— Reply to this email directly, view it on GitHub https://github.com/bwlewis/irlba/issues/70#issuecomment-1829644640, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABCZHR4FZBBTYAO2ILXV3TYGXDY7AVCNFSM6AAAAAA7DXI7PGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRZGY2DINRUGA . You are receiving this because you were mentioned.Message ID: @.***>
Is there any solutions now ? I meet the same problem and it does not work even I changed Matrix version changes to 1.6-1.1. seurat v 4.4.0 seuratobject v4.1.4 The below is the code.
scRNA1 <- RunUMAP(scRNA1, reduction = "pca", dims = 1:20)
Warning: The default method for RunUMAP has changed from calling Python UMAP via reticulate to the R-native UWOT using the cosine metric To use Python UMAP via reticulate, set umap.method to 'umap-learn' and metric to 'correlation' This message will be shown once per session 09:40:18 UMAP embedding parameters a = 0.9922 b = 1.112 09:40:18 Read 12279 rows and found 20 numeric columns 09:40:18 Using Annoy for neighbor search, n_neighbors = 30 09:40:18 Building Annoy index with metric = cosine, n_trees = 50 0% 10 20 30 40 50 60 70 80 90 100% [----|----|----|----|----|----|----|----|----|----| **************************************************| 09:40:19 Writing NN index file to temp file C:\Users\Admini\AppData\Local\Temp\RtmpiCBaL8\filed0c4d6a7319 09:40:19 Searching Annoy index using 12 threads, search_k = 3000 09:40:19 Annoy recall = 100% 09:40:20 Commencing smooth kNN distance calibration using 12 threads with target n_neighbors = 30 09:40:20 Initializing from normalized Laplacian + noise (using irlba) Error in irlba::irlba(L, nv = n, nu = 0, maxit = iters) : function 'sexp_as_cholmod_sparse' not provided by package 'Matrix'
@bwlewis Any ETA for a new version of irlba on CRAN? Just to be clear, you don't need to make any change to the package, only bump its version and resubmit to CRAN. Thanks!
I am using mac M3, after install gfortran-12.2-universal.pkg then I can install
install.packages("Matrix", type = "source")
install.packages("irlba", type = "source")
I am using mac M3, after install gfortran-12.2-universal.pkg then I can install
install.packages("Matrix", type = "source") install.packages("irlba", type = "source")
I also met the same problem and your method solved it, thanks!
I have tried installing Matrix and irlba packages from source after installing gfortran-12.2-universal.pkg as suggested. However, I get the following error while trying to install Matrix package. Here is the screenshot. Can anyone help please?
Installing gfortran-ARM-13.2-Sonoma.dmg for Sonoma 14 macOS made it possible for me to install the Matrix and irlba packages from source and this in turn solved the error "function 'as_cholmod_sparse' not provided by package 'Matrix'".
I have tried installing Matrix and irlba packages from source after installing gfortran-12.2-universal.pkg as suggested. However, I get the following error while trying to install Matrix package. Here is the screenshot. Can anyone help please?
Thank you all!
@VitorAguiar Your comment is very misleading! Nobody said that you had to install Matrix v1.6_1.1 to fix the "as_cholmod_sparse" issue. If your version of Matrix is >= 1.6.2, all you need to do to fix this issue is reinstall irlba from source. And that does solve the problem.
Also why would you install an archived version of Matrix? Current version is 1.6-5 and this is what everybody uses so it should work with Seurat. The fact that Seurat v4.1.1 required at least v1.6-3 doesn't mean that you must install this exact version.
Anyways it's unfortunate that 5 months have passed and irlba version still didn't get bumped (with resubmission to CRAN) as this would have solved the problem once for all, and by doing so avoided a lot of confusion and frustration.
@hpages I was just sharing what worked for me. Your suggestion to reinstall irlba from source and current version of Matrix did not solve the problem for me. That's why I had to install an archived version of Matrix, one that was new enough to work with Seurat. Since you feel that my comment was misleading, I am deleting it.
@VitorAguiar The breaking change in Matrix happened when they went from version 1.6.1 to 1.6.2. So why installing Matrix 1.6.5 (current version) vs 1.6-3 would make any difference as far as the "as_cholmod_sparse" issue is concerned? Therefore when you say that
install.packages("Matrix", type = "source")
install.packages("irlba", type = "source")
didn't fix the problem for you, but that specifically installing Matrix 1.6.3 did solve it, I suspect that something else was going on in your case. And since we don't have the full story (i.e. output of commands + sessionInfo()) I guess we'll never know.
In any case it is very very unlikely that installing Matrix 1.6.3 instead of the current Matrix was necessary. Just wanted to clarify this in case other people seeing your comment might think this is a good idea .
In case anyone is having issues with this solution and also has renv installed, beware that the source install may be ignored in favor of a cached version. This solved the issue for me: renv::install("irlba", type = "source", rebuild = TRUE)
Solved matrix errors by downgrading to version 1.6-5 and then installing irlba from source
install.packages("https://cran.r-project.org/src/contrib/Archive/Matrix/Matrix_1.6-5.tar.gz", repos = NULL, type = "source")
install.packages("irlba", type = "source")
