irlba

irlba copied to clipboard

Consistency in sign.

6

I know that principal value decomposition is defined up to sign, but I think that for reproducibility, `irlba()` should be consistent. Ideally, it should follow the same convention as the...

eliocamp

BLAS/LAPACK routine 'DLASCL' gave error code -4

62

``` Error in irlba(t(x), nv = n) : BLAS/LAPACK routine 'DLASCL' gave error code -4 ``` Any ideas?

ShanSabri

Enable use of other matrix representations in C code.

5

This PR allows the use of any double-precision matrix class inside the `fastpath` C code. For example: ```r library(irlba) mat

LTLA

Expose C++ multiplication interface

2

Hey there! My lab spends a lot of time taking SVDs of, let's say, "nearly nice" matrices. For these matrices, we can often compute `t(A) %*% x` and `A %*%...

alexpghayes

** On entry to DLASCLS parameter number 4 had an illegal value

5

I got this message when running `irlba(large_mat, nv = 100)`. Is it something I can ignore or what is happening here? Also, the R seems to be stuck after outputting...

albert-ying

Inconsistent results with complex matrices

2

I'm using {irlba} to perform svd on complex matrices and found that the results where inconsistent. In my real code, I had an issue that the resulting principal values depended...

eliocamp

Different results obtained on the same input stored using dense or sparse matrices

2

I've been using the `irlba` package on the same input stored both as a dense and as a sparse matrix; I noticed that the PCA output is influenced by the...

AndiMunteanu

Rationale behind some of the bits and pieces in the C code

4

I've been writing a C++ port of **irlba**'s C code (https://github.com/LTLA/CppIrlba) for use in some other applications of mine. It's been... tough going, but I think I've gotten it right,...

LTLA

irlba's implementation of prcomp loses rotation rownames

1

Hi, I've rewritten my code to replace prcomp with prcomp_irlba (so much faster 🙌 ), and got an error because, downstream, I call rownames of the rotation matrix, however in...

leahrosen

Convergence criteria

2

I work on Soft-Impute and Soft-SVD algorithms described in [Matrix Completion and Low-Rank SVD via Fast Alternating Least Squares](https://arxiv.org/pdf/1410.2596.pdf). They proposed to keep track the change of Frobenius norm in...

dselivanov