t;>,
"rcpp-devel@lists.r-forge.r-project.org
<mailto:rcpp-devel@lists.r-forge.r-project.org>"
mailto:rcpp-devel@lists.r-forge.r-project.org>>
Subject: Re: [Rcpp-devel] Speed of RCppEigen Cholesky decomposition on
sparse mat
---
>
> Message: 1
> Date: Tue, 27 Nov 2018 15:33:55 +0100
> From: Serguei Sokol
> To: "Hoffman, Gabriel" ,
> "rcpp-devel@lists.r-forge.r-project.org"
>
> Subject: Re: [Rcpp-devel]
This is really a brilliant observation!
Best,
Yixuan
On Tue, Nov 27, 2018 at 9:34 AM Serguei Sokol wrote:
>
> Le 26/11/2018 à 18:23, Hoffman, Gabriel a écrit :
> > I am developing a statistical model and I have a prototype working in R
> > code. I make extensive use of sparse matrices, so the
Le 26/11/2018 à 18:23, Hoffman, Gabriel a écrit :
I am developing a statistical model and I have a prototype working in R
code. I make extensive use of sparse matrices, so the R code is pretty
fast, but hoped that using RCppEigen to evaluate the log-likelihood
function could avoid a lot of memor
I guess BLAS plays a minor role here. The operation here is a sparse
Cholesky decomposition, and if I'm correct the Matrix package in R uses
SuiteSparse/CHOLMOD to do the factorization, while Eigen uses its own
implementation. The underlying numerical algorithms may be different.
Best,
Yixuan
On
If I recall correctly, Eigen does not use the BLAS, but has reprogrammed
the operations in C++. If you want to take advantage of a fast system BLAS,
try Armadillo, or in this case RcppArmadillo.
Thanks,
Avi
On Mon, Nov 26, 2018 at 12:23 PM Hoffman, Gabriel
wrote:
> I am developing a statistica
Gabriel,
On 26 November 2018 at 17:23, Hoffman, Gabriel wrote:
| I am developing a statistical model and I have a prototype working in R
| code. I make extensive use of sparse matrices, so the R code is pretty
| fast, but hoped that using RCppEigen to evaluate the log-likelihood
(lowercase c g
I am developing a statistical model and I have a prototype working in R
code. I make extensive use of sparse matrices, so the R code is pretty
fast, but hoped that using RCppEigen to evaluate the log-likelihood
function could avoid a lot of memory copying and be substantially
faster. However, in
