You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Feynman Liang (JIRA)" <ji...@apache.org> on 2015/06/30 22:33:04 UTC
[jira] [Comment Edited] (SPARK-7210) Test matrix decompositions for
speed vs. numerical stability for Gaussians
[ https://issues.apache.org/jira/browse/SPARK-7210?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14608976#comment-14608976 ]
Feynman Liang edited comment on SPARK-7210 at 6/30/15 8:32 PM:
---------------------------------------------------------------
Some relevant links:
[A R cookbook recipe|http://thirteen-01.stat.iastate.edu/snoweye/hpsc/?item=cookbook&subitem=ex_mvn] describing and implementing mvn likelihood using Cholesky decomposition.
[Another R reference with the same cholesky trick|http://gallery.rcpp.org/articles/dmvnorm_arma/], except this one computes the density.
[The C++ code called by R's `mvnfast` package |https://github.com/mfasiolo/mvnfast/blob/master/src/rmvnCpp.cpp]
was (Author: fliang):
Some relevant links:
[A R cookbook recipe|http://thirteen-01.stat.iastate.edu/snoweye/hpsc/?item=cookbook&subitem=ex_mvn] describing and implementing mvn densities using Cholesky decomposition.
[Another R reference with the same implementation as above|http://gallery.rcpp.org/articles/dmvnorm_arma/]
[The C++ code called by R's `mvnfast` package |https://github.com/mfasiolo/mvnfast/blob/master/src/rmvnCpp.cpp]
> Test matrix decompositions for speed vs. numerical stability for Gaussians
> --------------------------------------------------------------------------
>
> Key: SPARK-7210
> URL: https://issues.apache.org/jira/browse/SPARK-7210
> Project: Spark
> Issue Type: Improvement
> Components: MLlib
> Reporter: Joseph K. Bradley
> Priority: Minor
>
> We currently use SVD for inverting the Gaussian's covariance matrix and computing the determinant. SVD is numerically stable but slow. We could experiment with Cholesky, etc. to figure out a better option, or a better option for certain settings.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org