[
https://issues.apache.org/jira/browse/MAHOUT-45?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12591347#action_12591347
]
Ted Dunning commented on MAHOUT-45:
-----------------------------------
Sounds right to me. Note that there may be some ordering constraints ... but I
think you are right that Givens is considered easier to parallelize.
Block householder with k x k blocks may be easy to parallelize across k
processes. The idea would be that instead of distributing the different
rotations to different processors, you would do the single block reflection in
parallel. The decomposition of each block could then be done completely in
parallel as well.
Probably it would be good to outline the approach for both algorithms and see
how it looks.
> Matrix QR decomposition
> -----------------------
>
> Key: MAHOUT-45
> URL: https://issues.apache.org/jira/browse/MAHOUT-45
> Project: Mahout
> Issue Type: Improvement
> Components: Matrix
> Reporter: Sergey Chickin
> Priority: Minor
> Attachments: MAHOUT-45.patch
>
>
> Matrix QR decomposition and appropriate determinant calculator
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
