You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by "Trevor Grant (JIRA)" <ji...@apache.org> on 2017/06/23 04:19:00 UTC
[jira] [Commented] (MAHOUT-1942) Algorithms should be applicable to
incore matrices
[ https://issues.apache.org/jira/browse/MAHOUT-1942?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16060394#comment-16060394 ]
Trevor Grant commented on MAHOUT-1942:
--------------------------------------
Move to close this as "won't fix"
It has become evident that majority of algorithms fall into one of two categories-
1) Highly parallelizable- an incore version is distributed out via a map/reduce function, the distributed version is simply a wrapper
2) A not Highly parallelizable- In which case the optimal implementation for the incore and distributed version differ significantly.
By convention then, in the algorithms framework, models and fitters are presumed to be distributed unless `InCore` is the prefix of the class name.
> Algorithms should be applicable to incore matrices
> --------------------------------------------------
>
> Key: MAHOUT-1942
> URL: https://issues.apache.org/jira/browse/MAHOUT-1942
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.13.1
> Reporter: Trevor Grant
> Fix For: 0.13.2
>
>
> The functions in `org.apache.mahout.math.algorithms` should be able to be applied to incore matrices as well.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)