Jake should probably make the call regarding the actual restartable SVD (e.g. the LanczosState and related stuff). As for the LinearOperator stuff I've been working on, it touches a lot of stuff in mahout-math so I think waiting for 0.6 would be more appropriate than rushing it into 0.5.
-Jon On Wed, May 4, 2011 at 5:17 AM, Sean Owen (JIRA) <[email protected]> wrote: > > [ > https://issues.apache.org/jira/browse/MAHOUT-319?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13028724#comment-13028724] > > Sean Owen commented on MAHOUT-319: > ---------------------------------- > > It sounds like this is still in progress -- may we move to 0.6? I didn't > push it myself since there has at least been some noise in the last 2 weeks, > and some patch up. > > > SVD solvers should be gracefully stoppable/restartable > > ------------------------------------------------------ > > > > Key: MAHOUT-319 > > URL: https://issues.apache.org/jira/browse/MAHOUT-319 > > Project: Mahout > > Issue Type: Improvement > > Components: Math > > Affects Versions: 0.3 > > Reporter: Jake Mannix > > Assignee: Jake Mannix > > Fix For: 0.5 > > > > Attachments: MAHOUT-319.diff, MAHOUT-319.patch > > > > > > LanczosSolver, DistributedLanczosSolver, and HebbianSolver all keep > copious amounts of memory-resident data which is lost if the app crashes or > is killed (OOM, forgetting to run in a screen session, and losing net > connectivity to the server running it, etc...). > > These algorithms (and many other Mahout processes!) should enable a > pluggable "persist state" mechanism (to HDFS, RDBMS, local disk, key-value > store, etc), and similarly, a way to pick up and start from such a state. > > -- > This message is automatically generated by JIRA. > For more information on JIRA, see: http://www.atlassian.com/software/jira >
