I haven't been able to track all of this, but I think things are going in an odd direction.
- DenseVector is not a reasonable default for something that appears to be intended to support various flavors of sparse matrices. - null rows are not a problem. Interpret them as all zero - setting the number of rows and columns does not imply allocation. Neither does a legal access. The only thing that implies allocation is setting a value to non-zero. On Mon, Jul 11, 2011 at 9:06 PM, Lance Norskog (JIRA) <j...@apache.org>wrote: > > [ > https://issues.apache.org/jira/browse/MAHOUT-756?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13063705#comment-13063705] > > Lance Norskog commented on MAHOUT-756: > -------------------------------------- > > When it is this confusing, UnsupportedOperationException is permissible. > There is no obvious right thing to do. What if there are different Vector > classes and some empty slots? > > Using DenseVector/Matrix is the default policy in other places, so let's go > with that. If your program runs out of memory, you'll eventually find all of > the DenseVectors and track down the problem. In the meantime, it will be > slow but will work. > > > VectorList (Matrix implementation) does not maintain cardinality getters > correctly > > > ---------------------------------------------------------------------------------- > > > > Key: MAHOUT-756 > > URL: https://issues.apache.org/jira/browse/MAHOUT-756 > > Project: Mahout > > Issue Type: Bug > > Components: Math > > Reporter: Lance Norskog > > Priority: Minor > > Attachments: VectorList.patch > > > > > > VectorList (implements Matrix) is dynamically expandable, row-wise. There > are three different ways to query the size of a Matrix, and VectorList does > not correctly supply these values. > > > > -- > This message is automatically generated by JIRA. > For more information on JIRA, see: http://www.atlassian.com/software/jira > > >
