[
https://issues.apache.org/jira/browse/MAHOUT-269?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12805636#action_12805636
]
Ted Dunning commented on MAHOUT-269:
------------------------------------
I don't understand your point. If there are any values at all in the dense
vector, then it won't return MIN_VALUE.
On the other hand, for a sparse vector, there should be a check to return 0 if
the dimension of the vector is >0 and there were no non-zero elements.
What to return in the case of a 0-dimensional vector is not something I would
worry about.
> Vector.maxValue() returns Double.MIN_VALUE for vectors with all negative
> entries.
> ---------------------------------------------------------------------------------
>
> Key: MAHOUT-269
> URL: https://issues.apache.org/jira/browse/MAHOUT-269
> Project: Mahout
> Issue Type: Bug
> Components: Math
> Affects Versions: 0.1, 0.2
> Reporter: Jake Mannix
> Priority: Minor
> Fix For: 0.3
>
>
> {code}
> @Override
> public double maxValue() {
> double result = Double.MIN_VALUE;
> for (int i = 0; i < size(); i++) {
> result = Math.max(result, getQuick(i));
> }
> return result;
> }
> {code}
> Should be:
> {code}
> @Override
> public double maxValue() {
> double result = 0;
> for (int i = 0; i < size(); i++) {
> result = Math.max(result, Math.abs(getQuick(i)));
> }
> return result;
> }
> {code}
> Right? MaxValue should be returning the max of the absolute value, not the
> real max, right? And the maxValue of the zero vector is 0, not MIN_VALUE,
> yes?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
