[
https://issues.apache.org/jira/browse/MATH-1024?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13748616#comment-13748616
]
Phil Steitz commented on MATH-1024:
-----------------------------------
I vaguely remember this being discussed once before. It is worth searching the
archives to see what, if any, rationale we had at the time for zeroing the
default in QR. It is OK, IMO, for the defaults to be different if there is a
good numerical reason for it (i.e., one algorithm can handle nearer-singular
matrices than the other).
> LU and QR have different default singularity thresholds
> -------------------------------------------------------
>
> Key: MATH-1024
> URL: https://issues.apache.org/jira/browse/MATH-1024
> Project: Commons Math
> Issue Type: Wish
> Affects Versions: 3.2
> Reporter: Evan Ward
> Priority: Minor
> Attachments: ToleranceTest.java
>
>
> LU uses 1e-11 and QR uses 0. This means by default QR never throws a
> Singularity exception. This caused divergence to go unreported in
> GaussNewtonOptimizer when it uses QR. I think a consistent default should be
> chosen. I will make the threshold explicit in GaussNewton.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
