Github user vrilleup commented on the pull request:
https://github.com/apache/spark/pull/964#issuecomment-47722138
@mengxr Yes, you are right. Keeping API simple might be more important than
flexibility. I think tol can be set to a default value (e.g. 1e-10, which is
the default in matlab). But maxit is related to k, 300 would be enough for most
cases, or max(300, k*2) if k is large. For the naming, how about
computeTruncatedSVD? (I saw this term used in many papers, also in my thesis).
or computeSVDs (like svds in matlab)?
I think separating svd and svds in two functions is better than deciding
dense/sparse impl internally for the user. It's hard to enumerate all use cases
and decide an impl logic.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
