BUG: special/cdflib: fix fatal loss of precision issues in cumfnc #7858
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Replace gammaln(a+b)-gammaln(a)-gammaln(b) -> -betaln(a,b) to avoid loss of precision in cancellations for large a, and make use of gammaln(a+1) == gammaln(a) + log(a) for a > 0. Whether the previous code worked correctly or not for very large values depends on floating point issues --- it is dangerous to rely on gammaln(a) and gammaln(a+b) returning exactly the same floating point value when a==a+b up to fp precision, as this is apparently not guaranteed by gfortran optimizations.
|
LGTM, merged. Thanks Pauli |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Replace gammaln(a+b)-gammaln(a)-gammaln(b) -> -betaln(a,b)
to avoid loss of precision in cancellations for large a, and
make use of gammaln(a+1) == gammaln(a) + log(a) for a > 0.
Whether the previous code worked correctly or not for very large values depends
on floating point issues --- it is dangerous to rely on
gammaln(a) and gammaln(a+b) returning exactly the same floating point
value when a==a+b up to fp precision, as this is apparently not
guaranteed by gfortran optimizations.
Fixes gh-7828
There in principle may be other issues with this function since
it does not appear to be properly tested, but this is at least one obviously
correct fix...