-
Notifications
You must be signed in to change notification settings - Fork 100
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add ChainRules definitions for gamma(a, x)
, loggamma(a, x)
, and gamma_inc
#305
Conversation
Codecov Report
@@ Coverage Diff @@
## master #305 +/- ##
==========================================
- Coverage 89.31% 88.35% -0.97%
==========================================
Files 12 12
Lines 2640 2653 +13
==========================================
- Hits 2358 2344 -14
- Misses 282 309 +27
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
Great, thanks! It would be good to resolve the testing question. |
Soon it will be possible to test (existing and new) partially implemented differentials properly: JuliaDiff/ChainRulesTestUtils.jl#140 |
Should this one be update now that JuliaDiff/ChainRulesTestUtils.jl#140 is merged? |
The drone runs seems to have connectivity issues. Not sure why. |
I added ChainRules definitions for
gamma(a, x)
,loggamma(a, x)
, andgamma_inc
according to https://functions.wolfram.com/GammaBetaErf/GammaRegularized/introductions/Gammas/ShowAll.html. Similar to the Bessel functions, derivatives with respect to the first argumenta
are not implemented since they are given in terms of a (regularized) hypergeometric function.Unfortunately, it seems due to the error thrown for derivatives of the first argument these rules can't be tested with
test_frule
andtest_rrule
since these functions always unthunk all derivatives. It would be good if one could somehow exclude some derivatives from the tests completely and not only from the finite differencing checks. I'll open an issue in ChainRulesTestUtils regarding this question.(BTW I assume that due to this issue the derivatives of the Bessel functions are not tested currently)