A probabilistic programming language that combines automatic differentiation, automatic marginalization, and automatic conditioning within Monte Carlo methods.
This pull request is WIP that creates utility functions to implement the standard CDF functions without using Boost.
incomplete beta function created, ibeta (Note: These are regularized, meaning that the range of the function is in [0, 1])
incomplete gamma (lower and upper) created
Regimes where asymptotic series work better are added too
constant.birch created where constants have been placed (Note: For specific implementations, constants might change, like TINY/STOP constants and this is reflected in the PR)
test_cdf.birch epsilon threshold changed.
TODO:
Test utility functions to ensure that they work in the appropriate way (potentially using boost)
Digamma is not yet changed but it might be required in the future
Thanks for all the efforts on this @pranavsubramani, this puts us one step closer to speeding up compile times, and eliminating some complex dependencies for GPU and other hardware support. Now merged!
This pull request is WIP that creates utility functions to implement the standard CDF functions without using Boost.
ibeta
(Note: These are regularized, meaning that the range of the function is in[0, 1]
)constant.birch
created where constants have been placed (Note: For specific implementations, constants might change, like TINY/STOP constants and this is reflected in the PR)test_cdf.birch
epsilon threshold changed.TODO:
Digamma
is not yet changed but it might be required in the future