This PR improves performance of binomial with 10-50% speedups with larger gains for BINV algorithm. Mostly optimizations were adding strictness and avoiding allocations.
@idontgetoutmuch it seems using larger than 10 threshold would be more efficient but didn't make necessary measurements for picking better one. So I left it as is. I plan to make release once this PR is merged
This PR improves performance of binomial with 10-50% speedups with larger gains for BINV algorithm. Mostly optimizations were adding strictness and avoiding allocations.
@idontgetoutmuch it seems using larger than 10 threshold would be more efficient but didn't make necessary measurements for picking better one. So I left it as is. I plan to make release once this PR is merged