stillwater-sc / universal / 23123625508

85%

master: 84%

push

github

feat(unum1): implement math functions (#577)

* feat(unum1): implement math functions for unum Type I

Implement standard math library via double-precision delegation:
sqrt, exp, log, log2, log10, pow, sin, cos, tan, asin, acos, atan,
sinh, cosh, tanh, abs, min, max, floor, ceil, trunc.

All functions handle NaN propagation and domain errors (e.g.,
sqrt(negative) = NaN, log(0) = NaN, asin(2) = NaN). The ubit is
automatically set by operator=(double) for inexact results.

Move abs() from unum_impl.hpp to math_functions.hpp to avoid
redefinition.

Resolves #569

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix(unum1): guard non-finite results and fix min/max NaN handling

Address CodeRabbit review findings:

1. Non-finite native results (inf from exp/pow/sinh/cosh) now map to
   NaN via from_native() helper, since unum Type I has no infinity
   encoding. Also guard operator=(double) against infinity input.

2. min/max now explicitly propagate NaN instead of relying on
   unordered comparison semantics which produced argument-order-
   dependent results.

Add test cases for overflow-to-NaN and min/max NaN propagation.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

65 of 65 new or added lines in 2 files covered. (100.0%)

43397 of 51580 relevant lines covered (84.14%)

6109012.89 hits per line