[
This project collects various ideas about mathematical constants for C++.
Most of these constants are now in the standard for C++20. These versions under namespace emsr will work under C++17.
There are several directions this could go once the basics are in place.
-
template<typename Constant, typename NumTp> class floating_point_constant { NumTp fact; };
-
Constant is able to belch out mantissa bits - as many as needed for the type of float NumTp. Actually NumTp could be floating-point, rational or fixed-point.
-
Math operations involving floating_point_constant would modify the factor and accumulate operations using template programming. Ratios need to cancel the constant.
-
Only evaluate at the end and with the best precision.
-
This could work a lot like units.