Deterministic math functions in C# using decimals. Gotta match em all.
There's been some concern for a while about floating-point determinism in C#. This would be where the exact floating-point result of
calculations on float or double types can be different, depending on the platform that runs the program, its JIT compilation
settings, and so on. Visual Studio offers a checkbox for "Deterministic," which may be enough, but just in case it isn't...
Determon was mostly created to help support replays in games with procedural generation; in these cases, absolute mathematical
accuracy is less important than absolute reliability across all machines that can run the application. The MathM class provides
both of these; the ApproxM class only provides the second. The ApproxM class seems to be quite a bit faster than MathM, but both
are rather slow compared to Math and MathF. This is because all the math here operates on decimal numbers, rather than float
or double; this gives us determinism at the expense of speed.
Determon is licensed under MIT; see the file LICENSE for more. The MathM class is based primarily on Ramin Rahimzada's DecimalMath library, in a larger repo of C# helper classes. The approximation code is not from the same repo, and draws from a fairly wide range of approximations. The 1955 research study by RAND Corp, "Approximations for Digital Computers," was very useful for the inverse trigonometric function approximations (it uses Taylor series, but these seem to be good enough in this case). Some other approximations were published on the web in various articles. The Sqrt() methods in both the precise and approximate math are from Stack Overflow, answering this question.