For SecuCalc I planned to add highly accurate numbers, so more than the default 64-bit floating point ones, relatively fast, HOWEVER it's hard and so much to do…
sin, cos, exp, log, sinh/cosh maybe, tan maybe, asin, acos, atan, …

and I want it to be accurate… addition, multiplication, subtraction, and division are already working well :) (if I convert the 64-bit floating point numbers to my own format, calculate, and convert back, the result is right :) | the format is 64 bits exponent, 1 bit sign (8 probably because of Java…), 32 * n (n e |N) bits mantissa; so numbers up to 10^5,550,000,000,000,000 are possible)

18167704s ago, by Antonio