You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is probably a good idea to use exp2 as the fundamental exponential function (it has fewer roundings) and then just implement exp and exp10 with an extra multiple by log2(e) and log2(10) respectively. A similar strategy can be used for logarithms.
The text was updated successfully, but these errors were encountered:
Hey @oscardssmith , thanks for your interest in MultiFloats.jl! This is a great idea, as it simplifies the argument reduction step from subtracting an integer multiple of log(2) to simply subtracting an integer. I'll give this a shot, and if the benchmarks turn out favorably, you can expect to see this implemented in the next release.
It is probably a good idea to use
exp2
as the fundamental exponential function (it has fewer roundings) and then just implementexp
andexp10
with an extra multiple bylog2(e)
andlog2(10)
respectively. A similar strategy can be used for logarithms.The text was updated successfully, but these errors were encountered: