On Convergence Rates of Root-Seeking Methods

[Claudio L.]

(03-13-2018 11:22 PM)Mike (Stgt) Wrote:  BTW, to compute 7/5*inverse(sqrt(1-1/50)) to 50'000 digits the runtime is 0.81 seconds, about the same as for sqrt(2), but the delta of this and the previous sqrt takes ~10% more than the root itself.

I've used this one for quick calculations:
http://preccalc.sourceforge.net

It has some limited programmability, I always check newRPL's output against this one and Wolfram Alpha to make sure.