Dana Jacobsen

RWH's numpy version is very fast and uses less memory than the others. His pure Python version is also quite fast. Both run circles around the current SymPy implementation, noting...

Thank you very much for the report, and also big thanks for the confirmation on 32-bit perl with ivsize == 4. Definitely something to fix!

I'm definitely interested. I looked at it a little a few months ago, and added "rootmod: a way to get all the roots" to the TODO. It should be easy...

@hvds I added XS for the general root case. I'll work on the PP code for it. Currently it shortcuts to the allsqrtmod code when given k=2. I'm thinking I...

Thanks. It looks like this was fixed with the big sqrtmod / rootmod rewrite last August. But it's been far too long since a release. ``` $ perl -MMath::Prime::Util=sqrtmod -wle...

The GMP code still needs a lot of changes to rootmod / sqrtmod, which is the main bottleneck for my release. I would like to get the GMP backend out...

Ah. Tonelli-Shanks and Cipolla work modulo a prime. It got substantially more complicated with both composite modulos as well as arbitrary k-th roots. Tonelli-Shanks still exists in the C code...

GMP code with full sqrtmod implementation has been pushed. My rootmod implementation (in C and PP) uses znlog to solve a discrete log, which isn't in GMP, so that would...

Also seems to be fixed in the sqrtmod / rootmod rewrite. Negative inputs are supported, and the documentation indicates we'll use the absolute value (just like Pari/GP). I tried to...

Hi JJ, thanks for the help! There are some generic things that apply to all projects. Quality checks in the documentation (spelling, missing, malformed, unclear, ambiguous, etc.). Build procedures. Test...