I wrote a few different calculations of d(gamma_p)/da
and I’d like to document where the different versions perform best and where I should use limits/approximations. I know I have at least one where the current version does as badly as 1.0e-3 for error and the new version is much better.
I was trying to modify Vincent’s Mathematica function for computing gradients. For d(gamma_p(a,z)/da
I’d like to get it to return the point the gradient was calculated at along with the gradient itself to make testing less confusing in C++. So the table I’m hoping for has the columns:
a z d(gamma_p)/da d(gamma_p)/dz
... ... .... ...
Any chance I could get a hand with getting this to work? So far I’ve only managed to screw it up.
https://www.wolframcloud.com/objects/86879aa3-c1cc-4265-9ab6-407ec35446ab