Kaya Unalmis

Quadrature in other coordinates will require more nodes since NFP periodicity is los

Also after this make use of partial summation in tranforms and basis

We should do partial summation regardless, but there also exists a closed form solution to this root finding problem. It is given by equation 19.36 in Boyd's chebyshev and fourier...

It won't beat ~

It could also be useful for initial guess if convergence has ever been an issue

That's a good idea, thanks Rory. I've been doing some reading form Boyd's spectral book and thinking for the optimal method to compute these quantities over the last week. my...

(if you're hooking in ffts for chebyshev double fourier, the order for transform from real to spectral should coeffs = rfft2(dct(f)) since dct returns a real output).

Won't the speed improve a bunch if we do `rfft2(dct(f, radial axis), flux axes)` and the inverse `idct(irfft2)`

I marked the two I knew were done and added couple

Yes I think using partial summation techniques would be significant. Some thoughts - The given example in 3 may require #1207 first, otherwise the nodes would change in the Newton...