I haven't added the SIMD yet, but they're already a bit faster than using Zernike at low degree because of the symmetries that reduce the number of Givens rotations required and also the square layout of the coefficients is more friendly on the cache I think.
@dlfivefifty @ikaroruan
I haven't added the SIMD yet, but they're already a bit faster than using Zernike at low degree because of the symmetries that reduce the number of Givens rotations required and also the square layout of the coefficients is more friendly on the cache I think.