PFA implementation
- Dominant language
- Julia
- Stars
- 74
- Forks
- 8
- Avg merge
- 13h 43m
- Merged PRs (30d)
- 1
Description
I think the major missing algorithm in this package is the prime-factorization algorithm for mixed-radix FFTs where the factors are co-prime. I'm not entirely sure, but I think this is why the composite benchmark is substantially worse than the rest relative to FFTW for small-size ffts.
This one has some serious considerations for a few reasons:
- which radices to choose in a many-composite algorithm?
- How to effectively incorporate this into the callgraph?
- Is this _actually_ any better? I have no idea
For completeness, I should note that I've previously implemented Rader's algorithm at some point (I'm not sure if in C++ or julia...) and it did _not_ work well at all unless your prime is length $2^k - 1$, which I'll leave until someone actually struggles with the use-case.
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.