Caleb Zulawski
Caleb Zulawski
Thanks for the PR! Sorry I took so long to get to this, I've been pretty busy lately. I like the direction this is going. @drubinstein had an idea though,...
With the parameter pack the dimensions are still constants, so I think so.
Yeah, I'll have to think about that one more. I'm pretty sure it's possible, but not obvious to me either at the moment. I think it's probably possible with a...
This works: ```c++ #include template struct data_impl; template using data_impl_t = typename data_impl::type; template struct data_impl { using type = T; }; template struct data_impl { using type = data_impl_t[First];...
That is possible, but the problem is that makes initialization frustrating--you need to also initialize it in one giant array. Using a multidimensional array is more ergonomic, I think.
Consider solving these with Taylor series, or a modified method such as Horner's method
For now, I've at least renamed them from ASSERT_(condition) to COTILA_DETAIL_ASSERT_(condition) to avoid any macro name reuse.
I've only implemented complex-to-complex transforms (no real-to-complex or real-to-real, see #5), so both the inputs and outputs are `Complex` or `Complex`. In memory, each complex number is just equal to...
I do believe that the performance difference between Fourier and RustFFT comes down to SIMD (for powers of two, at least). It might be worth looking into the arithmetic operation...