Martin Möddel

From a slack discussion that took place yesterday I can contribute the following example, where a simple comment seems to switch the behavior of a cell from [non-interactive to interactive...

I managed to get it working ```julia function test_avx!(xre::Vector{T}, xim::Vector{T}, βre::T, βim::T, Are::Matrix{T}, Aim::Matrix{T}, k::Int) where {T

> +1 for complex number support. It's also an opportunity to be faster than C compiled with gcc/clang which does a bad job of vectorizing complex arithmetic. Speaking of which...

> Did you try `test!`, using those StructArrays as arguments? > You noted that `@avx` being faster wasn't surprising because the `StructArrays` save you from any shuffling. LLVM will also...

> Disappointing, I really expected this to be good enough: Would have been nice. > Interesting. Sounds like I'll have to change the heuristic. It was actually a post on...

> I added the ability to manually specify the unroll factor in v0.3.8, via specifying `unroll = U` or `tile = (U,T)` where U and T are integer literals (ie,...

You are right. This is also how I solved my problem in the end. Using `reinterpret` and hand crafting a kernel for my function ```julia function test!(x::Vector, y::Vector, beta) for...

The coefficients in the polynomial are inherently irrational (require sqrt for calculation) so simply using `Rational` would not do the trick. What are you trying to achieve?

In principle one could make different normalizations possible (#6), which would yield the polynomials you desire. However, this refactoring of the package would require some time, which I currently do...

Hi Felix, it does make sense to include your extensions into our package. Maybe with two small modifications 😉 1. I think it makes sense to check for integer overflow....