Stephan Hilb

You can rewrite your kernel `mK` according to my previous example. Also I cannot reproduce your example: the following does not allocate for me on Julia 1.8.3 ```julia using BenchmarkTools...

I understand what you want to do, but you did not apply my example to your setting: If you have the remaining channel as last component it reads: `k =...

You are right, the API is actually documented fairly well, but end-user examples and a generated documentation website is still missing. This will come eventually. Regarding boundary conditions: accesses outside...

I don't think the readme is the place for that. You can look at the documentation of `Kernel` from within Julia, where the parameters are explained and the function `something`...

Thank you for the suggestions. I referred to the documentation in `src/types.jl`: ``` Kernel{X}(f) Create a kernel with axes `X` wrapping a kernel function `f`. The window function defines a...

I'm not sure I understand what you are searching for. It will already throw an error if the kernel makes out-of-bounds accesses on arrays without an extension defined. This is...

Just use an extension: `map(k, extend(a, StaticKernels.ExtensionReplicate()))` will return the same dimensions as `a`. As long as you are only interested in the inner part, any extension will do the...

Your output and input array for the kernel operation is the same, which means that you almost certainly compute the wrong thing here since you overwrite relevant input data. This...

Ok, in that case you can ignore my example since I assumed it was unintended. If you have this data dependency, then there are there are almost no options for...

right, type inference through `Base.promote_op` sometimes gives up early. 1. Quick fix for you: annotate your window function return type: `mW -> (sum(mW) / kernelNumPx)::Float64` 2. What we could do...