Wonseok Shin

Another related capability I would like: ```julia @computed struct MyType{N,V

I guess you have calculated a wave emanating from some source. To calculate the overlap integral, you need to calculate the waveguide mode separately. This can be done by setting...

Thanks for pointing out the problems! 1. In the one-liner, the last argument `true` should be `false`. The name of this argument is `inspect_only`, which, when set `true`, forces the...

Hi Abhijeet, GPU support is certainly one of the features that I would like to add, but it is currently not a priority.

Memory is the limitation of direct methods for solving linear systems. For GPU I will implement iterative methods, which are extremely memory-efficient.

> This feature will be available thru the upcoming `qtTerm` which will handle pan/zoom, multiple windows and hover effects. Unfortunately, it's still WIP (but should be available before JuliaCon). Any...

I don't think checking the sameness of the entries inside `Statistics.mean` is realistic, because it will slow down the performance `mean(x)` significantly when all entries of `x` are identical: ```julia-repl...

The implementation of the above proposal should look something like this: ```julia function zscore2(X) μ = mean(X) Z = similar(X) Z .= X .- μ μ, σ = mean_and_std(Z) zscore!(Z,...

> Have you tried calling `std(Z, mean=0)` instead of `mean_and_std(Z)`? Though I'm not sure to what extent assuming that the mean is exactly zero can be a problem. Is this...

> Note that there are more efficient ways to do this. First, `x[1]` shouldn't be called for each entry. Second, `all` short-circuits, so its cost is negligible when all entries...