Niko Savola

This brings to mind that some large PCells might take a few seconds to update. In this case if one is updating many parameters, it might be nice to have...

> Thanks, but do you have a proof of benfits (e.g. runtime improvement on critical execution paths), or is that a hypothetical idea? One suggestion is that you try that...

I also wonder if this `matrix.to_string` call is somehow inefficient or is it just called a lot of times https://github.com/KLayout/klayout/blob/f7ef538f343a208603288ed3925b561b00d2336a/src/db/db/dbMatrix.cc#L353-L360

Hello, I could work on this issue for UnitaryHack. I guess a smooth gradient would look good for all of these? Changes in `R_hat` seem to be much smaller than...

> Hey @nikosavola , you still up for this? do you need some guidance? I concentrated the code from https://github.com/netket/netket/pull/1098 a bit by having just one `color_curve`, which is expected...

> (Unrelated, but you also tried the GCNN on the J1-J2 right? Did it work well? I've seen a plot on your repository that looked as if it was not...

If someone knows of a way to _hash_ current state of Ansys or somehow check if the simulations were edited, this should be easy.

You can update the syntax 👍

Should this issue be closed as of #116 ?

I'm a bit confused about the tests, do they currently require some previous version to compare to? Additionally, is it enough to compile everything with: ```bash cd src make make...