Simon Blanke

Depending on the start configuration of positions the particles might stick to the edges of the search-space when attempting to spiral: 

This looks much better, but there is nothing happening after convergence: 

I added some fixes for the rotation-matrix and added a decay-parameter to change the step-size (or more like: radius-factor) over time. In the following search-path-gif you can see what happens,...

From the search-path you can see how the algorithm is mostly exploration at the moment: 

Hello @djokester, you are very welcome to work on this (or another) issue. I have not started working on this algorithm yet. But there are some things you should keep...

Hello @djokester, I was able to implement the direct search, based on the explanation [in this fantastic article](https://link.springer.com/article/10.1007/s10898-020-00952-6#:~:text=Introduced%20in%201993%2C%20the%20DIRECT,hyperparameter%20(the%20desired%20accuracy).). It was much easier to understand this algorithm after I understood and...

Hello @alex-hh, thanks for opening this issue. Your problem sounds very interesting! I have never seen (or understood) this kind of optimization problem. Is the length L always the same...

Hello @alex-hh, very good. Gradient-Free-Optimizers does only support numerical search-spaces (numpy arrays), but the python package [Hyperactive](https://github.com/SimonBlanke/Hyperactive) is build on top of Gradient-Free-Optimizers and supports numbers, strings and even functions...

Okay, so the `distribution`-parameter is similar to a [numpy distribution](https://numpy.org/doc/1.16/reference/generated/numpy.random.normal.html#numpy.random.normal). Those distributions are continuous, but the GFO search-space is discrete. GFO automatically rounds the new position to the next integer....

Hello @alex-hh, > Is it straightforward currently to have this kind of sampling of candidate positions in GFO? It really depends on how this looks like in code. Could you...