Jan Kwakkel

> > Another more sophisticated direction is to use adaptive sparse grids ([see this thesis](https://repository.tudelft.nl/islandora/object/uuid%3A5fafb525-2017-4521-8af6-29034327c4a8?collection=education)) > > Looks very interesting. Do we have this in the workbench? Do we want...

> I'm new to the sensitivity analysis libraries in the ABM context. The sensitivity analysis of my past work, on economic models, is mainly guided by intuition. I am eager...

I agree with @rht that showing how other tools can be used in conjunction with MESA is the way to go. However, if you try to use numpy.qmc or SALib...

> > The EMA workbench has a MESA example: https://emaworkbench.readthedocs.io/en/latest/examples/example_mesa.html so it would be easy to point to that rather than repeat it. > > From my reading of the...

If I understand it correctly, this PR is superseded by #2003. So this PR can be closed?

Thanks for checking this. It was also on my list of things to check. I am surprised by the overhead of weakrefs. I knew it is done in pure python...

So, I ran into the slowness again. While playing with #1994, I moved `Cell.agents` to `Cell._agents` and turned `Cell.agents` into a property that returned an AgentSet. On WolfSheep, this doubles...

So the problem seems to be WeakkeyDictionary which is a pure python implementation. It might be possible to substantially improve our speed without losing the use of weakrefs.

A quick update: I looked into the current implementation of WeakkeyDict. It is basically a clever wrapper around a normal dict with `weakkey.ref()` keys. Some minor speedups may be possible...

Another experiment. I changed the implementation of Agentset.shuffle in 2 ways: I shuffle the weakrefs themselves, and it defaults to inplace updating (quick dirty check, can be changed also within...