Ernesto Carrella

A working example of building and running a model in JS. We aren't quite there in the presentation yet, but we are getting close.

Introducing a new R package to estimate parameters from simulations via reference tables

A brief comment on a very cool paper coming to terms with under-identification in agent-based models by tricking a genetic algorithm to do some more exploration than what we usually get from calibration techniques

A few notes on estimating parameters in agent-based models, trying to tie together the literature and identify the disparate objectives we are trying to achieve

A few notes on estimating parameters in agent-based models, trying to tie together the literature and identify the disparate objectives we are trying to achieve

A short guide to setup MASON with Intellij

A quick post-mortem on a 3-day burst of agent-based modelling from scratch

A guest lecture at the KKP in Indonesia

You built a simulation model and now you want to tune its parameters to data; here's an easy way to do it. A pared down example about this arXiv preprint

A cool article on the philosophy of simulation science is up on JASSS. I like the idea and the execution but I think the fear of “equifinality” is overstated.

I completed my dissertation defense on Friday. It went reasonably well. I was pleased I could embed the Dart version of the model straight into my slides. I am amazed that RMarkdown + ioslides + dart work so well together. The only thing I couldn’t do was to have the model learn on its own when a slide was called so that I always need to press “reset” before “start” when a slide containing the model shows up. Minor inconvenience but still.

Conservation notice: this is a simple model I cobbled up in about a week both as a spike to test the capabilities of google maps APIs and as a demo for a post-doc job. The code is open-source on MIT license. It was programmed in pure dart.

A very good book. To read. Detailed, learned and thorough. Also deeply flawed.

If you are like me and spent months looking for these, you are welcome. For everyone else, these are just common approximative models used in control theory to fit processes to. Unfortunately they are always expressed as Laplace transforms. S-space isn’t very useful to me, I really need processes in time. All credit to this stackexchange answer. An explanation of its math is here. Thank you, kind strangers.

I want to try and apply PID controllers to monopolistic competition. Partially because monopolistic competition is more interesting, but partially because I don’t like the way some ABMs model it.

Unit testing is awesome. Writing your assumptions in code so that they’ll always hold no matter what else changes, what’s there not to like? But the real question is, are extremely poorly written unit tests also awesome?