Choose timezone
Your profile timezone:
Powered by Indico
v3.3.5-pre
1) Roberto to chase Gulli Johannesson regarding CMA implementation and description.
2) Martin to chase Joeri Hermans regarding AVO
3) AMPGO: Roberto has implemented it, and will add it to the paper.
4) Active learning is being deprecated for this paper. We can remove it from the draft. Martin will do this after putting the draft on Overleaf.
5) Will has added a section on pyBAMBI and Multinest to the framework. PyBAMBI has yet to be interfaced with Bob's framework. Multinest is available through pyscannerbit.
6) Barry has sequential neural likelihoods working on his machine. Needs to implement it in Bob's framework. Having done that, Barry will attempt the same for calibrated approximated likelihood ratios.
7) Martin to chase Jakob on Metric Gaussian Variational Inference
8) MCMC: Csaba has nearly finished the MCMC section. Will put it in Overleaf when possible.
9) There is a massive collection of sampling algorithsm available in the Pagmo package at https://esa.github.io/pagmo2/. Pagmo vs Bob: we should try and parallelise algorithms in Bob's framework that are not currently parallelised. Roberto will attempt to write an interface between pagmo and Bob's framework, but we decided to stick with Bob's framework.
10) Zach will interface pyopt to Bob's framework to get Bayesian optimisation working.
11) Melissa to implement and document particle filter algorithm. Bob to let her know that this is still required.
12) Roberto reminds us that we were going to use GAMBIT likelihoods for testing. MSSM7 without nuisance parameters exists as a fast approximation. EWMSSM couldn't be interpolated with a deep neural network. Roberto is trying with a Gaussian process.
Timescales: Get code and algorithm descriptions done by Christmas. Get GAMBIT fast likelihood by Christmas.
Martin to propose list of tests for the different algorithms and send round document for discussion.
Martin and Roberto to push Judita for metrics by Christmas.