Skip to content

Physics-Based Model

Ideas

  1. Refactor Physics Model
  2. Get Toy Datasets
  3. Implement Different samplers
  4. Do generic visualizations

Samplers

We want to work with some of the generic samplers as well as some of the SOTA methods. We also want to use things out of the box and we don't want to reinvent algorithms.

  • Affine-Variant Sampling Ensemble Sampler - emcee
  • Nested Sampling - dynesty
  • NUTS - pymc3
  • SMC - pymc3

We can also integrate some of this code based on some code by Dan! We can see an example here where we integrate emcee with pymc3. If the formatting is the same, we should also be able to use the nested sampler from dynesty.

Visualization

We want to do generic visualizations. Meaning we should be able to use the same functions for other models. And it will also be using a common platform. In this case, it will be arviz.