todo.tasks

TODO:
    ☐ Posterior samples @high

Test:
    ☐ ESS: retry until sample succeeds

Features:
    ☐ Predictive joint log-pdf

＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿
Archive:
 ✓ Split everything up in `run.py` and `process.py` @done (22-04-14 13:48) @project(Refactor experiments)
 ✓ Make `reproduce.sh` which runs all @done (22-04-14 13:48) @project(Refactor experiments)
 ✓ Add vix and kemar data to WBML @done (22-04-14 13:48) @project(Refactor experiments)
 x Simple estimates of ESS @cancelled (20-02-04 15:09) @project(TODO / MCMC diagnostic plots)
 x Plot traces of dimensions @cancelled (20-02-04 15:09) @project(TODO / MCMC diagnostic plots)
 x Improve mixing of MCMC by incorporating more into the prior in ESS @cancelled (20-02-04 15:09) @project(TODO)
 ✓ --model flag option @done (20-02-02 17:54) @project(TODO)
 ✓ --fix-noise flag option @done (20-02-02 17:20) @project(TODO)
 ✓ Save samples @critical @done (20-02-01 22:07) @project(TODO)
 ✓ --quick flag option @done (20-02-01 21:38) @project(TODO)
 ✓ Take samples at random positions in summarise_samples @done (20-02-01 21:24) @project(TODO)
 ✓ Separate printing ELBO and variables from training @done (20-02-01 21:24) @project(TODO)
 ✓ Strip out mean-field approximation @done (20-02-01 21:16) @project(TODO)
 ✓ Reuse samplers and samples whenever possible @done (20-02-01 20:30) @project(TODO)
 ✓ Reduce amount of output @done (20-02-01 20:30) @project(TODO)
 ✓ Estimate ELBO value from many samples in SMF approximation @done (20-02-01 20:29) @project(TODO)
 ✓ Can omega now be trained? @done (20-01-24 21:40) @project(TODO)
 ✓ Thoroughly test `expq`. @done (20-01-17 21:54) @project(TODO)
 ✓ Double check argument names. @done (20-01-17 21:54) @project(TODO)
 ✓ Port over tests from GP-RV. @done (20-01-11 20:27) @project(TODO)