pints
pints copied to clipboard
pints-team
Probabilistic Inference on Noisy Time Series
Not sure why. Seems to be the initialisation rather than the actual iterations? Or maybe there's lots of Nones returned before 5 full iters are reached? Either way, would be...
Maybe we could have slightly better docs for this page: https://pints.readthedocs.io/en/latest/diagnostics.html#mcmc-diagnostics? For example what are the expected inputs and outputs. It took me a few minutes to figure it out,...
In the Hamiltonian MCMC [notebook](https://github.com/pints-team/pints/blob/master/examples/sampling/hamiltonian-mcmc.ipynb) is at the very bottom the number of divergent transitions displayed. In the notebook a number of examples are explored, each of which define a...
I know we'd run out of symbols at some point, but the `sigma0`s in [MCMCController](https://pints.readthedocs.io/en/latest/mcmc_samplers/running.html#pints.MCMCController) and [OptimisationController](https://pints.readthedocs.io/en/latest/optimisers/running.html#pints.OptimisationController) mean subtly different things, one is covariance matrix $\Sigma$ and the other is...
Something we haven't done much in PINTS, is adding multiple _classes_ in a test module (file), and putting the stuff required to set up the data for each test in...
- docs state it returns `(fig, ax)` when in fact it returns `(fig, array-of-axes)` - some info about the interpretation should be given in notebook: https://github.com/pints-team/pints/blob/master/examples/plotting/mcmc-autocorrelation.ipynb - notebook ends in...
"Since 1.96 is the 0.975 quantile of the standard normal distribution" Should we include these in the plot by default?
The Hierarchy of methods in docs: https://pints.readthedocs.io/en/latest/#hierarchy-of-methods needs updating. E.g. It's missing the Haario ones (under the Adaptive methods).
The autograd notebook stopped worked when numpy 2 came out #1530 But now their github says the project is no longer maintained, and people should use JAX instead: https://github.com/HIPS/autograd So...
We could add it to the `extras` list of dependencies (only requires numpy/scipy/matplotlib itself) https://github.com/dfm/corner.py