Jesse Grabowski

Iskrev (2009) describes a method of testing local parameter identification using empirical autocorrelation matrices [here](https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=451ae8fb8a80d1a8e686c2a773ab4cb51534c063). This would be a nice feature to add.

Exploring VI implementations

4

Checking out how VI works with a simple from-scratch implemention ---- 📚 Documentation preview 📚: https://pymc--7799.org.readthedocs.build/en/7799/

Add `model_to_minibatch` transformation to convert all `pm.Data` to `pm.Minibatch`

9

## Description A pain point for me when testing different algorithms (e.g. MCMC vs VI) is that I don't want to write a 2nd version of the model with `pm.Minibatch`...

### Description This is an oft-requested distribution, and devs have always said no because it's not useful at all in the context of MCMC sampling (see [here](https://discourse.pymc.io/t/the-inverse-wishart-prior-for-covariance-matrix/1318/3), also #4606, #975,...

Replace `assert` checks in `logprob.scan` with real error handling

1

### Description There are two annoying assert checks that I commonly run up against when writing time series models [here](https://github.com/pymc-devs/pymc/blob/ef26ae88e87c2120c2700d062e404ed1f777d358/pymc/logprob/scan.py#L115) and [here](https://github.com/pymc-devs/pymc/blob/ef26ae88e87c2120c2700d062e404ed1f777d358/pymc/logprob/scan.py#L116). Other users are hitting them too, see for...

Compute Metropolis average accept probability with `logsumexp`

5

In the Metropolis step stats that we report on the progressbar, we have a probability of acceptance. This is currently computed in a very direct way, just `np.mean(np.exp(log_accept_prob))`. We can...

Fix progress bar error when nested `CompoundStep` samplers are assigned

9

#7721 reports an error in the presence of nested `CompoundStep`. Here's a prettier version of what pymc gives for the example in that issue: ``` CompoundStep ├─CompoundStep │ ├─ Metropolis:...

Add `rng_fn` to CAR/ICAR

5

### Description I was talking to @theorashid who linked me to [this case study](https://mc-stan.org/learn-stan/case-studies/icar_stan.html) of CAR priors. It seems like they're just MvNormals, but with degenerate covariance matrices. We can...

Logp inference fails for exponentiated distributions

7

### Description Suppose I want to make a log-normal "by hand": ```py def d(mu, sigma, size=None): return pt.exp(pm.Normal.dist(mu=mu, sigma=sigma, size=size)) with pm.Model() as m: y = pm.CustomDist('y', 0, 1, dist=d,...

New CholeskyCorr transform

5

## Description For drawing from an LKJCholeskyCorr distribution. Seems nice, but needs a forward transform. Marked as maintenance because our current transformation breaks for n > about 5. ## Related...