pymc

pymc copied to clipboard

Refactor `get_tau_sigma` and support lists of variables

2

**What is this PR about?** This PR refactors the `get_tau_sigma` logic in order to support lists of `Variable`s and fixes https://github.com/pymc-devs/pymc/issues/6987. N.B. The original code had a positivity check for...

tvwenger

Allow Truncation of CustomDist

1

This is now possible: ```python import pymc as pm import numpy as np def maxwell_dist(scale, size): return pm.math.sqrt(pm.ChiSquared.dist(nu=3, size=size)) * scale scale = 5.0 x = pm.CustomDist.dist(scale, dist=maxwell_dist) trunc_x =...

ricardoV94

Add KL Divergence helper

11

## Description ## Related Issue - [ ] Closes issue: # - [ ] Related issue (not closed by this PR): # ## Checklist - [ ] Checked that [the...

ferrine

Use same API for defining internal and external Nuts kwargs

2

### Description User on discourse [reported](https://discourse.pymc.io/t/how-can-i-set-the-maximum-tree-depth-for-the-nuts-method-from-the-numpyro-library/12258): > > How can I set the maximum tree depth for the NUTS method from the numpyro library? > The way described in the...

ricardoV94

Implement logprob inference for binary operations

4

### Description Using CDFs it should be simple to derive the logp of graphs of the form: ```python import pymc as pm x = pm.Normal.dist() y = x < 0.5...

ricardoV94

Remove `givens_dict` argument in sample posterior predictive internal function

21

https://github.com/pymc-devs/pymc/blob/473c9528e56fc291e323e30082c8f4295dbd6149/pymc/sampling/forward.py#L618 If I pass variables to cmopile forward function, would it be a correct do-inference?

ferrine

### Description These are needed for many methods like `MargnalModel.recover_marginals`. And debugging utilities like https://github.com/pymc-devs/pymc-experimental/issues/217 Alleviates the need for #6721

ricardoV94

BUG: LKJCorr breaks when used as covariance with MvNormal

10

### Describe the issue: LKJCorr backwards sampling is currently broken and does not properly constrain values to a positive semidefinite matrix early enough. This leads to compilation randomly failing when...

velochy

ENH: Measurable `dot` and `matmul`

4

### Before ```python def scaled_mv_normal(mu, cov, R): return pm.MvNormal.dist(mu=R @ mu, cov= R @ cov @ R.T) ``` ### After ```python def scaled_mv_normal(mu, cov, R): return R @ pm.MvNormal.dist(mu=mu, cov=cov)...

jessegrabowski

Add icdf functions for Beta, Gamma, Chisquared and StudentT distributions

24

**What is this PR about?** Adds ICDF functions to Beta, Gamma, Chisquared and StudentT distributions. Issue https://github.com/pymc-devs/pymc/issues/6612 This is a work in progress. The quantile functions depend on two main...

amyoshino