pymc
pymc copied to clipboard
pymc-devs
Bayesian Modeling and Probabilistic Programming in Python
It's probably because they are not pure RandomVariables. We need to update the `model_graph` logic to accommodate them
In PyMC discourse, I asked this [this](https://discourse.pymc.io/t/value-error-in-chains-for-a-gaussian-process/9894), and I reproduce here my code for convenience: ```python X = np.array([32, 38, 94, 83, 99, 78]) y = np.array([1702, 1514, 683, 269,...
## Description of your problem Getting an exception that I can't pass both `samples=500` and `keep_size=True` cant both be provided, but as a user I only pass samples, and then...
**Please provide a minimal, self-contained, and reproducible example.** ```python # %% import pymc as pm import numpy as np from functools import partial from pymc import sampling_jax import pytest rstate...
## Description of your problem PyMC sideliner here, who is finally taking a proper crack at it with 4.0. I find the distribution docstrings difficult to parse when viewing them...
Closes #5762. This PR removes `GaussianRandomWalkRV` altogether by defining an `rv_op` as a cumulative sum of distributions. The most recent version of AePPL, i.e. 0.0.31, is now able to retrieve...
Addressing #5659 Kept the original dims in the log_jac_det by changing ``` def log_jac_det(self, value, *inputs): return at.sum(value[..., 1:], axis=-1) ``` to ``` def log_jac_det(self, value, *inputs): return at.sum(value[..., 1:],...
@canyon289 to resolve #5669, I have made some small changes mainly to aesaraf.py to include the dim names in the new_size tensor (https://github.com/pymc-devs/pymc/commit/6585f9021a17a7f0c9c833bf4700fab240d2164b) This solves the issue mainly and `test_model`...
addresses #5791 - [x] Inferring dims from named pandas indexes - [x] Inferring dims from DataArrays - [x] Inferring coords from DataArrays - [ ] A unit test - [...
This tries to address #4807 Right now, it passes all unit tests, but doesn't use the fast sampling procedure of `sample_posterior_predictive` The code is also not feature-complete.
