pymc
pymc copied to clipboard
pymc-devs
Bayesian Modeling and Probabilistic Programming in Python
## Description of your problem The initial point evaluation, primarily in `init_nuts` is the one and only place in the `pm.sample` control flow when we want to add jitter. That's...
This was added to facilitate sampling from the prior as a starting point for SMC, but with the new functions added in #4983 this is no longer needed. https://github.com/pymc-devs/pymc/blob/6d2aa5ddebed01d81c2ab66b9d4bd02194f82508/pymc3/sampling.py#L1983-L1999
As we did with test/initial values we should remove the value variable transformation from the `.tag` if possible. Probably saving it in a `model.value_transforms` dictionary that can be used directly...
## Description of your problem The `MarginalSparse` Gaussian Process uses `DensityDist` internally to compute the `marginal_likelihood` ([here](https://github.com/pymc-devs/pymc3/blob/main/pymc/gp/gp.py#L733-L756)). We should refactor that code to use a proper RandomVariable instead ## Versions...
```python import pymc3 as pm import numpy as np returns = pd.read_csv(pm.get_data("SP500.csv"), index_col="Date") returns["change"] = np.log(returns["Close"]).diff() returns = returns.dropna().astype('float32') returns.head() with pm.Model() as model_pymc3: step_size = pm.Exponential("step_size", 10.) volatility =...
## Description of your problem **Please provide a minimal, self-contained, and reproducible example.** See implementation here https://github.com/pymc-devs/pymc3/blob/c1efb7ad51e34e6bd8b292ef226a7e191adbdb82/pymc3/theanof.py#L92 for float32 precision, it converts values to int16 for float64 precision, it converts...
In anticipation of the big v4 release we should start to write a document walking people through the upgrade process. Specifically, this document should contain: * Information about the Theano->Aesara...
`model.logpt`, `model.dlogpt` and `model.d2logpt` would be a bit more intuitive without the `t` suffix. Historically that was done to distinguish the `tensor` graphs from the compiled functions, but those are...
Would be nice to add code-block examples that show how to use frequent `pm.math` functions with simple PyMC models in the docstrings: * switch * stack * softmax * ...