pytensor
pytensor copied to clipboard
pymc-devs
PyTensor allows you to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays.
### Description When we have a JAXOp in the final graph in a non-jax backend we may want to manipulate the JAX Op for efficiency. We could rewrite `Blockwise(JAXOp)` ->...
This PR removes all occurrences of the `micromamba-version: "1.5.10-0"` specification from GitHub workflow files, as the underlying issue that necessitated this pin has been resolved. ## Background The micromamba version...
**Disclaimer: this is still 100% on hack status, and I don't understand half of the things I did** When we tried #811 it was obvious that numba compile times were...
### Context for the issue: Add an example of how Pytensor backend switching compares to a pure implementation in that backend.
Not ready for review!, includes commits that have to be removed or from other PRs
### Description A quick simple implementation seems to be 8x faster, avoiding the whole multiple-dispatch approach. Also it should give us more flexibility to unify with variadic/commutative patterns as explored...
### Description We're seeing BLAS warnings in the windows CI for `pymc-extras`. There are two. [One is the usual linker failure](https://github.com/pymc-devs/pymc-extras/actions/runs/18078168486/job/51437649154?pr=584): CI log snippet ``` 2025-09-28T18:44:42.2527549Z [1m[31mERROR [0m pytensor.graph.rewriting.basic:basic.py:1757 Rewrite...
Fixes pymc-devs/pytensor#541 ## Problem `AdvancedSubtensor` currently takes `SliceType` and `NoneType` variables as inputs, which prevents using `Blockwise/vectorize` trivially in cases that would otherwise be valid: ```python import pytensor import pytensor.tensor...
# Named Dimensions Refactor: Objects Instead of Strings I'm still working on this, but thought it might be helpful to share what I have so far... ## The Key Change...
### Description There are some complex code examples added in #1614, but they don't run as part of the doctests because we explicitly ignore `pytensor.link` module (I don't remember if...