pytensor icon indicating copy to clipboard operation
pytensor copied to clipboard

PyTensor allows you to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays.

Results 395 pytensor issues
Sort by recently updated
recently updated
newest added

### Description If the number of repetitions in `pt.repeat` is homogeneous, we could replace it by `repeat(x, unique_value)`. Under the hood this ends up as an alloc, skipping the non-C...

graph rewriting
performance

### Description Users should be able to set ityes and otypes for as_jax_op/wrap_jax directly if they want to avoid automatic inference, just as in as_op.

### Description ScalarOps are parametrized by an `output_types_preference` that determines the output types from the input types in `make_node` (unless an Op overrides `make_node`). This is used indirectly via `output_types`....

docs
Op implementation

## Description Adds pt.pack and pt.unpack helpers, roughly conforming to the `einops` functions of the same name. These helps are for situations where we have a ragged list of inputs...

enhancement

### Description I tried it in this commit: https://github.com/pymc-devs/pytensor/pull/1592/commits/07e50e20e4ae952a7aba7b68cf1a97373fa3eba5, some tests failed which need to be checked. Nothing crazy should be needed, maybe I did a small error in the...

beginner friendly
graph rewriting

### Description Scan and Composite/ScalarLoop are our only ops with inner graph that support equality /merging. The way equality / hashing is done it's to use the generated c_code or...

enhancement
Op implementation
scan
OpFromGraph

### Description We take small performance hits by using `scipy.linalg` functions rather than directly calling lapack functions in our `perform` methods. These wrappers include a lot of data validation checks...

help wanted
beginner friendly
performance
linalg

### Description OPs are supposed to be immutable/ hashable. The hash of the Scan Op is defined based on the graph with which it is created originally. However the equality...

bug
scan

### Description I'm investing implementing ALS in pytensor which is usually implemented with sparsity constructs (see [implicit](https://benfred.github.io/implicit/api/models/cpu/als.html) for reference). I quickly looked around and saw this [older thread](https://discourse.pymc.io/t/sparse-linear-system-solving-inside-pymc/14963/2) where someone...

enhancement
gradients
linalg

### Describe the issue: Due to changes in https://github.com/pymc-devs/pytensor/pull/1582, as part of https://github.com/pymc-devs/pytensor/issues/1550, the following codeblock(s) in https://pytensor.readthedocs.io/en/latest/gallery/optimize/root.html fail: ### Reproducable code example: ```python import pytensor import pytensor.tensor as pt...

bug