pymc-devs / pymc-devs/pytensor

Useless `dimshuffle("x","x")`

Open
#1,244 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

graph rewriting
Dominant language
Python
Stars
644
Forks
208
Avg merge
2d 14h
Merged PRs (30d)
16

Description

Description
import pytensor
import pytensor.tensor as pt

x = pt.tensor("x", shape=(None, None))
y = x.dimshuffle("x", "x")
pytensor.function([x], y).dprint()
# DeepCopyOp [id A] 1
#  └─ DimShuffle{order=[x,x]} [id B] 0
#     └─ x [id C]

Whenever a DimShuffle only has x and it has as many entries as the inputs, it's basically a specify_shape(x, (1,) * x.type.ndim). If the dimensions were known to be length1, the DimShuffle would be removed correctly

This is a low priority one since the DimShuffle is little more expensive than the specify_shape

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by running the Python reproducer in the issue and inspecting the DimShuffle optimization entry point. Trace the case where the order contains only "x" and matches the input dimensionality. Done means the graph uses the specified-shape form instead of retaining the redundant DimShuffle, with the existing known-length-one behavior preserved.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
compilers, performance
Issue type
Feature
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.