pymc-devs/pytensor

View on GitHub

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

Stars: 644
Forks: 209
Open beginner issues: 14
Indexed issues: 341
Avg merge: 2d 14h
Merged PRs (30d): 16

Dominant language: Python
License: No license data
Last GitHub push: Sep 18, 2026
Latest indexed: Sep 19, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: help wanted

Beginner friendly Everything

Open All

Numba linker: fgraph_to_python puts fgraph inputs in global_env, preventing numba caching Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

pymc-devs/pytensor#2428 · 3 comments · Opened Sep 18, 2026
Feature request: backend-dispatch registration hook for third-party packages Open

Difficulty 5/5 Over a week Newbie friendliness 45/100

pymc-devs/pytensor#2427 · 5 comments · Opened Sep 18, 2026
Make Sort/ArgSort axis a static Op property Open

enhancement refactor

Difficulty 4/5 3-5 days Newbie friendliness 55/100

pymc-devs/pytensor#2426 · Opened Sep 18, 2026
MLX: Subtensor and IncSubtensor fail on a symbolic integer index under mx.compile Open

bug indexing mlx

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

pymc-devs/pytensor#2422 · Opened Sep 17, 2026
Numba MvNormalRV raises LinAlgError on a NaN covariance: np.linalg calls inside the njit body Open

enhancement linalg numba

Difficulty 3/5 1-2 days Newbie friendliness 84/100

pymc-devs/pytensor#2421 · Opened Sep 16, 2026
Blockwise of an OpFromGraph core keeps the inner graph's unknown static shape Open

OpFromGraph shape inference vectorization

Difficulty 3/5 1-2 days Newbie friendliness 68/100

pymc-devs/pytensor#2419 · 6 comments · Opened Sep 15, 2026
local_useless_unbatched_blockwise emits squeeze(expand_dims(x)) that nothing collapses, so factorizations stop merging Open

graph rewriting performance vectorization

Difficulty 3/5 1-2 days Newbie friendliness 68/100

pymc-devs/pytensor#2416 · 2 comments · Opened Sep 14, 2026
psd_solve_to_chol_solve only tracks b_ndim=2, so solve(A, b, assume_a="pos") with a vector b factors A twice Open

graph rewriting linalg performance

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

pymc-devs/pytensor#2415 · Opened Sep 14, 2026
BUG: ImportError: cannot import name 'batched_dot' from 'pytensor.tensor.blas' Open

bug

Difficulty 3/5 1-2 days Newbie friendliness 48/100

pymc-devs/pytensor#2407 · 3 comments · Opened Sep 7, 2026
Move the _is_provably_positive helpers out of tensor/subtensor.py Open

maintenance refactor

Difficulty 4/5 3-5 days Newbie friendliness 50/100

pymc-devs/pytensor#2404 · Opened Aug 28, 2026
test_mlx_eigh compares eigenvector signs, which are arbitrary Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

pymc-devs/pytensor#2394 · Opened Aug 26, 2026
MLX: gradient of batched advanced indexing fails in broadcast_shapes Open

bug gradients indexing mlx

Difficulty 3/5 1-2 days Newbie friendliness 68/100

pymc-devs/pytensor#2387 · 1 comment · Opened Aug 26, 2026
Consider `mlx-addons` for MLX linalg dispatches Open

enhancement help wanted linalg mlx

Difficulty 5/5 Over a week Newbie friendliness 30/100

pymc-devs/pytensor#2383 · 4 comments · 1 reaction · Opened Aug 26, 2026
MLX AdvancedIncSubtensor fails when Blockwise adds a batch dim Open

backend compatibility bug mlx

Difficulty 3/5 1-2 days Newbie friendliness 70/100

pymc-devs/pytensor#2382 · Opened Aug 25, 2026
ENH: Expose a 'stop at named variables' mode in `pprint` and add LaTeX printers Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

pymc-devs/pytensor#2381 · 2 comments · Opened Aug 25, 2026
numba max reduction is 10x slower than sum on the same array Open

numba performance

Difficulty 4/5 3-5 days Newbie friendliness 50/100

pymc-devs/pytensor#2367 · 2 comments · Opened Aug 20, 2026
pt.flatten drops static shape on a fully-known input Open

bug shape inference

Difficulty 3/5 1-2 days Newbie friendliness 68/100

pymc-devs/pytensor#2366 · 3 comments · Opened Aug 20, 2026
PyTorch backend warns on every read-only constant: torch.as_tensor on non-writable array Open

bug torch

Difficulty 2/5 1-3 hours Newbie friendliness 72/100

pymc-devs/pytensor#2365 · 1 comment · Opened Aug 20, 2026
scan_push_out_seq doesn't push a sequence-dependent Blockwise out of the loop Open

enhancement graph rewriting performance scan

Difficulty 3/5 1-2 days Newbie friendliness 58/100

pymc-devs/pytensor#2357 · 1 comment · Opened Aug 16, 2026
Prune axis invariant offsets in (log)softmax Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

pymc-devs/pytensor#2344 · Opened Aug 12, 2026
pt.tensor("x") without shape= gives misleading error — default implies optional when it isn't Open

Difficulty 2/5 1-3 hours Newbie friendliness 68/100

pymc-devs/pytensor#2338 · 1 comment · Opened Aug 10, 2026
TensorType.shape is a mutable plain attribute Open

Difficulty 2/5 1-3 hours Newbie friendliness 76/100

pymc-devs/pytensor#2330 · 3 comments · Opened Aug 9, 2026
XTensorType.dims and .shape are silently mutable, leaving ndim out of sync Open

Difficulty 2/5 1-3 hours Newbie friendliness 72/100

pymc-devs/pytensor#2329 · 1 comment · Opened Aug 9, 2026
BUG: matrix multiplication causes compiler error on Windows Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 45/100

pymc-devs/pytensor#2321 · 3 comments · Opened Aug 6, 2026
JAX linker: Cannot compile Convolve1d without static mode Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2317 · 1 comment · Opened Jul 31, 2026
MLX linker: Composite op fails with mixed-shape inputs Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

pymc-devs/pytensor#2316 · 1 comment · Opened Jul 31, 2026
Add specialized float32 numba dispatches for transcendental functions Open

numba performance request discussion

Difficulty 5/5 Over a week Newbie friendliness 48/100

pymc-devs/pytensor#2314 · 2 comments · Opened Jul 30, 2026
C implementation of `gammaincc` silently loses precision for large `a ≈ x` Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

pymc-devs/pytensor#2291 · Opened Jul 13, 2026
BUG: Clang linker fails on macOS 27 Beta ARM64 — `ld: library 'd64' not found` Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 58/100

pymc-devs/pytensor#2268 · 1 comment · 1 reaction · Opened Jun 30, 2026
How to handle `monotonically_increasing/decreasing` attributes on non-unary OPs Open

pymc-devs/pytensor#2267 · 2 comments · 1 assignee · Opened Jun 29, 2026
BUG: NaN gradient (VJP) from pt.log(pt.sigmoid(x)) for large x in JAX backend Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 52/100

pymc-devs/pytensor#2264 · 3 comments · Opened Jun 28, 2026
DOC: Add pre-commit commands to CONTRIBUTING.md Open

Difficulty 1/5 Under an hour Newbie friendliness 86/100

pymc-devs/pytensor#2215 · 1 comment · Opened Jun 11, 2026
PERF: Enable BlasOpt and dispatch gemm/gemv/ger to fused MLX kernels Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

pymc-devs/pytensor#2213 · 2 comments · Opened Jun 11, 2026
Add first-class python backend Open

backend compatibility help wanted

Difficulty 5/5 Over a week Newbie friendliness 35/100

pymc-devs/pytensor#2193 · 3 comments · Opened Jun 1, 2026
An Elemwise feeding `dest[idx].set(...)` / `.inc(...)` isn't fused into the write (extra temp + copy) Open

enhancement graph rewriting memory opt numba performance

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2192 · Opened Jun 1, 2026
Gradient of pt.minimum is wrong in numba mode due to fast_math assoc flag Open

gradients numba request discussion

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2187 · 10 comments · Opened May 28, 2026
MetaType `__props__` based equality overrides class custom implementations Open

Op implementation

Difficulty 5/5 Over a week Newbie friendliness 35/100

pymc-devs/pytensor#2185 · Opened May 28, 2026
Module-level Op instances break LSP signatures, Sphinx docs, and source links for expm, det, inv, lstsq Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

pymc-devs/pytensor#2184 · 1 comment · Opened May 28, 2026
DOC: include description for the section with vectorize Open

docs

Difficulty 1/5 Under an hour Newbie friendliness 68/100

pymc-devs/pytensor#2183 · 1 comment · Opened May 28, 2026
Don't allow AdvancedSubtensor and AdvancedIncSubtensor non-consecutive advanced indices Open

indexing Op implementation

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2166 · Opened May 24, 2026
support for `__len__` Open

enhancement

Difficulty 3/5 1-2 days Newbie friendliness 50/100

pymc-devs/pytensor#2151 · 1 comment · Opened May 20, 2026
Canonicalize associative Op argument order and add distributive factorization for Add/Sub Open

graph rewriting performance

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2140 · Opened May 16, 2026
Lift Elemwise through SetSubtensor Open

graph rewriting

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2138 · Opened May 15, 2026
Investigate infer_shape database Open

graph rewriting needs info Op implementation

Difficulty 5/5 Over a week Newbie friendliness 35/100

pymc-devs/pytensor#2137 · Opened May 15, 2026
Useless integer casts in Alloc/Reshape shape arguments are not canonicalized away Open

beginner friendly graph rewriting

Difficulty 3/5 1-2 days Newbie friendliness 74/100

pymc-devs/pytensor#2135 · Opened May 14, 2026
Add CAReduce(DimShuffle(x)) -> CAReduce(x) rewrite when DimShuffle is a no-op for the reduction Open

graph rewriting

Difficulty 3/5 1-2 days Newbie friendliness 45/100

pymc-devs/pytensor#2132 · 1 comment · Opened May 13, 2026
DOC: add citation files Open

docs

Difficulty 2/5 1-3 hours Newbie friendliness 65/100

pymc-devs/pytensor#2119 · 1 comment · 1 reaction · Opened May 9, 2026
Keep jit_fn through pickle Open

performance

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pymc-devs/pytensor#2108 · Opened May 1, 2026
ENHANCEMENT: Scan has no MLX dispatcher — stateful recurrent logps fail to compile Open

enhancement mlx

Difficulty 5/5 Over a week Newbie friendliness 45/100

pymc-devs/pytensor#2097 · Opened Apr 28, 2026
PERF: minor MLX dispatch optimizations (Cast, DimShuffle, Argmax) Open

enhancement mlx

Difficulty 4/5 3-5 days Newbie friendliness 58/100

pymc-devs/pytensor#2094 · Opened Apr 28, 2026

Showing the newest 100

This page lists what was indexed most recently. Advanced filter has the whole inventory, narrowed by language, difficulty and how long a task takes.

Open advanced filter

pymc-devs/pytensor

Showing the newest 100

Get new issues in your inbox