pymc-devs / pytensor

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

https://pytensor.readthedocs.io

Repository from Github https://github.compymc-devs/pytensor

pymc-devs/pytensor Issues

BUG: Compatibility Issue for pytensor on macOS 15.4
Updated 5 months ago13
Consider using fft for convolution to reduce cost of multiple convolutions with shared inputs
Updated 5 months ago2
User facing split does not match numpy
Updated 5 months ago
Sparse `MulSD` Op gradient raises when dense is scalar
Updated 5 months ago
BUG: `convolve1d` `mode="same"` indexing error
Closed 6 months ago1
Numpy 2.x C implementation for AdvancedIncSubtensor1
Updated 6 months ago1
Do not allow passing None for arbitrary arguments in Function
Updated 6 months ago
Nans in JAX multinomial dispatch
Closed 6 months ago3
Add Ops for specialized dot products with structured matrices
Updated 6 months ago3
Add correlate helper function
Updated 6 months ago
Convolve1d rewrites
Updated 6 months ago
`get_vector_length` incorrectly returns for shared variable without static shape
Closed 6 months ago3
`pt.arange(int)` has `.type.shape` set to `None`
Closed 6 months ago1
Add helper for `special.kn` that just calls `special.kv` under the hood
Closed 6 months ago2
Add numba overload for `Nonzero`
Closed 6 months ago
Implement generalized rewrite to replace ops that compute partial outputs by ops with all the outputs
Updated 6 months ago
Numba lapack routines should use int64 for shapes
Updated 6 months ago
Support new `jax.random.key` classes
Updated 6 months ago
Mitigate the PyPI API token
Closed 6 months ago2
BUG: Latest PyTensor versions not available in conda?
Updated 6 months ago1
`linspace` and `retstep=True` fails
Updated 6 months ago1
Numba symmetrical solve prints spurious ill-conditioned matrix
Closed 6 months ago
Replace 0 sized nodes by zeros
Updated 6 months ago
Rewrite nested inc/set_subtensor on zeros
Updated 6 months ago
Add option not to print inner graphs in debug_print
Updated 6 months ago
atleast_Nd should only accept one positional argument
Closed 6 months ago
Scan save mem rewrite masks issues with `steps=0`
Updated 6 months ago
Scan NIT-SOT with 0 steps have wrong shape
Updated 6 months ago
Cleanup root objects
Updated 6 months ago
Inplace on sit-sot / mit-sot when nsteps is symbolic
Updated 6 months ago
Allow inner-function inplace of mit-mot in Scan backend
Updated 6 months ago
Expose `vecdot`, `vecmat` and `matvec` helpers
Closed 6 months ago
Remove `enable_slice_literals` from `numba.dispatch.basic`
Closed 6 months ago
Replace our `Tri` op with an `OpFromGraph`
Updated 6 months ago1
Find an alternative to MKL for OSX on Intel
Updated 6 months ago2
Add `C` and `Python` modes
Updated 6 months ago
CGemv fails with strides0 x
Updated 6 months ago
Add `transposed` argument to `pt.linalg.solve`
Closed 6 months ago
Transform graph to make large constants symbolic inputs
Updated 7 months ago1
Implement `infer_shape` automatically from `gufunc_signature`
Updated 7 months ago
Run doctests on the docs
Updated 7 months ago
Deprecate `broadcast_shape` in favor of `broadcast_shapes`
Updated 7 months ago1
Consider specialized DimShuffle C-implementation
Updated 7 months ago
Numba RandomVariables should fallback to object mode
Updated 7 months ago
Useless `dimshuffle("x","x")`
Updated 7 months ago
Numba `solve_triangular` returns wrong results when `overwrite_b=True`
Closed 7 months ago
`solve_triangular` gradients are wrong for certain options
Updated 7 months ago
Allow jacobian and hessian of arbitrary dimension expressions
Updated 7 months ago
Port SymbolicRandomVariable Op to PyTensor and get rid of many RVs in PyTensor
Updated 7 months ago
Stop running numba tests in object mode for coverage
Updated 7 months ago