aesara 2.9.4
0
A library for defining, optimizing, and efficiently evaluating mathematical expressions involving mu
Contents
A library for defining, optimizing, and efficiently evaluating mathematical expressions involving multi-dimensional arrays.
Stars: 1179, Watchers: 1179, Forks: 155, Open Issues: 331The aesara-devs/aesara
repo was created 4 years ago and the last code push was 1 months ago.
The project is very popular with an impressive 1179 github stars!
How to Install aesara
You can install aesara using pip
pip install aesara
or add it to a project with poetry
poetry add aesara
Package Details
- Author
- None
- License
- BSD-3-Clause
- Homepage
- None
- PyPi:
- https://pypi.org/project/aesara/
- GitHub Repo:
- https://github.com/aesara-devs/aesara
Classifiers
- Scientific/Engineering/Mathematics
- Software Development/Code Generators
- Software Development/Compilers
Related Packages
Errors
A list of common aesara errors.
Code Examples
Here are some aesara
code examples and snippets.
GitHub Issues
The aesara package has 331 open issues on GitHub
- Constant folding overrides broadcastable pattern
- Update the high-level API documentation
ifelse
should be available in theaesara
namespace- Added Ops for np.linspace, np.logspace and np.geomspace
- Limit the use of
equal_computations
inScan
rewrites - Prevent
Scan.__eq__
from walking inner-graphs - Make sure we don't constant fold
Shape
s before shape inference tensor.zeros
fails with symbolic scalar values- Remove
mode
-based test runs - Remove
floatX
-based test runs - Reconsider
Alloc
's canonical form and rewrites - Implement a generalized
Op.[grad, R_op, L_op]
forCAReduce
- Implement Op for
linspace
and similar numpy functions - Make
Sparse
tensorType
s extendTensorType
- Remove
MaxAndArgmax