pytensor 2.8.10

Optimizing compiler for evaluating mathematical expressions on CPUs and GPUs.

PyTensor is a Python library that allows you to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays. It is built on top of NumPy. PyTensor features:

• tight integration with NumPy: a similar interface to NumPy’s. numpy.ndarrays are also used internally in PyTensor-compiled functions.

• efficient symbolic differentiation: PyTensor can compute derivatives for functions of one or many inputs.

• speed and stability optimizations: avoid nasty bugs when computing expressions such as log(1 + exp(x)) for large values of x.

• dynamic C code generation: evaluate expressions faster.

• extensive unit-testing and self-verification: includes tools for detecting and diagnosing bugs and/or potential problems.