TensorKit.jl v0.17.1

A feature-and-performance patch release adding a matrix exponential (exponential/exponential!), broader Enzyme AD support, and several trivial-symmetry performance improvements.

Highlights

• Matrix exponential: new exponential and exponential! for tensors (Exponential #465)
• Enzyme AD: forward and reverse rules for TensorOperations contractions, linear algebra, and VectorInterface, plus reverse rules for index manipulations (Forward AD tests and cleanup for vector interface #436, Forward rules for TensorOperations calls #437, Add Enzyme forward and reverse rules and tests for VectorInterface #440, Forward and reverse Enzyme tests and rules for linalg #449, Forward and reverse Enzyme rules and tests for TensorOperations #451, Enzyme reverse mode rules and test for index manipulations #466)
• Performance: reduced overhead in the TensorMap constructor and trivial-symmetry code paths (Bypass some overhead in TensorMap constructor, bypass overhead in sectorequal/sectorhash for trivial symmetry #476, perf: Trivial symmetry performance improvements #479, Trivial tensors fast path into TensorOperations machinery #463, Mark more error strings lazy #478)
• GPU: consolidated GPU logic into a GPUArrays extension and dropped the explicit cuTENSOR dependency. Importing cuTENSOR is still recommended for best performance, as it enables the corresponding TensorOperations extension (Consolidate duplicated GPU logic into new GPUArrays extension #460, Remove explicit dependence on cuTENSOR for CUDA ext #455)
• Docs: added docstrings for catdomain and catcodomain (docs: Add docstrings for catdomain and catcodomain #485)

Full Changelog

See CHANGELOG for the complete list of changes.