v0.16.1

[lkdvos]

Requires #364 first.

Release notes:

A small patch release with bug fixes, performance improvements, and early GPU/extension support.

Highlights

• Initial CUDA support for factorizations and preparatory work for CUDA-backed workflows (#336, #325).
• GPU-friendly truncation implementations and norm optimizations for large tensors (#349, #351).
• Better preservation of tensor storagetype when converting between TensorMap representations (#357).

Notable changes

Added

• CUDA support groundwork for factorizations and GPU extensions (#336, #325).
• Mooncake and Adapt extension integration scaffolding (#352, #344).
• Exported TimeReversed symbol for convenience (#337).

Changed

• convert(TensorMap, t) retains storagetype during conversions (#357).
• transpose specialization for DiagonalTensorMap to improve correctness/performance (#335).
• Unified CartesianSpace and ComplexSpace constructors (#334).
• TensorOperations AD cleanup and test reorganization (#343, #339).

Fixed

• Correct handling of real tensors with complex scalartype (#360).
• Stable ordering for diagonal eigenvalues (#350).
• Divide-by-zero safety in show for empty tensors and ensure svd_vals(::DiagonalTensorMap) returns SectorVector (#329, #333).
• Various type-stability and small test fixes (several commits).
• Adding tensors of different types now correctly promotes (#364)

Performance

• GPU-friendly truncation routines reduce host/GPU overhead for large truncations (#349).
• norm improvements reduce allocations and speed common cases (#351).

Upgrade notes

• This is a patch release; the public API remains compatible with v0.16.0.
• CUDA support is initial. To use CUDA features, enable the CUDA/cuTENSOR stack and the optional extensions (see Project.toml extras).

Contributors

Thanks to: Lukas Devos, Katharine Hyatt, Boris De Vos, Yue Zhengyuan, and others for contributions and reviews.