[lkdvos]

TensorOperations.jl v5.8.3

This release substantially reduces ncon overhead and fixes several AD and allocator issues; it also adds a cuTENSOR workspace option.

Highlights

• ncon has much less overhead: network analysis and tree construction are type stable, intermediates are partitioned as (open A; open B) to avoid repartitioning copies for symmetric tensors, and ncon (including traces) can now be differentiated with Zygote.
• cuTENSORBackend(; workspace = :min | :default | :max) controls cuTENSOR's workspace preference for contractions.
• BufferAllocator now sizes its buffer correctly after an overflow, and AD rules (ChainRules, Enzyme, Mooncake) cover tensorfree!(C, allocator).
• Compat extended to GPUArrays 12 and JLArrays 0.4.

Full Changelog

See CHANGELOG for the complete list of changes.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
see 5 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.