v3.9.0.14

What's Changed

New feature

• Feature: Add support for ML EXX in training script. by @sunliang98 in #6479
• Feature: Add deepks_out_base to support simultaneous output of numpy files of base and target functionals by @xuan112358 in #6483
• Feature: support NVTX profiling via timer_enable_nvtx flag by @wangtianxiang in #6495

Fix

• Fix: Name conflict in pyabacus. by @ErjieWu in #6466
• Fix: Value error of vdrpre in DeePKS. by @ErjieWu in #6476
• Fix: resolve compile error with USE_ELPA=OFF + BUILD_TESTING=ON and switch to nvtx3 headers when CUDA_VERSION >= 12090 by @wangtianxiang in #6497
• [Bug] Fix dsp compilation error by @Critsium-xy in #6499
• Fix: Fix crash in Debug build with multi-GPU due to forced cudaSetDevice(0) by @wangtianxiang in #6498

Refactor

• Delete useless codes in SIAB directory by @mohanchen in #6467
• Refactor: remove the INPUTw input script and transfer parameters to INPUT (SIAB refactor 1/3) by @kirk0830 in #6475
• add output information about device count by @dzzz2001 in #6484
• Update out_dmk and out_dmr by @mohanchen in #6481
• [Refactor] Move hardware initializer out from esolver code by @Critsium-xy in #6494
• Build(deps): Bump actions/setup-python from 5 to 6 by @dependabot[bot] in #6492
• Removed the temporary variable DMRGint_full when transitioning from 2D block parallelism to serial in Hcontainer(develop) by @zgn-26714 in #6489

Tests&Docs

• Test: Add test for out_alllog to ensure the output log filenames are correct by @AsTonyshment in #6482

Perf

• Perf: Optimize Davidson by fusing operators, offloading CPU computation to GPU, and reducing memory transfers by @wangtianxiang in #6493

Version Updates

• Update version to 3.9.0.14. by @ErjieWu in #6504

New Contributors

• @wangtianxiang made their first contribution in #6495

Full Changelog: v3.9.0.13...v3.9.0.14