v3.11.0-beta.3

New feature

• [Feature] Add the USE_KML compile option and change the values of abstol/orfac for KML by @ZhouXY-PKU in #7370

Bug Fixes

• Fix potential Data race in OpenMP by @mohanchen in #7345
• fix(dsp): correct dspDestoryHandle to use cluster ID (MY_RANK % dsp_count) by @Cstandardlib in #7357
• Fix: Prevent missing EXX in RT-TDDFT evolution when using estep_per_md > 1 evolution strategy by @AsTonyshment in #7360
• fix(gint): use fp64 accumulators on the mixed-precision gint path (Useful Information for correctly supporting mixed precision calculations of LCAO) by @dzzz2001 in #7368
• fix(dftu): resolve conflict between yukawa_potential and uramping by @dyzheng in #7369
• [Fix] Avoid to use OMP in fft_cpu.cpp when using KML by @ZhouXY-PKU in #7374
• fix(sdft): add CT (Chebyshev Trace) iter_header for pure SDFT without diagonalization by @Cstandardlib in #7388
• Fix: use correct \psi for GPU in cal_tau for ELF by @Flying-dragon-boxing in #7378

Refactoring

• Refactor: generalize the rule to switch the io backend of abacuslite by @kirk0830 in #7347
• move source_base/memory.h to source_base/memory_recorder.h by @mohanchen in #7348
• Update write_dipole.cpp by @mohanchen in #7349
• refactor(device): remove dead code from DeviceContext, add dsp_count guard by @Cstandardlib in #7361
• Modify headers in source_basis to lower the dependency's depth by @Critsium-xy in #7364
• [Refactor] Remove parameter.h is some files in source_basis by @Critsium-xy in #7365
• [Rafactor] Remove parameter.h in rest files in source_basis by @Critsium-xy in #7373
• Update out_chg 2 and out_pot 3, and elf by @mohanchen in #7379
• Refactor: clean up and fix TDOFDFT evolve_ofdft module by @lyb9812 in #7389

Performance

• add OpenMP in RI_2D_Comm::split_m2D_ktoR() by @linpeize in #7350
• perf(gint): optimize phi_operator kernels (Useful Information for some special optimization commands that have not been used in ABACUS before) by @dzzz2001 in #7366

Documentation & Tests

• Toolchain: Allow controling parallel jobs for ABACUS build after toolchain setup by @Growl1234 in #7376
• Test: separate from #7304 for test cases only by @dyzheng in #7382

Version Update

• Update version to v3.11.0-beta.3. by @ErjieWu in #7397

Full Changelog: v3.11.0-beta.2...v3.11.0-beta.3

v3.11.0-beta.2

New Features

• [Feature] Add Windows Installer for ABACUS by @Critsium-xy in #7267
• [Feature] Allow selecting which version to install for windows installer by @Critsium-xy in #7278
• [Feature] abacusw90: A Wannier90 interface for ABACUS by @ZhouXY-PKU in #7248
• Feature: Implement moving spatial gauge for RT-TDDFT Ehrenfest dynamics (developer preview) by @AsTonyshment in #7300
• Feature: Enable NCCL library for GPU-Direct collective communications in plane wave (BPCG, stochastic KG, etc.) by @Flying-dragon-boxing in #7301
• Add new search algorithm by @19hello in #7337
• Support non-ASCII (CJK) paths in Windows installer by @Critsium-xy in #7342

Bug Fixes

• Fix a GPU compiling issue by @mohanchen in #7255
• fix some format issues by @mohanchen in #7259
• add a small threshold (e.g., 1.0e-15) when writing occupation numbers by @mohanchen in #7257
• Fix: switch to support the solid spherical harmonics that is consistent with other functionalities by @kirk0830 in #7258
• [Fix] Using WARNING_QUIT when setting gamma_only=1 for Wannier90-related calculations by @ZhouXY-PKU in #7266
• Add guard for hsolver rank-deficient cases by @Cstandardlib in #7284
• Fix(pw): suppress spurious warning when no plane waves exist for a k-point by @Cstandardlib in #7282
• [Fix] Fix the wrong paths in the wannier-interface CI/CD workflow by @ZhouXY-PKU in #7293
• fix docker.intel by @mohanchen in #7256
• [Fix] Eliminate unnecessary vkb CPU allocation on GPU path by @Cstandardlib in #7296
• Fix: Fix ELF for kpar > 1 by @sunliang98 in #7307
• Toolchain: Fix package installing issue by @Growl1234 in #7315
• Fix TITLE() in module_xc by @linpeize in #7330
• Fix a matrix comparison in exx symmetry by @maki49 in #7335
• fix output by @mohanchen in #7317
• Fix: Fix missing EXX contribution in RT-TDDFT evolution steps by @AsTonyshment in #7344

Refactoring

• Build(deps): Bump actions/upload-pages-artifact from 4 to 5 by @dependabot[bot] in #7262
• Build(deps): Bump actions/upload-artifact from 4 to 7 by @dependabot[bot] in #7263
• Do not output time.json. If timer information is required, please extract it from running.log instead. by @mohanchen in #7261
• [Refactor] Move abfs-vector3_order from source_base to module_ri by @Critsium-xy in #7270
• Update outputs of LCAO and DeePKS by @mohanchen in #7275
• Remove temporary output file in DeePKS. by @ErjieWu in #7276
• delete libpaw （useless) by @mohanchen in #7273
• remove PAW-related codes by @mohanchen in #7277
• Build(deps): Bump conda-incubator/setup-miniconda from 3 to 4 by @dependabot[bot] in #7287
• Toolchain: Packages updating by @Growl1234 in #7285
• Toolchain: Remove notice of downloading from master branch by @Growl1234 in #7303
• Build(deps): Bump actions/setup-python from 5 to 6 by @dependabot[bot] in #7309
• Build(deps): Bump actions/checkout from 4 to 6 by @dependabot[bot] in #7308
• Refactor: update #include in all files by @linpeize in #7289
• Toolchain 202601 by @QuantumMisaka in #7310
• [Refactor] Remove some dependencies in module_base by @Critsium-xy in #7311
• Refactor: EXX PW now uses the new Parallel_device communication interface by @Flying-dragon-boxing in #7312
• [Refactor] Remove unnecessary manual MPI_Reduce in hsolver dav code by @Cstandardlib in #7314
• Remove parameter.h include of files in source_base by @Critsium-xy in #7318
• Toolchain: Use "filter_setup" functional for a cleaner setup file by @Growl1234 in #7319
• Remove dependencies in rest files in source_base by @Critsium-xy in #7333
• [Refactor] Shorten verbose names in module_gint by @dzzz2001 in #7336

Documentation

• Move windows_installer document's path by @Critsium-xy in #7274
• Doc: Enhance the doc of scf_thr and scf_ene_thr by @sunliang98 in #7280
• Add example Co6Nb4Os2 by @mohanchen in #7316

Version Update

• Update version to v3.11.0-beta.2 by @AsTonyshment in #7343

Full Changelog: 3.11.0-beta.1...v3.11.0-beta.2

3.11.0-beta.1

New Features

• Feature: add OpenMP in XC_Functional_Libxc::v_xc_libxc() by @linpeize in #7130
• Feature: Implement ext-WT KEDF for OFDFT by @sunliang98 in #7231
• Feature: use new format of CSR file for out_mat_hs2 and out_mat_ds, out_mat_t, out_mat_r, out_mat_xc2 (Useful Information for output format update of H(R), S(R) and other matrices that based on NAO basis set) by @dyzheng in #6991
• feat(input, kpoints): support shifted kspacing mesh with koffset and kmesh_type by @Cstandardlib in #7247

Bug Fixes

• [Build] Fix SCALAPACK LIBRARY find logic by @Cstandardlib in #7214
• FIx: fix cuda compilation warning by @dzzz2001 in #7215
• Fix: Fix UPF100 negative lmax handling by @sunliang98 in #7221
• Fix: Enhance the warning of ntype by @sunliang98 in #7239
• fix bug in Matrix_Orbs22::cal_grad_overlap_matrix() by @linpeize in #7254

Refactoring

• Refactor onsite_proj COMPONENTS and add dspin examples by @mohanchen in #7211
• [Refactor] Remove redundant real2recip by @Cstandardlib in #7228
• Toolchain: Remove redundant exporting variables by @Growl1234 in #7230
• refactor: remove Array_Pool in favor of std::vector by @dzzz2001 in #7234
• refactor: find scalapack automatically by @Cstandardlib in #7235
• Update CMake to formally name the five levels of ABACUS executables (Useful Information for knowing Five Compilation Levels of ABACUS) by @mohanchen in #7236
• Update examples and interfaces directories by @mohanchen in #7238
• Toolchain: Enable float FFTW linking for building with MKL by @Growl1234 in #7245
• Upload new script for drawing absorption spectra by @AsTonyshment in #7246

Tests

• [Tests] Add test for small cell by @Cstandardlib in #7216
• Change threshold in 01_PW/208_PW_CG_float by @19hello in #7241
• Add CI/CD workflow for Toolchain by @QuantumMisaka in #7244

Documentation

• [Docs] Fix out_chg docs by @Cstandardlib in #7227
• Doc: Update the doc of out_dmk by @sunliang98 in #7240

Version Update

• Update version to v3.11.0-beta.1 by @sunliang98 in #7249

Full Changelog: v3.9.0.27...3.11.0-beta.1

v3.9.0.27

New Features

• Feature: support the setting on atomic magnetic moment via ase.atoms.Atoms.magmom by @kirk0830 in #7124
• Enable compatibility of GNU compilers + MKL by @Growl1234 in #7139
• feat(gint): add single-precision (fp32) support for grid integration module (Useful Information for implementing mixed (single/double) precision calculations for grid integral operation by @dzzz2001 in #7149
• feat(gint): enable mixed-precision (fp32/fp64) support for GPU path by @dzzz2001 in #7207

Bug Fixes

• Fix: Fix toolchain installation of Cereal by @Flying-dragon-boxing in #7116
• Fix: add cccl include path for CUDA >= 13.0 where thrust moved to cccl by @Flying-dragon-boxing in #7155
• Fix: remove device=cpu restriction for EXX PW calculation by @Flying-dragon-boxing in #7156
• Fix: add runtime check for cudaHostRegister in EXX PW by @Flying-dragon-boxing in #7157
• 2026PKUCourseHW5：verify output stream open in genelpa matrix save helpers by @chaomianpkp in #7175
• Fix: Fix several bugs in KEDF get_energy functions by @sunliang98 in #7169
• 2026PKUCourseHW5: Fix(io): check nxnynz before glo_order allocation in read_wfc_pw by @chaomianpkp in #7180
• 2026PKUCourseHW5:mathzone_add1.cpp/case9-Variable initialization by @lxz2023 in #7181
• Fix: implement special treatment in abacuslite I/O on the SCF unconverged case by @kirk0830 in #7195
• Fix: port sdft memory op to DSP by @Cstandardlib in #7203
• Fix the issue of inconsistent behavior between estep and md_nstep. by @ESROAMER in #7198
• Fix: add EXX erfc parameter handling for pw basis type by @Flying-dragon-boxing in #7166

Refactoring

• Refactor: split timer::tick() to timer::start() and timer::end() by @linpeize in #7020
• Remove the use of std::system in DeePKS_equiv. by @ErjieWu in #7122
• Refactor: Remove the ML training code by @sunliang98 in #7167
• 2026PKUCourseHW5: Replace C-style cast with static_cast in read_pp_complete.cpp by @Howard-Shao09 in #7183
• Fix IntArray class and a few math classes with memory management and code style improvements by @mohanchen in #7112
• DFTU: Use ScalapackConnector instead of direct math library calls by @lanshuyue in #7196
• Add initialization checking in BFGS, and replace the old BFGS with a new one in CG-BFGS. by @19hello in #7117
• Toolchain: Use fixed commit hash instead of master branch by @Growl1234 in #7193
• Refactor ESolver: for pw basis by @mohanchen in #7205

Tests

• Tests: Lower the threshold for CI/CD test 096_PW_PBE0_FM by @Flying-dragon-boxing in #7118
• Build(deps): Bump codecov/codecov-action from 5 to 6 by @dependabot[bot] in #7190

Documentation

• Build(deps): Bump actions/configure-pages from 5 to 6 by @dependabot[bot] in #7191
• Build(deps): Bump actions/deploy-pages from 4 to 5 by @dependabot[bot] in #7189
• Update version to v3.9.0.27 by @DanielZhao0432 in #7206

New Contributors

• @chaomianpkp made their first contribution in #7175
• @lxz2023 made their first contribution in #7181
• @Howard-Shao09 made their first contribution in #7183
• @Growl1234 made their first contribution in #7139
• @DanielZhao0432 made their first contribution in #7206

Full Changelog: v3.9.0.26...v3.9.0.27

v3.9.0.26

What's Changed

Feature

• Feature: add class PotXC_FDM and Pot_Cosikr by @linpeize in #6962
• [Feature] Add gemv_op_mt for DSP by @Cstandardlib in #7009
• [Feature] Allow compile LCAO version on DSP by @Critsium-xy in #7013
• Feature: Multi-GPU support for RT-TDDFT by @AsTonyshment in #7026

Fix

• Fix: correct the unit conversion between Plumed and ASE in example by @kirk0830 in #6993
• Fix buffer overflow in H_Ewald_pw::rgen for very small unit cells / high ecutwfc by @Copilot in #7025
• [Fix] Fix mxr, use estimate for hardcoded instance by @Cstandardlib in #7036
• Fix DSP count check problem by @Critsium-xy in #7035
• Fix: use direct import of full_3x3_to_voigt_6_index by @kirk0830 in #7042
• Fix: dsp memory op by @Cstandardlib in #7056
• Fix: Add DSP gemm pack with auto memcpy to buffer by @Cstandardlib in #7060
• Fix: Fix DSP gemv_op_mt compile by @Cstandardlib in #7079

Refactor

• Refactor: Optimize memory management in RT-TDDFT to enable large-scale GPU calculations by @AsTonyshment in #6995
• Refactor: remove dependence of Tensor for cg interface by @Cstandardlib in #7000
• Refactor: Use a unified hpsi_func/spsi_func for all ks_solvers by @Cstandardlib in #7002
• refactor esolver_ks_pw by @mohanchen in #7008
• Refactor device parameter in esolver_ks_pw by @mohanchen in #7016
• Refactor: add nullptr for uninitialized pointer by @linpeize in #7027
• Refactor: add nullptr for uninitialized pointer by @linpeize in #7069
• Refactor: add nullptr for uninitialized pointer by @linpeize in #7078
• Esolver refactor for LCAO and pw by @mohanchen in #7012
• Try removing template dependency of ESolver by @mohanchen in #7014
• We remove the template of ESolver_KS !!! by @mohanchen in #7015
• Remove template of ESolver_KS_pw: just a first step by @mohanchen in #7024
• Let’s try removing the template of ESolver_KS_pw again by @mohanchen in #7028
• add const for variable functions by @linpeize in #7037

Others(Docs / Build / Tests / quality / version)

• Code quality improvements by @mohanchen in #7001
• Build(deps): Bump actions/checkout from 4 to 6 by @dependabot[bot] in #6988
• Build(deps): Bump crazy-max/ghaction-github-pages from 4 to 5 by @dependabot[bot] in #6990
• Build(deps): Bump actions/upload-artifact from 6 to 7 by @dependabot[bot] in #6989
• Build(deps): Bump docker/setup-buildx-action from 3 to 4 by @dependabot[bot] in #7006
• Build(deps): Bump docker/login-action from 3 to 4 by @dependabot[bot] in #7005
• Build(deps): Bump docker/metadata-action from 5 to 6 by @dependabot[bot] in #7004
• Build(deps): Bump docker/build-push-action from 6 to 7 by @dependabot[bot] in #7003
• Docs: fix default value for ks_solver by @Cstandardlib in #7011
• [Tests] Remove sum_cube.cpp that has been replaced by sum_cube.py by @Cstandardlib in #7029
• add output-specification in doc (Useful information for formats of output files) by @mohanchen in #7030
• Put on the link of output-specification in doc by @mohanchen in #7031
• Stabilize RI RPA/EXX and Ewald EXX method merge on develop by @AroundPeking in #7019
• Update version to v3.9.0.26 by @ieiue in #7070

New Contributors

• @PKUJZX made their first contribution in #7038
• @samomoyed made their first contribution in #7043
• @RXL6105505822 made their first contribution in #7048
• @JohnHe233 made their first contribution in #7064

Full Changelog: v3.9.0.25...v3.9.0.26

v3.9.0.25

What's Changed

Feature

• feat: add built-in help system with fuzzy search for INPUT parameters (Useful Information for adding new input parameters and explanations) by @zhubonan in #6935
• Feature: support pyabacus to run abacus directory (Useful information for the following developments of pyabacus) by @dyzheng in #6943
• Feature: ABACUS light-weight ASE plugin implementation by @kirk0830 in #6950

Fix

• Fix GPU single precision energy error in dav_subspace solver by @zhubonan in #6946
• Fix memory recorder by @Critsium-xy in #6952
• Fix: Fix RT-TDDFT segmentation fault caused by unconverged initial SCF by @AsTonyshment in #6954
• Fix C++17 inline constexpr to support C++11 by @dzzz2001 in #6957
• Build: create symbolic link after build by @kirk0830 in #6967
• Fix some minor formatting issue for the input-main.md by @zhubonan in #6983
• fix: nspin = 2 for init_chg=dm by @dyzheng in #6975
• Fix pw_diag_nmax default value by @Cstandardlib in #6985

Refactor

• Optimize vloc_of_g by skipping negligible analytic terms by @A-006 in #6932
• Refactor: Device module modernization and GPU initialization consolidation (Useful information for GPU coding) by @dzzz2001 in #6936
• Refactor: Unify CUDA spherical harmonics (sph_harm) implementation (Useful information for using Ylm in both CPU and GPU coding) by @dzzz2001 in #6947
• Refactor: support modern cuSOLVERMp with NCCL by @dzzz2001 in #6951
• Update CMakefile, refactor some files in source_io by @mohanchen in #6959
• Refactor output format of Mulliken charge analysis, pchg by @mohanchen in #6969
• Refactor the test directories of 02_NAO_Gamma and 03_NAO_multik, and several related output formats by @mohanchen in #6973
• Add interfaces directory by @mohanchen in #6978

Test

• [Tests] Set DOT_NUM_THREADS and force clean workspace to make the doxygen test and the container test more robust. by @ZhouXY-PKU in #6906
• Tests: Use sum_cube.py in place of sum_cube.cpp by @Cstandardlib in #6948
• [Test] Add a new category in Unit test for estate by @Critsium-xy in #6977
• Tests: Simplify MODULE_HSOLVER test to reduce time consumption by @Cstandardlib in #6980
• Update 01_PW 02_NAO_Gamma 03_NAO_multik directories by @mohanchen in #6979
• Tests: Do not run time-consuming PERF_MODULE_HSOLVER_KERNELS in CI/CD by @Cstandardlib in #6982
• [Test] Update the classification of unit tests by @Critsium-xy in #6987

Doc

• Docs: update cuSolverMP documentation for HPC SDK version differences by @dzzz2001 in #6955
• docs: Update example and docs for GPU-LCAO acceleration by @QuantumMisaka in #6961
• Docs: update the documentation of abacuslite by @kirk0830 in #6984

Version Update

• Update version to v3.9.0.25 by @ZhouXY-PKU in #6986

New Contributors

• @zhubonan made their first contribution in #6946

Full Changelog: v3.9.0.24...v3.9.0.25

v3.9.0.24

What's Changed

Feature

• [Feature] Update LAPACK solver by @Critsium-xy in #6870
• Feature: Introduce ML_Base class and add ML_EXX support by @sunliang98 in #6884
• Feature: add Hessian operator <\phi|\nabla_x\nabla_y|\phi> by @dyzheng in #6888
• Support using hybrid functionals for RT-TDDFT calculations. by @ESROAMER in #6886
• Feature&Doc: support fix_axes and fix_ibrav for relax_new=false and optimized doc by @dyzheng in #6858
• Feature: support ELF with non-collinear spin (nspin = 4) by @dyzheng in #6912

Fix

• Fix: change uplo to U for lapack_hegvd to be consistent with former routine by @Cstandardlib in #6868
• [Fix]Update source files for NEP installation script by @ZhouXY-PKU in #6871
• Fix: eliminate compile warning of iteration index(int) undefined behavior by @Cstandardlib in #6874
• Fix: cusolverMP BLACS grid mapping order causing errors by @dzzz2001 in #6885
• Fix a readin bug, add pchg test in integrate tests, update some examples by @mohanchen in #6878
• Fix: enhance several warning messages by @sunliang98 in #6908
• Fix dependency cycles by @mohanchen in #6902
• Fix: Minor fixes for EXX PW by @Flying-dragon-boxing in #6877
• Fix: Hefei-NAMD interface of file syns_nao.csr and parameter cal_syns by @dyzheng in #6910
• Fix: Robust handling of bandgap boundary conditions in ElecState and add unit tests by @A-006 in #6917
• Fix some initialization problems by @mohanchen in #6921
• Fix: useless output with EXX and symmetry on by @dyzheng in #6934

Refactor

• [Refactor] Update init_wfc message to be user-friendly by @Cstandardlib in #6876
• Refactor: save memory for kinetic and overlap force and stress by @dyzheng in #6881
• Refactor: Delete additional wavefunction initialization in lcao_others by @AsTonyshment in #6887
• Refactor/device check migration by @mohanchen in #6895
• Migrate exx_info by @mohanchen in #6896
• Try to delete global.h, the use of which is harmful for the whole package by @mohanchen in #6898
• refactor(gint): remove obsolete temp_gint directory by @dzzz2001 in #6899
• Refactor&Doc: simplify read_atoms.cpp and update stru.md doc by @dyzheng in #6903
• Reorganize the pw module source files by @mohanchen in #6913
• Refactor: Replace goto with while loop in occupy.cpp by @Cstandardlib in #6919
• Refactor: rename to delete new from names of operators in source_lcao by @dyzheng in #6920
• Refactor Parallel_Reduce: Replace manual gather-reduce with MPI_Allreduce by @A-006 in #6922
• Refactor Parallel_Reduce and optimize reduction operations by @A-006 in #6925
• Refactor: Refine ELF Output for Noncollinear Spin (nspin=4) by @sunliang98 in #6914
• Refactor: Skips unnecessary ACE operator updates during the final convergence step by @Flying-dragon-boxing in #6938
• Reorganize the source_psi and source_io directories by @mohanchen in #6923

Test

• Test: Add GPU tests for dav_subspace by @Cstandardlib in #6869
• [Tests]Fix md_nsteps in 04_FF (50DP&101NEP) by @ZhouXY-PKU in #6872
• [Tests] Update PEXSI download link in Dockerfile.intel to use a more robust GitHub source by @ZhouXY-PKU in #6894

Others

• ci: use OIDC for codecov-action by @njzjz-bot in #6891
• Some small updates, including timer, and eliminate clean function in esolver by @mohanchen in #6892
• Docs: Fix doc and warning for init_chg, nspin=1 by @Cstandardlib in #6929

Full Changelog: v3.9.0.23...v3.9.0.24

v3.9.0.23

What's Changed

Feature

• Feature: init_chg hr for initializing charge density with hrs1_nao.csr by @dyzheng in #6791

Fix

• Fix kpar check for DSP by @Cstandardlib in #6825
• Fix local pseudopotential energy calculation with k-point parallelization by @A-006 in #6827
• Fix: Resolve memory leak when using solvation model in relaxing geometry by @LKFEIYI in #6835
• [Fix] Fix Docker directory permissions in workflow by @ZhouXY-PKU in #6845
• Fix: Al ESP result discrepancy & add wavefunction normalization by @lyb9812 in #6853
• Fix: Yet another wrong index in act_op_kpar for spin-polarized calculations in EXX PW by @Flying-dragon-boxing in #6847
• Fix: correct nvc++ compiler compatibility issues by @dzzz2001 in #6854
• Fix: doc of soc by @dyzheng in #6856
• Fix&Doc: do not overwrite k-point weights for non-MP k-point lists by @dyzheng in #6857
• Fix: no k weights for get_wf by @dyzheng in #6859
• Fix: skip MPI test for lapack UT by @dyzheng in #6861
• Fix cudaErrIllegalAddress by using correct device pointer by @Cstandardlib in #6862

Refactor

• [Refactor] Replaced all exposed mathematical library interface by @Critsium-xy in #6828
• Simplify Lmax and rmesh in Matrix_Orbs and Center2_Orb by @linpeize in #6829
• Refactor class DensityMatrix. by @linpeize in #6831
• Add check for init_wfc to use proper atomic/random starting wavefunction by @Cstandardlib in #6833
• add DensityMatrix<complex,complex> by @linpeize in #6834
• Refactor implicit solvation model by @LKFEIYI in #6836
• Refactor: Remove dependency on shell commands for file operations by @A-006 in #6826
• Refactor: reduce FFT calls in Leps2 by @LKFEIYI in #6839
• [Refactor] Complete clean exposed mathematical library interfaces by @Critsium-xy in #6838
• [Refactor] Complete rework of diago_lapack by @Critsium-xy in #6843
• Refactor and Fix Parallel Handling of Plane Wave Counts in Eigenvalue Output by @A-006 in #6844
• Refactor ModuleESolver::init_esolver() by @linpeize in #6849
• Refactor: simplify snap_psibeta_half_tddft.cpp by @dzzz2001 in #6846

Test

• Add new tests in the integrate tests of LCAO gamma only calculations and update the index to count relax/MD steps, now all steps start from 1, not 0 by @mohanchen in #6821

Others

• Optimize install_openmpi.sh script by @19hello in #6863
• Bump version to v3.9.0.23 by @Cstandardlib in #6866

Full Changelog: v3.9.0.22...v3.9.0.23

v3.9.0.22

What's Changed

Feature

• Feature: Add a new parameter out_freq_td to control output frequency in RT-TDDFT by @AsTonyshment in #6796
• Feature: add exx_cs_inv_thr to get inverse Coulomb by @AroundPeking in #6793

Fix

• Fix: indptr overflow in output big HR. by @ErjieWu in #6803
• Fix: Add a threshold to avoid E_gap(k)=0 for insulators by @LKFEIYI in #6802
• Fix: Update NEP interface to match NEP_CPU's update by @MoseyQAQ in #6804
• fix: GPU memory leak in deriv_drhoc_scc by @dzzz2001 in #6809
• Fix:: replace deprecated CUDA_VERSION and handle CUDA 13+ architecture requirements by @Flying-dragon-boxing in #6810

Performance

• perf(TDDFT): Add CUDA acceleration for snap_psibeta_half function (Useful information about largely improves the snap_psibeta_half function) by @dzzz2001 in #6808

Refactor

• Simplify Lmax in Center2_Orb and Matrix_Orbs by @linpeize in #6806
• Simplify Lmax and ORB_gaunt_table in Center2_Orb and Matrix_Orbs by @linpeize in #6819

Others

• Update version to 3.9.0.22. by @ErjieWu in #6820

New Contributors

• @LKFEIYI made their first contribution in #6802
• @AroundPeking made their first contribution in #6793

Full Changelog: v3.9.0.21...v3.9.0.22

v3.9.0.21

What's Changed

Fix

• Fix: Fix an extremely serious bug that causes incorrect results in GPU RT-TDDFT calculation by @AsTonyshment in #6767
• Fix energy jumps occur in rt-TDDFT calculations by @ESROAMER in #6779
• [Fix] Report and exit for unsupported DSP KPAR configuration by @Cstandardlib in #6771
• [Fix]Enable abacus to be compiled with CXX 17 & CUDA 13 by @ZhouXY-PKU in #6777

Feature

• Add average iterative diagonalization step output by @Cstandardlib in #6795

Refactor

• Refactor and Optimize snap_psibeta_half_tddft by @dzzz2001 in #6765
• [Refactor] Support different CUDA versions in one single cuda_compat.h by @Cstandardlib in #6770

Test

• Test: Update the parameter and reference results of 23_OF_LPS by @sunliang98 in #6774
• Test: Update CI/CD tests for RT-TDDFT and add hybrid gauge and parameter td_dt tests for CPU and GPU by @AsTonyshment in #6776
• [Test] Use WARNING_QUIT instead of exit, Add and refactor unittests for ParaGlobal by @Cstandardlib in #6781

Others

• Build(deps): Bump actions/upload-artifact from 5 to 6 by @dependabot[bot] in #6783
• Update version to v3.9.0.21 by @xuan112358 in #6797

Full Changelog: v3.9.0.20...v3.9.0.21