v3.9.0.9

What's Changed

Feature

• Feature&Doc&Tests: Support for deepks_v_delta=-1&-2 by @ErjieWu in #6245
• Add deepks_out_freq_elec for outputting deepks labels during electronic steps by @xuan112358 in #6325
• Feature: long-range corrected hybrid functionals implementation with Spencer-Alavi truncation Coulomb potential by @1041176461 in #6351

Fix

• Fix: reinterpret_cast bug for c10::complex. by @ErjieWu in #6321
• Reformat running log, delete 'printe' command in INPUT and fix the output eigenvalue information issue by @mohanchen in #6326
• Fix: skip mix_dmr after the last iteration by @dzzz2001 in #6331
• Fix: Fix Compile Error on CUDA Version > 12.9 by @Flying-dragon-boxing in #6337
• Fix: Fix calculation type inconsistency across MPI ranks, so that the log file names are correct when out_alllog = 1 by @AsTonyshment in #6356
• Fix: Add MPI synchronization for ekb band energy tensor, and now GPU CI/CD test case 07_NO_EDM_TDDFT_GPU is functional again by @AsTonyshment in #6354

Refactor

• change Exx_Info_Global::coulomb_param from unordered_map to map (Do not recommend using unordered_map, which has defects in C++11) by @linpeize in #6318
• change INPUT exx_ for linear combination of coulomb_param by @linpeize in #6316
• [Refactor] Rename module_basis to source_basis || module_cell to source_cell by @Critsium-xy in #6319
• Refactor: release the limit (<150) on length of comment lines of INPUT by @kirk0830 in #6329
• Refactor: clean Davidson code, remove deprecated overlap matrix scc related code by @Cstandardlib in #6330
• [Refactor] Rename module_elecstate and module_psi and module_hamilt_pw by @Critsium-xy in #6334
• Remove INPUT parameter nbands_istate and unify the naming of real space wave function and partial charge density by @AsTonyshment in #6336
• new gint module (This is a big PR that replaces old integral codes) by @dzzz2001 in #6296
• [Refactor] Rename module_md, module_relax, module_hamilt_lcao and module_io by @Critsium-xy in #6340
• [Refactor] Move module_ri to source_lcao/module_ri and module_rdmft to source_lcao/module_rdmft and correct some documents by @Critsium-xy in #6344
• Refactor: new ccp param framework by @1041176461 in #6350
• Refactor: Refactor KEDF management in ESolver_OF by @sunliang98 in #6349
• [Refactor] Improve type safety with std::complex by @kluophysics in #6346
• [Refactor] Rename module_tddft and hamilt_ofdft and hamilt_stodft and hamilt_pwdft by @Critsium-xy in #6353
• Remove useless and outdated write_psi_r_1 interface and improve related docs and examples by @AsTonyshment in #6345
• Add "#TOTAL ENERGY#" after each ion step by @mohanchen in #6342

Full Changelog: v3.9.0.8...v3.9.0.9

v3.9.0.8

What's Changed

Fix

• Fix: fix a bug of out_mat_dh by @dzzz2001 in #6283

Feature

• Add Mt_Gemm for the nonlocal_pw by @A-006 in #6253
• Feature: universal customized parameterization on exchange-correlation functionals by @kirk0830 in #6270
• Feature: add complex erf function by @1041176461 in #6301
• add exx nscf file check by @linpeize in #6288
• add write_libxc_r by @linpeize in #6293
• Combine relax_new with relax_old by @19hello in #6303
• print out geometry information during bfgs_trad relax by @lanshuyue in #6314

Refactor

• Reconstruct the LCAO codes for the parameter 'PARAM.inp.out_interval', enable a unified control of output information every few ionic steps. by @mohanchen in #6272
• Update the 'get_s' command and related description. Update the 'DOS' output filenames. Move some functions in 'after_all_runners' to 'after_scf'. Change 'out_dm' to 'out_dmk' and change 'out_dm1' to 'out_dmr' by @mohanchen in #6278
• Move DeePKS printing message from screen to log file. by @ErjieWu in #6281
• Refactor: replace ENABLE_DEEPKS & ENABLE_MLKEDF with ENABLE_MLALGO by @haozhihan in #6271
• Refactor: Migrate ML KEDF Descriptor Calculation & Output to module_io by @sunliang98 in #6287
• Refactor: Combine some checking functions in DeePKS. by @ErjieWu in #6285
• Refactor: Optimize the code structure of K_Vectors by @Flying-dragon-boxing in #6263
• Change examples and update DOS files, etc. by @mohanchen in #6291
• [Refactor] Move certain files to source_main by @Critsium-xy in #6290
• refactor Exx_Info::coulomb_param for RI by @linpeize in #6295
• Update output filenames for PCHG by @mohanchen in #6299
• [Refactor] Rename module_esolver to source_esolver by @Critsium-xy in #6300
• [Refactor] Rename module_hsolver to source_hsolver by @Critsium-xy in #6305
• Refactor: the ccp parameters to support additional Coulomb matrix construction methods. by @1041176461 in #6307
• [Refactor] Rename module_base to source_base by @Critsium-xy in #6311
• Refactor: change the name of keyword force_thr_ev2 to force_zero_out for clarity by @kirk0830 in #6313
• remove deprecated INPUT exx parameters by @linpeize in #6317

Test

• Test: Add version check workflow for release validation by @sunliang98 in #6273

Version Update

• Update version.h by @xuan112358 in #6320

New Contributors

• @lanshuyue made their first contribution in #6314

Full Changelog: v3.9.0.7...v3.9.0.8

v3.9.0.7

What's Changed

Fix

• Fix: MPI communication errors due to inconsistent R-coordinates in sparse matrix generation​ by @AsTonyshment in #6233
• add check and update code format in exx by @linpeize in #6255

Feature

• Add more types bandgap of labels for deepks by @Chentao168 in #6226
• Feature: Support outputting partial charge and wave functions under PW basis when kpar > 1 by @AsTonyshment in #6242
• Feature: Spin-polarized calculations for EXX PW by @Flying-dragon-boxing in #6260
• Feature: implement k continuity initialization strategy & kernel by @jieli-matrix in #6171

Refactor

• Update out_dm and out_dm1 parameters and related documents by @mohanchen in #6237
• Update the file names controlled by the out_pot command by @mohanchen in #6240
• Refactor&Docs: Simplify deepks_bandgap realization and add input value checks. by @ErjieWu in #6241
• Refactor: Standardize OFDFT screen output to match KSDFT format by @sunliang98 in #6249
• Change the file names of output charge densities and kinetic energy densities by @mohanchen in #6254
• Exx ccp parameter by @linpeize in #6256
• Change 'istate.info' to 'eig.txt', add explanations for the parameter "out_band" (open files eigs1.txt and eigs2.txt). In the 'after_all_runners" subroutine in SDFT, use after_all_runners from ESolver_KS by @mohanchen in #6257
• Update the output filenames of H(k), S(k), T(k) and Vxc. In addition, modify the filenames of H(R), S(R), T(R), and Vxc by @mohanchen in #6262

Test

• Test: Rename the integrate tests of OFDFT. by @sunliang98 in #6246

Docs

• Docs: EXX PW Documents by @Flying-dragon-boxing in #6259

Version Updates

• Bump version v3.9.0.7 by @Flying-dragon-boxing in #6275

Full Changelog: v3.9.0.6...v3.9.0.7

v3.9.0.6

What's Changed

Feature

• Feature: Support cube format real-space wave function output for Plane Wave basis set by @AsTonyshment in #6217
• Update the output and readin of electronic wave functions in plane wave basis set by @mohanchen in #6223
• Update the output filenames when out_wfc_lcao turns on by @mohanchen in #6205
• Reactor FFT format and add RAII for resource handler by @A-006 in #6156

Fix

• Fix: correct the names of ut generated temp files in std::remove func by @kirk0830 in #6208
• Fix: Replace out-dated parameter bands_to_print with out_pchg and update related codes and docs by @AsTonyshment in #6211
• Fix: Fix minor INPUT parameter abuse in get_wf CI/CD test by @AsTonyshment in #6215
• Resolve the segmentation fault occurring in the pw float implementation by @A-006 in #6130
• fix a bug, when out_chg becomes 2, the read in function fails by @mohanchen in #6235
• Fix: if restart but can't find Hexx files, warning and run as usual (automatically initialized the density) by @xuan112358 in #6194

Refactor

• Refactor: delete paw code, will add back the paw code with a better plan. by @haozhihan in #6225

Test

• [Test] Separate unit test for module_base as an example by @Critsium-xy in #6197
• [Test] Add RT-TDDFT GPU integrate tests and fix some related bugs/typos by @AsTonyshment in #6201
• [Test] Make tests in module_io, module_hsolver and module_cell indenpendent by @Critsium-xy in #6218
• [Test] Make tests in Module_Psi, Module_MD and Module_RI indenpendent by @Critsium-xy in #6232
• [Test] Add tests in subdirectories to its module by @Critsium-xy in #6231

Version Updates

• Update version.h (v3.9.0.5) by @haozhihan in #6209
• bump version to 3.9.0.6 by @denghuilu in #6239

Full Changelog: v3.9.0.5...v3.9.0.6-patch

v3.9.0.5

What's Changed

Feature

• Feature: multi-k case support for out_dm=1 by @ErjieWu in #6181
• Update HSE examples, reduce the testing time, modify some outputs for HSE and LR calculations, reformat the output atomic positions, change 'jd' files in integreate tests to 'README' files by @mohanchen in #6182
• Update the documents for input parameters by @mohanchen in #6192
• [Toolchain] 2025-02 version update by @QuantumMisaka in #6189

Fix

• Fix: Enhance the computational stability of ELF. by @sunliang98 in #6166
• Refactor: toolchain fixes for Intel oneapi 2025 for #5518 by @kluophysics in #6173

Refactor

• [Refactor] Separate blas_connector.cpp into three files by @Critsium-xy in #6178

Perf

• Perf: add ccache support to cuda compiler by @dzzz2001 in #6187

Test

• Test: Add unit tests and integrate tests for new features of DeePKS. by @ErjieWu in #6140
• Update integrate tests to reduce running time, change improper output style and files by @mohanchen in #6179
• [Test] Now integrated test will run after other tests by @Critsium-xy in #6183
• Tear down the integration tests into more than 10 subdirectories by @mohanchen in #6186
• Updating the integraete tests, reorganize 02_NAO_Gamma and 03_NAO_multik directories by @mohanchen in #6188

Full Changelog: v3.9.0.4...v3.9.0.5

v3.9.0.4

What's Changed

Feature

• Feature: add out_mat_ds to print <phi|dphi> matrix by @dyzheng in #6149
• Feature: LDOS mode for pw by @YuLiu98 in #6167

Fix

• Fix: A bug caused by the incorrect update of DMR for deepks_out_labels>0 but deepks_scf=0 by @ErjieWu in #6155
• Fix: Fix periodic boundary condition search issue by enforcing search_pbc=true by @AsTonyshment in #6163
• Fix: add MPI_Finalize in QUIT function by @Critsium-xy in #6160
• Fix : fix the segment fault of md/relax of deepks by changing location of ld.init_DMR to before_scf by @xuan112358 in #6170

Refactor

• Refactor: reduce the read-in loop in sts_line by @YuLiu98 in #6148
• Refactor the rt-TDDFT integrate tests by @mohanchen in #6157
• Update namespace ModuleGint by @linpeize in #6169
• Keep refactoring the integrate tests by @mohanchen in #6165
• Update get_wf codes and output formats by @mohanchen in #6154

Perf

• Perf: Use phialpha in calculate_HR in DeePKS to avoid duplicate calculations. by @ErjieWu in #6159

Test

• delete some useless integrate tests (in my opinions, we are wasting time to run these tests, they should be covered by unittests) by @mohanchen in #6152
• Add Gtest for GPU pw_basis and pw_basis_k by @A-006 in #6087
• Delete some useless integrate tests by @mohanchen in #6164

Full Changelog: v3.9.0.3...v3.9.0.4

v3.9.0.3

What's Changed

Feature

• Feature: Make KG-KSDFT support GPU by @Qianruipku in #6013
• Add two LCAO base group GPU version compilation options in toolchain by @tang070205 in #6014
• add LBFGS method by @19hello in #6005
• Feature: Make sKG support GPU by @Qianruipku in #6046
• Feature: write Vxc(R) by @maki49 in #6061
• Feature&Docs: support the indent control on FmtTable by @kirk0830 in #6055
• Feature: print the expectation value of angular momentum operators Lx/Ly/Lz for ABACUS NAO basis by @kirk0830 in #6097
• Feature: LDOS in pw basis set by @YuLiu98 in #6100
• Feature : deepks_out_labels = 2 for label preperation in deepks training by @xuan112358 in #6104
• Feature: LDOS in lcao basis sets by @YuLiu98 in #6123
• Feature: Add gamma point extrapolation for EXX PW by @Flying-dragon-boxing in #6118
• Feature: add postprocess for stm plot by @YuLiu98 in #6137

Fix

• Fix : out_mat_dh will lead to different result with MPI-1core with MPI-4core by @xuan112358 in #6018
• [Fix] Fix DCU port compilation error by @Critsium-xy in #6017
• Fix: calculate magnetic moments with onsite_radius>0(DFT+U case) by @dyzheng in #6024
• Fix: Enhance the warning message when the XC name cannot be recognized. by @sunliang98 in #6025
• Fix: Angular momentum quantum number check in reading SOC pseudopot file by @AsTonyshment in #6027
• Add genelpa warning if matrix is not positive definite by @Cstandardlib in #6022
• Fix: Enable EXX PW ACE only when separate_loop is on by @Flying-dragon-boxing in #6043
• Fix: directly enter exx loop when init_wfc=="file" by @maki49 in #6019
• Fix: DFT+U force&stress with orbital_corr of some elements are -1 by @dyzheng in #6049
• Fix: add the print header for cusolvermp in scf info by @kirk0830 in #6038
• Fix: nan values in LCAO band-decomposed charge density (get_pchg) by @AsTonyshment in #6060
• Fix an output for debug by @maki49 in #6066
• Fix: add a check before calculating EXX force by @maki49 in #6067
• Fixing problem for running meta functionals with kinetic energy density mixed by @BariumOxide13716 in #6071
• Fix: modify warning output by @dzzz2001 in #6074
• Fix: Sign problem in DeePKS cal_pdm(). by @ErjieWu in #6082
• Fix: std::out_of_range of map when out_mat_t with multi-core by @YuLiu98 in #6090
• fix a typo(missing newline) in cuda.md by @OutisLi in #6103
• Fix: v_delta_precalc error in DeePKS. by @ErjieWu in #6106
• Fix: add exit 1 for shell UTs by @Qianruipku in #6122
• Fix: bug of divide_mpi_groups by @Qianruipku in #6120
• Fix some output information typo by @mohanchen in #6133
• Fix: Enhance the computational stability of tau_vw. by @sunliang98 in #6136
• Fix: Avoid hexadecimal output during relax process of HSE by @xuan112358 in #6146
• Fix: update weight in ldos by @YuLiu98 in #6144

Refactor

• Refactor: math_kernel_op by @Qianruipku in #6020
• Add templates for recip_to_real in the pw_basis by @A-006 in #6023
• Refactor symmetr.cpp, update some output formats in surfchem, etc. by @mohanchen in #6052
• Update output formats in running.log by @mohanchen in #6053
• Refactor: set dm_to_rho as a new esolver by @YuLiu98 in #6059
• Refactor: Remove some files in DeePKS and add some features. by @ErjieWu in #6065
• Refactor:Change potentials to module_pot by @A-006 in #6070
• Refactor: reduce the memory consumption of cal_gint_vlocal by @dzzz2001 in #6069
• Update ABACUS output formats, accept it after 3.10 version by @mohanchen in #6076
• Refactor: replace 3.14 using PI by @YuLiu98 in #6085
• Refactor: Remove C++17 syntax and optimize DMR calculation in DeePKS. by @ErjieWu in #6094
• Rafactor:remove GlobalC::rhopw by @A-006 in #6093
• Update Hamiltonian and Operator classes by @mohanchen in #6084
• [Refactor] Remove uncapsulted scalapack functions in dftu_occup.cpp by @Critsium-xy in #6099
• Refactor: Simplify some functions in DeePKS. by @ErjieWu in #6102
• Update DOS output formats (pw basis). by @mohanchen in #6113
• Refactor&Docs: add annotation beside commented-out lines in VDW module by @kirk0830 in #6117
• Refactor the LCAO DOS codes by @mohanchen in #6127
• Refactor: Move cal_tau to esolver_ks.cpp. by @sunliang98 in #6135
• Refactor: replace 3.14 using PI by @YuLiu98 in #6085
• Update DOS codes and related changes with respect to 'magnetzation' variable and default results of 204_NO_KP_NC example, reconstruct esolver by adding a new file pw_setup.cpp in esolver by @mohanchen in #6145

Perf

• Optimize: Compilation time of vdwd3_autoset_xcparam.cpp by @Qianruipku in #6042
• Perf: optimize cal_DMR and folding_HR by @dzzz2001 in #6068

Test

• Test: add test and output for dry run by @YuLiu98 in #6015
• test: fix Dockerfile.intel by @pxlxingliang in #5999
• Reopen unit test for DeePKS multi-k case. by @ErjieWu in #6039
• Tests: remove mock of print func in md/esolver tests by @YuLiu98 in #6073
• changing test 218 by @BariumOxide13716 in #6088
• Test: add test for pw LDOS by @YuLiu98 in #6109

Docs

• Doc: fix the format by @WHUweiqingzhou in #6008
• Update latest Intel oneAPI default compiler for cxx by @Cstandardlib in #6035
• exclude toolchain/install/ in .gitignore after toolchain run by @kluophysics in #6033
• [Toolchain] pack-run and package url by @QuantumMisaka in #6051
• Feature&Docs: support the indent control on FmtTable by @kirk0830 in #6055
• Docs: Correct intro to output in quickstart by @Cstandardlib in #6095
• Refactor&Docs: add annotation beside commented-out lines in VDW module by @kirk0830 in #6117

New Contributors

• @tang070205 made their first contribution in #6014
• @kluophysics made their first contribution in #6033
• @OutisLi made their first contribution in #6103

Full Changelog: v3.9.0.2...v3.9.0.3

LTSv3.10.0

BUG FIX

• Fix: Reopen UT in HContainer output. by @ErjieWu in #5781
• Fix: correct the typo and format error in doc by @kirk0830 in #5788
• Fix 1. get_pchg with nspin=4 and 2. out_mul with DFT+U and 3. fix updates in develop branch by @dyzheng in #5859
• Fix: modify orb info manually by @YuLiu98 in #5853
• Fix: parse_expression for scientific notation by @1041176461 in #5882
• Fix: Resolve compilation issue with Libxc 7.0.0 in ABACUS by @AsTonyshment in #5905
• Fix a bug and a magic number in module_exx_symmetry by @maki49 in #5848
• Fix the wrong symmetry analysis at nspin=2 by @maki49 in #5926
• Fix a bug about hcontainer in exx nscf by @maki49 in #5927
• Fix: fix cmake bug when USE_CUDA=1 and ENABLE_DEEPKS=1 by @dzzz2001 in #5929
• Fix: Fix segmentation fault in integrate test 312_NO_GO_wfc_get_wf by @AsTonyshment in #5970
• Fix : out_mat_dh will lead to different result with MPI-1core with MPI-4core by @xuan112358 in #6018
• Fix: Enhance the warning message when the XC name cannot be recognized. by @sunliang98 in #6025
• Update latest Intel oneAPI default compiler for cxx by @Cstandardlib in #6035
• Fix: Angular momentum quantum number check in reading SOC pseudopot file by @AsTonyshment in #6027
• Fix: directly enter exx loop when init_wfc=="file" by @maki49 in #6019
• Fix: DFT+U force&stress with orbital_corr of some elements are -1 by @dyzheng in #6049
• Fix: add the print header for cusolvermp in scf info by @kirk0830 in #6038
• Fix an output for debug by @maki49 in #6066
• Fix: add a check before calculating EXX force by @maki49 in #6067
• Fixing problem for running meta functionals with kinetic energy density mixed by @BariumOxide13716 in #6071
• Fix: modify warning output by @dzzz2001 in #6074
• Fix: Fix the Segmentation fault caused by zero atom case with LCAO basis. by @sunliang98 in #5821
• Fix: add const in module_lr by @maki49 in #5817
• Fix: support negative value in parse_expression by @1041176461 in #5826
• Fix: dmin is replaced by dmax by @haozhihan in #5829
• Fix a bug and a magic number in module_exx_symmetry by @maki49 in #5848

Others

• Refactor:Move the relaxation part of the ucell function by @A-006 in #5767
• Refactor: Remove global dependence of some functions in DeePKS. by @ErjieWu in #5778
• Refactor: decrease uninitialized member values and the risk of memory leaks in abacus by @A-006 in #5755
• Feature: binary format of backup charge density by @YuLiu98 in #5782
• Refactor: Remove some redundant variables and global dependence in DeePKS. by @ErjieWu in #5791
• [Feature] Add col-major encapsulation for blas kernels and add some lacking kernels from blas by @Critsium-xy in #5790
• Feature: enable init_chg=file for metagga by @YuLiu98 in #5792
• Refactor:Remove update_tau_pos in ucell by @A-006 in #5783
• Feature: add nbands_mul for setting multiply for default nbands and fix initial mag for SOC by @dyzheng in #5946
• test: correct INPUT of examples/spin_polarized/ATOM by @pxlxingliang in #5998
• Perf: optimize the stream strategy in module_gint by @dzzz2001 in #5845
• Perf: modify openmp strategy in module_gint by @dzzz2001 in #5898
• Doc: Enhance ocp and ocp_set documentation for clarity and error prevention by @AsTonyshment in #5896
• Docs: update docs about init_wfc by @YuLiu98 in #5912
• Perf: inline function of complexarray by @dzzz2001 in #5964
• Doc: add OpenMPI usage recommendations and fix doc error by @dzzz2001 in #5965
• Test: fix Dockerfile.intel by @pxlxingliang in #5999
• Doc: fix the format by @WHUweiqingzhou in #6008
• Feature: support the default as the value of dft_functional when initialize vdw by @kirk0830 in #5949
• Optimize: Compilation time of vdwd3_autoset_xcparam.cpp by @Qianruipku in #6042
• Perf: openmp for cal_force_stress by @dzzz2001 in #5956
• Perf: optimize cal_DMR and folding_HR by @dzzz2001 in #6068
• update to version 3.10.0 by @dyzheng in #6077

Full Changelog: v3.9.0...LTSv3.10.0

v3.9.0.2

What's Changed

Feature

• Feature: support the default as the value of dft_functional when initialize vdw by @kirk0830 in #5949
• Feature: Add HR output label for DeePKS. by @ErjieWu in #5968
• Feature: Hybrid functionals for planewave basis by @Flying-dragon-boxing in #5940
• Toolchain 202501 by @QuantumMisaka in #5980

Fix

• Fix: Fix crash in RT-TDDFT current calculation with length gauge when using > 20 MPI processes by @AsTonyshment in #5959
• Fix: Fix segmentation fault in integrate test 312_NO_GO_wfc_get_wf by @AsTonyshment in #5970
• Fix wavefunction output logic of ks pw by @Cstandardlib in #5974
• Fix: BPCG results not stable when using CUDA AWare MPI by @Qianruipku in #5976
• Fix: One segmentation fault related to EXX PW by @Flying-dragon-boxing in #5993

Refactor

• Delete conv_esolver in esolver, and add 'conv_esolver' as a parameter in iter_finish, after_scf, and update_pot functions by @mohanchen in #5941
• Refactor: before_scf of esolver_fp by @YuLiu98 in #5943
• Refactor: remove useless head files by @Qianruipku in #5945
• Refactor:Modify the output functions in elecstate. by @A-006 in #5954
• Refactor: Remove GlobalV in DeePKS and simplify some functions. by @ErjieWu in #5952
• Update after_scf in ESolver by @mohanchen in #5957
• Update ESolver, delete pelec->charge, instead using chr defined in ESolver_fp directly by @mohanchen in #5963
• Some small updates for the format by @mohanchen in #5972
• Refactor: in ESolver_KS_PW, calculate deband in iter_finish, not in hamilt2density by @Flying-dragon-boxing in #5973
• Refactor: decrease memory cost of get_s by @YuLiu98 in #5979
• Update timer, only print out those timers that consume more than 1% of total time by @mohanchen in #5984
• Refactor:Modify Weights and Ebands Functions in the elecstate Module by @A-006 in #5985
• Update some formats in LCAO_operator and ESolver by @mohanchen in #5987
• Refactor: move psi init to before_all_runners by @YuLiu98 in #5992
• [Refactor] Remove my_math.hpp by @Critsium-xy in #6002

Perf

• Perf: openmp for calaulate_HR in deepks_lcao.cpp by @dzzz2001 in #5937
• Perf: openmp for cal_force_stress by @dzzz2001 in #5956
• Perf: inline function of complexarray by @dzzz2001 in #5964
• Performance: Optimize para_gemm and para_linear_transform by @Qianruipku in #5967
• perf the compute of check_atom_stru and add openmp by @A-006 in #5962

Doc

• Doc: add OpenMPI usage recommendations and fix doc error by @dzzz2001 in #5965
• Fix: Fix a LaTeX rendering issue related to RT-TDDFT parameters in online documentation by @AsTonyshment in #5971
• Doc: polish Quick Start part of online doc by @WHUweiqingzhou in #6006

Test

• [Test] Remove pre-commit bot by @Critsium-xy in #5942
• Tests: Add GPU tests for out_wfc_pw & out_wfc_r by @Cstandardlib in #5989
• Tests: one integrate test for exx pw, only for verifying whether exx pw works by @Flying-dragon-boxing in #5994
• test: correct INPUT of examples/spin_polarized/ATOM by @pxlxingliang in #5998

Full Changelog: v3.9.0.1...v3.9.0.2

v3.9.0.1

What's Changed

Feature

• Feature: binary format of backup charge density by @YuLiu98 in #5782
• [Feature] Add col-major encapsulation for blas kernels and add some lacking kernels from blas by @Critsium-xy in #5790
• Feature: enable init_chg=file for metagga by @YuLiu98 in #5792
• [Feature] Add some GPU kernels to blas_connector by @Critsium-xy in #5799
• feature: parallel solve subspace diagonalization in dav_subspace by @pxlxingliang in #5549
• Feature: Support machine-learning based kinetic energy density functional for OFDFT by @sunliang98 in #5777
• [Feature] Complete all kernels' GPU implement in blas_connector.cpp by @Critsium-xy in #5833
• Feature: Add planewave parallization support for BPCG method by @Cstandardlib in #5849
• [Feature] Add vector_mul_vector, vector_div_vector and vector_add_vector in blas_connector and added some GPU tests. by @Critsium-xy in #5858
• [Feature] Add memory functions using AbacusDevice_t by @Critsium-xy in #5861
• Feature: add para_gemm to do parallel matrix multiply by @Qianruipku in #5870
• RT-TDDFT GPU Acceleration: RT-TD now fully support GPU computation by @AsTonyshment in #5773
• Feature: Make BPCG support band parallelism by @Qianruipku in #5873
• Solving linear equations to evolve the wave function in RT-TDDFT. by @ESROAMER in #5925
• First stage of add DSP FFT by @A-006 in #5878

Fix

• Fix: Reopen UT in HContainer output. by @ErjieWu in #5781
• Fix: correct the typo and format error in doc by @kirk0830 in #5788
• Fix: instable nonlocal pp in uspp by @YuLiu98 in #5798
• Fix: optimize lr_spectrum by @maki49 in #5805
• Fix some compile warnings about esolver by @maki49 in #5807
• Fix: Fix the Segmentation fault caused by zero atom case with LCAO basis. by @sunliang98 in #5821
• Fix: add const in module_lr by @maki49 in #5817
• [Fix] Fix the issue that single precision caculation is not runable on CUDA version by @Critsium-xy in #5825
• Fix: support negative value in parse_expression by @1041176461 in #5826
• Fix: dmin is replaced by dmax by @haozhihan in #5829
• Fix module_container test dependency on DCU by @Cstandardlib in #5838
• fix: memory leak when precision=single by @Qianruipku in #5839
• Fix: use gemm instead of einsum in BPCG by @Cstandardlib in #5827
• Fix a bug and a magic number in module_exx_symmetry by @maki49 in #5848
• Fix: wrong band output when kpar > 1 by @Qianruipku in #5847
• Fix: modify orb info manually by @YuLiu98 in #5853
• Fix the precision loss in Tensor output stream operator << by @AsTonyshment in #5866
• Fix:modify cheaktau type and add unittest by @A-006 in #5864
• Fix: Output a warning when the lattice vector is left-handed. by @sunliang98 in #5855
• Fix: Fixed a bug in RT-TDDFT where the electric field was not applied starting from the restart step by @AsTonyshment in #5877
• Fix: parse_expression for scientific notation by @1041176461 in #5882
• Fix: Solve mpi bug for DeePKS. by @ErjieWu in #5886
• Fix: Resolve compilation issue with Libxc 7.0.0 in ABACUS by @AsTonyshment in #5905
• Fix: wrong value of nmaxgr by @Qianruipku in #5906
• Fix: incorrect append mode for out_wfc_lcao by @YuLiu98 in #5914
• Fix: Update documentation for td_lcut parameter in RT-TDDFT by @AsTonyshment in #5917
• Fix: Add the correct RT-TDDFT electric field force acting on ions by @AsTonyshment in #5918
• Fix the wrong symmetry analysis at nspin=2 by @maki49 in #5926
• fix a bug about hcontainer in exx nscf by @maki49 in #5927
• Fix: useless output by @YuLiu98 in #5928
• Fix: fix cmake bug when USE_CUDA=1 and ENABLE_DEEPKS=1 by @dzzz2001 in #5929

Test

• Restore deepks unit test in gamma-only case and make H_V_delta have full size of Hamiltonian in multi-k case by @xuan112358 in #5785
• delete test print by @A-006 in #5795
• Tests: add checks for opening data files by hsolver tests by @Cstandardlib in #5823

Doc

• Doc: Enhance ocp and ocp_set documentation for clarity and error prevention by @AsTonyshment in #5896
• Docs: update docs about init_wfc by @YuLiu98 in #5912
• Update docs related to inputs, scf_thr and ase-abacus by @QuantumMisaka in #5922

Refactor

• Refactor:Move the relaxation part of the ucell function by @A-006 in #5767
• Refactor: Remove global dependence of some functions in DeePKS. by @ErjieWu in #5778
• Refactor: decrease uninitialized member values and the risk of memory leaks in abacus by @A-006 in #5755
• Refactor: Remove some redundant variables and global dependence in DeePKS. by @ErjieWu in #5791
• Refactor:Remove update_tau_pos in ucell by @A-006 in #5783
• Refactor: Enable the support of odd number of electrions in DeepKS orbital by @Liu-RX in #5793
• Refactor: update logic of init_chg by @YuLiu98 in #5801
• Refactor:remove cal_tau from ElecStateLCAO by @A-006 in #5802
• Refactor:Remove Ucell::update_pos_taud by @A-006 in #5794
• Refactor: refactor the constructors of Psi class by @haozhihan in #5761
• Refactor: Remove global dependence of descriptor, orbital_precalc, v_delta_precalc in DeePKS. by @ErjieWu in #5812
• Refactor: modify decimal digit of wfc to 8 by @YuLiu98 in #5820
• Refactor: Remove global dependence in force/stress calculation in DeePKS. by @ErjieWu in #5824
• Refactor: output true external_stress in dp/lj by @YuLiu98 in #5831
• Refactor:remove MPI part funcs of ucell by @A-006 in #5810
• Refactor: Use psi_initializer instead of wavefunc by @Qianruipku in #5775
• Remove TensorMap after cg call in HSolverPW by @Cstandardlib in #5797
• Refactor:Remove the read_atom_species and add read_lattice_constant by @A-006 in #5828
• Refactor: make module_lr/AX a general AO-to-MO transformer by @maki49 in #5834
• Refactor: Remove the global dependence of all remained functions in DeePKS. by @ErjieWu in #5835
• Refactor: Remove LCAO_Deepks from GlobalC. by @ErjieWu in #5844
• Refactor: remove GlobalC::Pkpoint by @Qianruipku in #5846
• Refactor: generalize the transition density matrix in module_lr by @maki49 in #5852
• Update pdm before outputting DeePKS labels by @xuan112358 in #5857
• [Refactor] Rem...