New Features

• TDDFPT: Add exciton descriptors (#3847)
• Efield implementation for xTB (#3883)
• Atom code: Add option to write xmgrace wavefunction file (#3962)
• GFN-xTB (#4005, #4190, #4209, #4242)
• Read/write external energy derivatives from/to trexio files (#4009, #4074)
• Pseudopotentials: Add 5f-in-core for trivalent and tetravalent actinides (#4068)
• Pseudopotentials: Add ccECP (#3940)
• RTBSE : Padé FT Refinement (#4115)
• HP-DFT modules and regtests (#4138)
• Add option to print space groups (#4271)
• Add SIRIUS DFTD3 and DFTD4 support (#4277)
• Atomic polarization tensors via numerical differentiation (#4287)
• RI-HFXk: Various improvements (#4291)
• Resonant Inelastic X-ray Spectroscopy (RIXS) Module (#4315)

New Libraries

• Improve DLA-Future integration (#4169, #4269)
• Upgrade to DeePMD 3.1.0 and switch to PyTorch backend (#3893, #4310)
• Make GRPP an internal dependency (#3966)
• Interface for greenX library (#4078)
• Add ace support (#4182)

Breaking Changes

• Remove old TDDFPT code (#4066)
• Restore old format for writing forces to .xyz files (#4294)
• RTBSE: Input structure changed (#3918)

Fixes

• Fix occurrence of ghost states with SVDs (#3911)
• Speedup loading of NequIP and Allegro models (#3912)
• TDDFPT%MGRID bug fix for GAPW (#3967)
• Fix typo in data/BASIS_ccGRB_UZH (#4017)

New Features

• BSE: Optical spectra, BSE@evGW(0), and lots more (#3628, #3659, #3781, #3793, #3815)
• Real time Bethe-Salpeter Propagation (#3691)
• Harris and EHT methods incl. LS solver (#3665, #3780, #3790)
• Print wannier states coefficients in AO basis (#3683, #3687)
• Add Z-matrix formalism and EVERY_N_STEP keyword for linear response (#3689, #3692)
• RPA based SIGMA Functionals (#3695)
• Framework to calculate response forces for external energy expressions based on KS orbitals (#3721)
• PAO: Add prediction from equivariant PyTorch models (#3738)
• gfn0-xTB and parallel DFT-D4 (#3765, #3679, #3685)
• Smeared occupation TDA (#3829)
• GW: Add keywords SIZE_LATTICE_SUM and KPOINTS_W (#3833)

New Libraries

• Add SMEAGOL for electron transport with NEGF (#3716)
• Add cuSOLVERMp generalized eigensolver (#3787)
• Add DLA-Future generalized and complex eigensolvers (#3799, #3813, #3819)
• Add TREXIO file format writer (#3792)

Breaking Changes

• Weighting of RSMD colvar is modified for the case of subsystem = list (#3818)

Fixes

• Re-init FFT scratch in case of PW env changes (#3661)
• Add asserts after malloc and realloc in DBM and grid code (#3706)
• Fix segmentation fault when performing CDFT-MD with many constraints (#3711)
• Fix thread-safety in fft_tools (#3729)
• Fix access to unallocated arrays in pme, spme, and ewalds (#3733)
• Fix handling of empty file names in the input (#3758)
• Fix bug in TDDFPT/Davidson restart (#3821)
• Fix bug in RTP/EMD restart (#3637)
• Add initialization step to i-Pi which some clients expect (#3747)

This is a minor release to fix an issue with the PW environment that can lead to stalls during MD (#3661).

Since this only affects MPI jobs, the ssmp binaries from the previous 2024.2 release are still up-to-date.

New Features

• Many new pages in the methods section of the manual
• ECP nuclear gradients (#3210)
• New basis sets (#3266, #3276 #3460, #3561)
• TDA Kernel methods (#3273)
• Bethe Salpeter equation for molecules (#3308, #3329)
• Stress prediction in NequIP & Allegro (#3428, #3445)
• GW for small unit cells with full k-points (#3448, #3505, #3513)
• Hedin shift for G0W0 (#3533)
• i-PI server functionality (#3420)
• Infrastructure for Kpoint symmetries (#3482)

New Libraries

• Machine learning with the DeePMD-kit (#3145)
• Dispersion correction with the DFTD4 library (#3501)
• GPU support via OpenCL (#3321, #3315, #3375)

Breaking Changes

• Increase ScaLAPACK default block size to 64 (#3184)
• Remove BROYDEN_MIXING_NEW option (#3346)
• Remove KP_RI_EXTENSION_FACTOR" keyword (#3223)
• Mark support for QUIP and PEXSI as deprecated (#3600)

Fixes

• Fix oscillator strength for TDDFT+SOC within length and velocity representation (#3201)
• External potential: Do not re-parse potential function at each grid-point (#3204)
• PWDFT: Print the band gap and total energy (#3211)
• Fix bug #3218 in screening of hfx derivatives (#3221)
• Fix bug #3217 in printout of eigenvalues and eigenvectors (#3230)
• Improve element comparison in reftraj (#3337)
• Allegro: Fix parallelization with virials (#3445)
• FFTW: Fix import/export wisdom file (#3492, #3496)

New Features

• Docs: Launch Sphinx-based manual (#2883)
• TDDFPT: Pseudopotentials, GAPW, GPW, and forces (#2895, #2932 , #2940, #3011, #3110, #3122)
• RI-HFX for K-points (with gradients and ADMM) (#2998)
• ADMM: input short cuts (#3118)
• Long-range quantum computing (WF) in short-range DFT embedding (#2924)
• Active space: Implement ERI calculation using half-transformed integrals (#3082)
• GW: open-shell periodic GW (#2920, #3045, #3125)
• G0W0/SOC: bandstructure, PDOS, local bandgap, periodic (#2994, #3130)
• NNP: Helium-Solute for interaction (#3043)
• Time-dependent Field for EMD (#3081)

New Libraries

• Add experimental support for DLA-Future eigensolver (#3143)
• Enabling calculations with ECPs libgrpp library (#3147)

Breaking Changes

• Remove SINGLE_PRECISION_MATRICES keyword (#3096, #3140)
• Abort run by default on SCF convergence failure (#3148)
• Drop support for NDEBUG (#3172)
• Production docker files moved to new repository (#3083)
• MD: Refactor REFTRAJ / EVAL_ENERGY_FORCES keyword (#2981)
• Drop CMake option CP2K_BUILD_DBCSR (#3044)

Fixes

• Correct LnPP2 Basis sets. Some typos corrected Jun-Bo Lu (#3100, #3102)
• Fix Wannier localization when using LOW SPIN ROKS (#3108)
• XAS_TDP: Fix bug leading to crashes when n_ranks>>>nex_atom (#2908)
• EMD + Time Dependent Electric Field (#3016)
• Fix parsing atom sites from CIF files (#3092)
• Fix parsing of CHARMM General Force Fields (#2956)

• GW: Periodic open-shell and Splitting of electronic states due to spin-orbit coupling (#2639, #2831)

• GTH pseudopotential database file with spin-orbit coupling (SOC) parameters added (#2848)

• RTP: TD Field Velocity gauge and projection TD-MOs (#2623, #2744)

• RTP: Linear density delta kick and restart (#2543)

• RTP: Enabled ADMM with GAPW (#2729)

• Implementation of the NVPT for APTs and AATs in velocity form (#2568, #2561)

• Intrinsic Atomic Orbitals (#2707)

• Machine Learning: Add PyTorch interface, Nequip and Allegro models (#2420, #2528, #2722)

• k-points: Implementation of the DIIS/Diag. solver (#2721)

• TDDFPT: SOC absorption (#2859)

• GAPW triplet excitation energies and forces (#2837, #2861)

• EC: Enable DC-DFT with HFX-ADMM for reference and DC calculation (#2780)

• Add cell symmetry HEXAGONAL_GAMMA_120 (#2758)

• Grid: Rename backends, change default to CPU (#2772, #2775, #2778)

• Grid: Enable GPU acceleration for large basis sets (#2787, #2793)

• FM: Add experimental support for NVIDIA cuSOLVERMp (#2860)

• Regtesting: Add --smoketest option (#2501)

• Add support for MPI Fortran 2008 bindings (#2486)

• Add support for Apptainer/Singularity containers (README)

• Add gradients for SOS-MP2 and RPA incl. benchmarks (#2208,#2271,#2473)
• TDDFT/Linear Response: Add GAPW/GAPW_XC and ADMM/GAPW options (#2200)
• TDDFT: Add excited state forces as property (#2363)
• RI-RPA: Allow for XC correction in ADMM RI-RPA (#2216)
• RTP: Velocity gauge and magnetic delta pulse (#2343)
• GW: Automatically extrapolate k-point mesh (#2229)
• xTB: Add vdW options (#2431)
• xTB: Fix electronic energy dependence on EPS_DEFAULT (#2287)
• Vibrational analysis: Raman Intensities (#2263)
• New pseudopotentials and basis sets (#2472, #2193)
• Improve NewtonX interface (#2443)
• Fist: Add LAMMPS style tabulated pair potentials (#2313)
• EC: Variational Density-Corrected DFT (DC-DFT) (#2322)
• Update active space interface (#2346)
• Helium: Add missing xyz output format (#2432)
• SIRIUS: Add support for libvdwxc (#2270)
• ELPA: Fix block size issue on GPU (#2407)
• Drop Support for MPI 2.0 (#2438)
• Add experimental CMake build system (#2364)
• Fix regtests on ARM64 (#1855)
• Start testing with Address Sanitizer (#2306)
• Start testing on macOS Apple M1 (sponsored by MacStadium)

This is a minor release to fix the outdated url for Spglib in the toolchain (#2262).

Since the actual source code has not changed the binaries from the previous 2022.1 release are still up-to-date.

• Migrate tensor operations to new sparse matrix library DBM (#1863)
• Add HIP support for PW (#1864)
• Drop support for GCC 5 (#1878)
• Add GAPW Voronoi integration (#1919)
• Remove deprecated sections LIBXC and KE_LIBXC (#1921)
• Add LibXC equivalents to ADMM exchange potentials (#1972)
• Improve support for metaGGA functionals (#1974)
• Use SPLA for offloading dgemm on GPUs in the mp2 module (#1951)
• TDDFT: enable state following using transition charge finger print (#1991)
• Add barostat for frozen atoms in absolute coordinate (#2000)
• Fix linkage of COSMA (#2021)
• Migrate to centralized __OFFLOAD_CUDA/HIP flags (#2027)
• Add low-scaling SOS-Laplace MP2 forces (#2031)
• Refactoring of basis set optimization code (#2068)
• Add k-points for the GW self-energy (#2073)
• CDFT: forces based on Hirshfeld partitioning (#2111)
• RPA: Add low-scaling gradients (#2131)
• MP2: Add more solvers (#2142)
• GW: Add 4-center Hartree-Fock and ADMM for exchange self-energy (#2145)
• Print vibrational modes for Newton-X (#2146)
• Add partially occupied Wannier states (#2154)
• Voronoi integration: Mitigated issues with symmetric structures, more diagnostic output (#2171)
• Add GAPW_XC for TDDFPT energies (#2178)

• Fix MacOS build (#1316)
• Add NEWTONX interface (#1794)
• Add Gromacs QM/MM support (see also)
• Add experimental support for HIP and OpenCL to DBCSR
• Adopt BSD3 license for new performance critical code (#1632)
• Add GAL21 forcefield (#1579)
• Upgrade to MPI_THREAD_SERIALIZED (#1564)
• Add new pseudopotentials and basis sets (#1547, #1551)
• Add analytical derivatives of the MO coefficients wrt nuclear coordinates (#1706)
• Add forces for RI-HFX (#1688)
• Add forces for TDDFT (#1670, #1759)
• Regularized RI for periodic GW (#1776)
• Add beadwise constraints to PINT (#1734)
• Add analytical stress tensor for NNP (#1783)
• Add ghost particles and tip scan for xTB (#1578)
• Add forces and stress tensor for MP2-based double-hybrids (#1647)
• Rewrite regtesting script in Python, arguments changed slightly (#1548)