Tags: electronic-structure/SIRIUS
Tags
[release] SIRIUS v7.7.0 (#1069) Changelog: - CI/CD is now based on UENVs - improve core generation in LAPW; in particular, MT potential is now augmented with the tail of free atom's potential - robust search of LAPW linearization energies E_{\nu} - constraint total unit cell magnetization for collinear case using two Fermi level approach - fix API to set Hubbard J parameters - fix compilation with clang-19 - fix number of auxiliary mpi grids s for spherical functions symmetrization
[release] SIRIUS v7.6.0 (#1012) Summary of changes: - update main licence to BSD 3-clause and simplify licence boilerplate in the sources - list of authors and contributors - generate header file with API functions for Julia bindings - direct reading of XML UPF-v2 files using pugixml library - cleanup of tests and unit tests - introduce real_occupation_matrix setting to enforce Hubbard matrix to be strictly real - add pseudopotential radial grid cutoff as a parameter instead of a hardcoded 10 a.u. value - add a code to generate smooth initial magnetisation at a slightly higher initialisation cost - 'atom' utility is updated to generate high-accuracy LAPW species - switch to GSL ODE solver for radial equation - 3-rd order radial derivatives for LAPW functions - API to set local and non-local occupancy matrices - API to return major, minor version and revision - issue a warning for non-optimal k-parallelization - remove unused input parameters from the schema and move some input parameters to a more logical location - use strong type for more indices - iterator over split index - save magnetisation in JSON output Fixes: - sirius_add_hubbard_atom_pair() contained a bug - init_atoms_to_grid_idx() is now called only in magnetic case - performance fix for symmetrisation of muffin-tin functions and occupation matrices - bug-fix in IORA o1 radial integrals - sirius_set_nonlocal_occupation_matrix(): count spin index starting from 0 - sirius_get_wave_functions(): handle spin index properly - linear solver for metals now properly handles k and k+q occupancies - compute and print DFT total energy and free energy separately
Update master branch (#948) * [update] bump version; add link to discourse.group (#930) * [refactor] get rid of template in radial_integrals_sum_L3() (#932) * apply @simonpintarelli proposed code change * get rid of one more tempale * default: -> throw, whitespace * fix * apply format --------- Co-authored-by: Simon Pintarelli <simon.pintarelli@cscs.ch> * [fix] subspace initialization in test_lr_solver (#934) * [cmake] remove std::filesystem target (#936) std::filesystem is part of C++17 which is required via CXX_STANDARD, there is no need to check for it explicitly * [fix] compilation error for ~memory_pool+tests (#939) * Ci/rebuild base image (#942) * CI: rebuild base images / pin spack to v0.21 * CI: use anonymous environments instead of dev-build with SPEC * avoid creating partitions with empty intervals in COSTA `custom_layout` --------- Co-authored-by: Anton Kozhevnikov <toxa81@gmail.com> * [fix] truncate very small occupation numbers to 0 (#943) * [refactor] minor cleaups (#938) * minor cleaups * apply format * [cmake] add missing REQUIRED (#945) * [fix] scalar-relativistic radial solver (#941) Introduce a few fixes for the scalar-relativistic radial solver. * [feature] port lr_addusdens to sirius (#937) Use a different sequence of operations to compute G+q augmentation charge. --------- Co-authored-by: Simon Pintarelli <simon.pintarelli@cscs.ch> Co-authored-by: Simon Pintarelli <1237199+simonpintarelli@users.noreply.github.com>
PreviousNext