History#

2026.8.7 – Bugfix: DDEC6 now works for charged systems (ions)

  • DDEC6 via Chargemol previously failed for any charged structure (an ion, or a charged BSSE fragment): ORCA’s orca_2aim always writes <Net Charge> 0.0 in the .wfx it produces, regardless of the molecule’s actual charge – confirmed on ORCA 6.1.1 for both a cation (Na+) and an anion (Cl-), even though the wfx’s own electron count is correct. Chargemol cross-checks the two and refuses to run, reporting “the quantum chemistry program you used to generate the wfx file contains a bug.” This step now corrects the <Net Charge> field to the structure’s actual charge (already known from the configuration) before handing the file to Chargemol, working around the ORCA bug rather than waiting on a fix from the vendor.

  • Right-clicking the node in the graphical flowchart editor did nothing – no Edit/Delete menu – because the Tk node never overrode right_click to actually post the popup menu (the base class only builds it; a subclass must call tk_popup itself). Added the same right_click override the cookiecutter template and other steps already use.

2026.7.14 – Control Chargemol’s core (OpenMP thread) count

  • The ncores option in the [atomic-charges-step] configuration is now wired to Chargemol via OMP_NUM_THREADS (it was previously documented but had no effect). ‘available’ (default) uses all cores the job/machine provides; an integer caps it. This matters most when running directly on a machine (no queuing system), where Chargemol’s OpenMP would otherwise use every core on the node.

  • The output now reports the number of threads used, and a line at the end of the step with the wall-clock time in seconds.

2026.7.13.1 – Bugfix: clearer Chargemol failures; skip incomplete densities

  • When Chargemol fails to produce charges, the error now quotes its real log (<input>.output) instead of the empty stdout.txt – so causes like “Could not find a suitable reference density” are shown directly. Chargemol writes its diagnostics to that log, not to stdout.

  • The atomic-densities directory is now used only if it actually contains the DDEC6 reference densities (c2_*.txt); an empty or incomplete directory is skipped in favor of the complete set bundled in the seamm-chargemol conda environment. This avoids a silent Chargemol failure when the configured directory exists but is not populated.

  • The bundled densities are now located by asking conda for the environment’s actual prefix, not by deriving it from the conda executable’s path. This fixes non-standard/HPC conda layouts where the derivation was wrong, so the densities silently were not found and DDEC6 failed.

2026.7.13 – Option to set the charges on the structure

  • New Set as the atomic charges on the structure option (default off). When on, the computed charges are also written to the structure’s standard per-atom charge attribute (in addition to the labeled charges_<label> column), so they travel with the structure – e.g. to write them to an extended-XYZ (extxyz) file for machine-learning training.

2026.6.29 – Run Chargemol from a dedicated conda environment

  • The installer now creates a ‘seamm-chargemol’ conda environment containing Chargemol (for DDEC6), and the step runs Chargemol in that environment – no hand-built installation or PATH setup is needed. Run ‘atomic-charges-step-installer install’ to set it up.

  • The reference atomic densities are found automatically inside the seamm-chargemol environment; the ‘DDEC reference densities’ setting is now needed only to point at a different copy.

2026.6.28 – Citations, charge normalization, and GUI fixes

  • The DDEC6 (Manz & Limas) and Bader (Henkelman) methodology papers are now cited for the charge method that is run.

  • The charges are normalized to the known net charge of the system with a small, reported uniform shift; this can be turned off.

  • Accepts an analytic wavefunction (.wfx) from a preceding ORCA step, in addition to Gaussian, for molecular DDEC6.

  • Fixed: the Results tab in the GUI was empty and selected results were not saved. The atomic charges, net charge, residual, and method are now listed and stored.

  • The charge method now offers Bader only when a density grid is available (a periodic/VASP density or explicit files); molecular densities offer DDEC6. Bader reports a clear message that it awaits periodic support.

2026.6.27 – Initial release of the Atomic Charges step

  • Computes DDEC6 atomic charges (via Chargemol) from a molecular Gaussian wavefunction (.wfx), storing them as a labeled charge set on the structure.