Troubleshooting =============== Unknown TOML key or section --------------------------- PhonoMC intentionally rejects aliases and misspelled keys. Move the key into one of the five documented sections and compare it with :doc:`input_files`. Boundary array length error --------------------------- ``boundary_position``, ``boundary_conditions``, and ``boundary_values`` must have the same number of entries. Periodic entries still require a placeholder value of ``0``. Boundary region did not match a facet ------------------------------------- Selectors are relative to the geometry bounding box and match facet centroids. Inspect the mesh bounds and widen the thin selector interval, for example from ``0.0`` to ``[-0.01,0.01]`` around a boundary plane. Periodic boundary validation failed ----------------------------------- Check that: - both regions are marked ``P`` - both boundary values are ``0`` - every periodic region occurs exactly once in ``periodic_pair`` - the pair matches the same number of facets - normals are opposite and areas/edge spectra are compatible Strict POSCAR loading failed ---------------------------- The selected ``material_folder`` must contain a readable POSCAR. PhonoMC no longer substitutes a unit-cell volume. Fix the path or POSCAR contents rather than bypassing the error. HDF5 loading failed ------------------- Verify that ``kappa-fbz.hdf5`` or ``kappa.hdf5`` exists and contains the required datasets with compatible dimensions. Use an absolute ``material_folder`` temporarily when diagnosing relative paths. ``qdelaunay`` is missing ------------------------ Install Qhull command-line tools: .. code-block:: bash sudo apt install qhull-bin OpenMP was not found -------------------- The project still builds serially. Install the compiler OpenMP runtime and reconfigure from a clean build directory. On Ubuntu: .. code-block:: bash sudo apt install libomp-dev Particle balance drifts ----------------------- Inspect ``absorbed``, ``injected``, ``recovered``, and ``net`` columns. Confirm periodic pairing, reservoir selection, time step, and mesh closure before increasing particle count. The 3D run is too slow ---------------------- First reduce ``particle_count``, ``iterations``, and ``grid_xyz``. Confirm the geometry and source region using a smoke test, then scale one dimension at a time and enable OpenMP.