Skip to content

3.4 Brillouin-zone sampling and supercell logic

These are unexecuted teaching inputs and starting models. Original diagrams are schematics, not calculated results. Validate version-specific syntax, licensed or authorized data, numerical convergence and the scientific model before using this workflow.

3.4.1 Model, units and provenance

Keep basis/potential files and executable versions traceable. Grid controls use Ry; common energy/force outputs use hartree and hartree/bohr. Read the unit in each output heading.

Shared inputs, conventions and evidence

Original schematic: Brillouin-zone sampling and supercell logic. No numerical results are claimed.
Original schematic: Brillouin-zone sampling and supercell logic. No numerical results are claimed.

3.4.2 Unexecuted inputs and explicit deltas

Use the accompanying instructions to identify the parent calculation and placement of every delta; a snippet is not automatically a standalone input. Preserve all blank-line and file-provenance requirements.

3.4.2.1 Input block 1

&KPOINTS
  SCHEME MONKHORST-PACK 4 4 4
&END KPOINTS

3.4.3 Worked investigation

Intuition. A k mesh integrates the electronic response of a periodic structure; a supercell changes real-space representation and folds reciprocal space. Gamma-only is a sampling choice, not a universal feature of CP2K. Use silicon B, the diagonalization route from case 11, and omit SMEAR for the initial insulating comparison if occupations permit. K-point support depends on the selected method, functional and build; do not extrapolate this GPW semilocal example to every advanced module.

Input delta. Insert under FORCE_EVAL/DFT, alongside SCF:

See input block 1 above.

Test 2×2×2, 4×4×4 and 6×6×6 meshes at the same conventional cell. Keep the mesh family and centering convention consistent; even/odd grids can differ in whether Gamma is included. Use diagonalization rather than leaving introductory OT active. The listed mesh sizes are starting tests, not a promise of silicon convergence.

Workflow. 1. Verify the actual k points and weights printed by the code. 2. Compare energy per atom and stress at fixed geometry. 3. Evaluate a small volume change at every mesh to test the energy derivative. 4. Construct a real-space supercell and derive the correspondingly folded mesh, then compare equivalent representations. 5. If using symmetry reduction, verify that the structure and chosen electronic state respect the symmetry. 6. Record the exact mesh and shift convention in a report; “dense sampling” is not reproducible.

Interpretation. Energy can appear converged while stress or a small phase-energy difference remains sensitive. A high-symmetry path used for a band plot does not replace a three-dimensional integration mesh. Comparing band extrema requires path coverage and often denser search than a decorative plotted line. This case establishes ground-state integration, not a full quasiparticle band-gap calculation.

Checks, pitfalls and exercise. Increasing k density should be separated from increasing basis or grid. For a slab, sample densely in periodic directions and use one point along the nonperiodic direction only when the chosen representation and solver support that setup. A liquid supercell can often be studied at Gamma, but that must be checked for its size and electronic character. Derive the reciprocal folding for a doubled cell and test whether an apparently cheaper Gamma run is truly equivalent to the reference mesh. Explain remaining differences from geometry, symmetry or numerics.

Diagram. Side-by-side real-space cell doubling and reciprocal-space contraction, with colored points folded onto the reduced zone. EN: real-space replication, reciprocal folding, weighted integration, same physical crystal.

3.4.5 Sources and further reading