7.3 A vertical excitation is a response calculation
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.
7.3.1 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
7.3.2 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.
7.3.2.1 Input block 1
&PROPERTIES
&TDDFPT
KERNEL FULL
NSTATES 6
NLUMO -1
MAX_ITER 100
CONVERGENCE [eV] 1.0E-5
RKS_TRIPLETS F
&END TDDFPT
&END PROPERTIES
7.3.3 Worked investigation
Question and intuition. The Kohn–Sham HOMO–LUMO difference is not, in general, an optical excitation energy. A response calculation couples occupied-to-virtual transitions through an interaction kernel. In the Tamm–Dancoff approximation (TDA), the resulting eigenvectors describe mixtures of orbital transitions and the eigenvalues give excitation energies. “Vertical” means that the nuclei remain at their ground-state geometry during the electronic excitation. This is appropriate for an electronic stick spectrum, but it omits vibronic structure and thermal ensemble broadening.
Supported scope. This exercise deliberately uses a neutral, closed-shell, isolated formaldehyde molecule, GPW, Γ-point sampling, and a converged ground-state reference. CP2K 2025.2 documents the modern section under FORCE_EVAL/PROPERTIES/TDDFPT. Its KERNEL FULL refers to the full interaction kernel within this module; it does not turn TDA into the full coupled excitation/de-excitation Casida problem. Do not confuse this route with the older DFT/TDDFPT section, real-time propagation, or the core-specific XAS_TDP module. The 2025.2 TDDFT method guide explains TDA; the 2025.2 TDDFPT reference is the keyword authority. The case does not claim compatibility of every functional, k-point, smearing, spin, or excited-state-gradient combination.
Prerequisites and original input delta. Optimize the molecule consistently and validate the ground-state electron count, multiplicity, and SCF convergence. Use an adequate orbital basis, including diffuse character if the target state requires it. Keep the ground-state DFT settings. Replace GLOBAL/RUN_TYPE with ENERGY, then add the following PROPERTIES section as a sibling of DFT inside FORCE_EVAL. If PROPERTIES already exists, merge its contents. This is valid CP2K section syntax, not a standalone molecular input.
See input block 1 above.
NSTATES is dimensionless; CONVERGENCE is explicitly in eV, avoiding its hartree default. NLUMO −1 asks the module to use all virtual orbitals in the finite basis. This is not an infinite-basis statement. The eigenvalue tolerance controls iterative accuracy, not the physical error of the functional.
Numbered procedure. 1. First obtain and archive a clean ground-state result. Confirm that the response run starts from the intended geometry, charge, and orbital solution. 2. Compute a small group of low states. Inspect response convergence and residual information, rather than accepting a spectrum solely because energies were printed. 3. Read excitation energies, transition moments, oscillator strengths, and reported transition amplitudes. Distinguish the first state from the first strongly absorbing state. 4. Increase NSTATES until the spectral window of interest is covered. Repeating six states with tighter convergence does not reveal a missing higher bright state. 5. Test basis quality and cell size separately. Then compare a semilocal reference with a justified hybrid setup, using Case 28 to control the approximation introduced by ADMM. 6. Broaden sticks externally using a declared Gaussian or Lorentzian width. Preserve sticks beside the smooth curve and identify the width as a visualization/modeling choice.
Interpretation and sanity. Oscillator strengths are dimensionless; energies are commonly printed in eV. A transition can be dark because symmetry or transition-density cancellation makes its electric-dipole coupling weak. An n→π* assignment should be supported by orbital/transition character, not a memorized energy range. Near-degenerate states can exchange order when settings change, so compare character rather than state index. Long-range charge-transfer and Rydberg states are especially sensitive to the functional and basis. Adiabatic semilocal/hybrid TDA does not reliably capture arbitrary double-excitation or strongly multireference physics. Singlet–triplet comparisons must specify which spin sector was requested; RKS_TRIPLETS T does not by itself provide spin–orbit-induced phosphorescence intensities.
Pitfalls and exercise. Do not add an arbitrary orbital energy shift merely to match a desired peak. Do not identify absorption and emission as the same vertical calculation: emission requires an appropriately relaxed excited-state geometry and careful state tracking. As a conceptual comparator, the ETH Zürich ultraviolet exercise discusses excited-state spectroscopy using NWChem; its input is not CP2K syntax. Exercise: design a table that separates iterative, basis, cell, and functional uncertainty. Explain how two states can swap labels while the physical assignment remains continuous. No numerical excitation energy is supplied here because this case has not been run.
7.3.4 Related calculations
- 7.2 From a dipole trajectory to an IR spectrum
- 7.4 PBE0 with ADMM: separate model change from acceleration