2.3 SCF convergence is an inner loop

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.

2.3.1 Model, units and provenance

Geometry in Å; electronic energy in hartree; vibrational wavenumbers in cm⁻¹. Check each printed field and keep thermal and standard-state terms distinct.

Shared inputs, conventions and evidence

Original schematic: SCF convergence is an inner loop. No numerical results are claimed.

2.3.2 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.

2.3.2.1 Input block 1

%chk=g07_dimer_scf.chk
%mem=2GB
%nprocshared=2
#p B3LYP/6-31+G(d) SP SCF=(Tight,MaxCycle=128) Int=UltraFine

Water dimer SCF loop baseline

0 1
O 0.000000 0.000000 0.000000
H 0.958000 0.000000 0.000000
H -0.240000 0.928000 0.000000
O 2.850000 0.000000 0.000000
H 3.090000 0.928000 0.000000
H 3.090000 -0.464000 0.804000

2.3.3 Worked investigation

2.3.3.1 Intuition and prerequisites

In a self-consistent-field procedure, the density creates an effective field and the orbitals obtained in that field create a new density. Convergence asks whether this loop has settled, not whether the molecule is in equilibrium. This case uses a water dimer as a slightly larger density problem, with a deliberately approximate geometry. You will compare a conventional SCF attempt with the fallback strategy XQC, while keeping the chemistry unchanged. There is no promise that this small system will fail conventionally; learning to handle an ordinary success is part of the exercise.

See input block 1 above.

2.3.3.2 Workflow

Inspect the geometry for atomic overlaps and a plausible donor–acceptor arrangement. Confirm twenty electrons.
Run the baseline and note iteration count, convergence diagnostics, and final energy. Do not delete a failed log.
Make an independent input replacing the SCF clause with SCF=(Tight,XQC,MaxCycle=128) and use a new checkpoint.
Compare final states, not only iteration counts. If orbital occupations, spin, or density character differ, the calculations may have found different stationary solutions.
If convergence is genuinely difficult, first check geometry, charge, basis linear dependence, and initial state. Change one diagnostic intervention at a time and preserve the reason in the run ledger.

2.3.3.3 Interpret check and limits

XQC is a fallback strategy, not an instruction to always run a different physical method. A larger cycle limit cannot cure an impossible charge state, wrong electronic root, or unreasonable geometry. Tight thresholds are numerical controls and should be tested against the property of interest. A successful SCF is still subject to wavefunction stability analysis in difficult open-shell or stretched-bond problems. Do not claim the XQC run is scientifically better simply because it converges faster; identical converged solutions should agree within the chosen numerical tolerance.

Exercise Write a diagnosis tree with separate branches for malformed input, impossible electron parity, oscillating SCF, stable SCF but failed optimization, and scheduler termination.