The Complete Relaxed Scan Algorithm¶
Putting every piece together, a relaxed coordinate scan in Gaussian proceeds as:
- Parse the scan specification. Identify the coordinate to scan (a Z-matrix variable, or a bond/angle/dihedral atom-index specification under
modredundant), its starting value \(q_0\), step size \(\Delta q\), and number of steps \(N\). - Set the scan coordinate to \(q_0\) for the first scan point.
- Freeze the scanned coordinate and run a full Berny geometry optimization over every remaining internal coordinate, starting from the current structure.
- Converge to a constrained stationary point. Confirm convergence, record the energy and geometry, and note the residual force (
-DE/DX) along the frozen coordinate — the Lagrange multiplier of the constraint. - Advance the scan coordinate to \(q_0 + \Delta q\) (and so on for subsequent points), using the just-optimized geometry as a warm-start guess for the next optimization.
- Repeat steps 3–5 for all \(N+1\) scan points.
- Assemble the results into the Summary of Optimized Potential Surface Scan table — the full \((q_n, E_n)\) profile — which can then be plotted directly as the reconstructed energy curve.
The result is a relaxed, one-dimensional cut through the full Potential Energy Surface: cheaper than a true reaction-path calculation, and — for a well-chosen scan coordinate — a close approximation to it.