Convergence

Overview

Optking utilizes a number of optimization presets which mirror and/or mimic the optmization behavior from a number of popular Quantum Chemistry packages. These may be selected with the G_CONVERGENCE keyword. Descriptions of each preset may be found below. See Notes [#fe] and [#ff] for clarification on what combinations of criteria are required or allowed.

For greater control one or more ctriteria can be selectively activated through use of the *<CRITERIA>_G_CONVERGENCE* keywords. In order to modify a preset both *FLEXIBLE_G_CONVERGENCE* and one or more *<CRITERIA>_G_CONVERGENCE* keywords must be selected in addition to the preset. Specifying *<CRITERIA>_G_CONVERGENCE* without *FLEXIBLE_G_CONVERGENCE* will cause the preset to be discarded and optking will ONLY consider the *<CRITERIA>_G_CONVERGENCE* keyword for convergence.

As an example the first set of options only changes the rms_force threshold. The second changes from QCHEM to GAU_TIGHT while loosening the rms_force threshold:

{"g_convergence": "gau_tight", "rms_force_g_convergence": 3e-5}
{"g_convergence": "gau_tight", "flexible_g_convergence": True, "rms_force_g_convergence": 3e-5}

Summary of convergence criteria for g_convergence defaults in optking (Same as in Psi4)

g_convergence	Δ E	Max Force	RMS Force	Max Disp	RMS Disp
NWCHEM_LOOSE [4]		\(4.5 \times 10^{-3}\)	\(3.0 \times 10^{-3}\)	\(5.4 \times 10^{-3}\)	\(3.6 \times 10^{-3}\)
GAU_LOOSE [6]		\(2.5 \times 10^{-3}\)	\(1.7 \times 10^{-3}\)	\(1.0 \times 10^{-2}\)	\(6.7 \times 10^{-3}\)
TURBOMOLE [4]	\(1.0 \times 10^{-6}\)	\(1.0 \times 10^{-3}\)	\(5.0 \times 10^{-4}\)	\(1.0 \times 10^{-3}\)	\(5.0 \times 10^{-4}\)
GAU [3] [6]		\(4.5 \times 10^{-4}\)	\(3.0 \times 10^{-4}\)	\(1.8 \times 10^{-3}\)	\(1.2 \times 10^{-3}\)
CFOUR [4]			\(1.0 \times 10^{-4}\)
QCHEM [1] [5]	\(1.0 \times 10^{-6}\)	\(3.0 \times 10^{-4}\)		\(1.2 \times 10^{-3}\)
MOLPRO [2] [5]	\(1.0 \times 10^{-6}\)	\(3.0 \times 10^{-4}\)		\(3.0 \times 10^{-4}\)
GAU_TIGHT [3] [6]		\(1.5 \times 10^{-5}\)	\(1.0 \times 10^{-5}\)	\(6.0 \times 10^{-5}\)	\(4.0 \times 10^{-5}\)
GAU_VERYTIGHT [6]		\(2.0 \times 10^{-6}\)	\(1.0 \times 10^{-6}\)	\(6.0 \times 10^{-6}\)	\(4.0 \times 10^{-6}\)

Footnotes

[1]

Default

[2]

Baker convergence criteria are the same.

[3] (1,2)

Counterpart NWCHEM convergence criteria are the same.

[4] (1,2,3)

Convergence achieved when all active criteria are fulfilled.

[5] (1,2)

Convergence achieved when Max Force and one of Max Energy or Max Disp are fulfilled.

[6] (1,2,3,4)

Normal convergence achieved when all four criteria (Max Force, RMS Force, Max Disp, and RMS Disp) are fulfilled. To help with flat potential surfaces, alternate convergence achieved when 100\(\times\)RMS Force is less than RMS Force criterion.

IRC Convergence

The IRC algorithm uses slightly different convergence criteria since the step sizes are of a fixed distance. The optimization ends when the forces are close to anti-parallel with the forces of the previous step. The threshold for checking the overlap of the forces can be controlled via irc_convergence. Alternatively an increase in energy along the MEP with any negative overlap of the forces is sufficient.

Individual points on the IRC are optimized in a constrained optimization (on a hypersphere of fixed radius) according to the convergence criteria of the table above.