Legacy .inp Input

The sectioned .inp format remains supported for existing input decks and controls that do not have a concise spelling. New calculations should begin with the recommended .oqp format or the Python API; workflow pages show those forms before their legacy .inp equivalents.

A compact .oqp file describes the same calculation in one readable line:

mrsf/bhhlyp/6-31g*
opt
geom="h2o.xyz"

This means an MRSF-TDDFT optimization of S0; OpenQP selects the required working reference automatically. Start with the .oqp Quick Start for state labels, SOC counts, and more examples.

Legacy .inp files use the sectioned format documented on this page and remain supported unchanged. Keep sectioned syntax in .inp and compact syntax in .oqp; changing formats is optional. The correction assistant is a secondary aid that produces an inspectable .resolved.oqp file, while the resolved canonical file remains the authoritative calculation record. OpenQP renders the short positional geometry above as the explicit canonical spelling geom="h2o.xyz".

The OpenQP repository ships a same-stem .oqp companion for every legacy example .inp. This makes the complete legacy example inventory available in both formats without removing the established input system.

The two formats intentionally expose different optimization detail. Concise .oqp geometry drivers always use the native OpenQP engine and have no lib selector. Traditional .inp files retain [optimize] lib=oqp, lib=geometric, or lib=scipy for compatibility; geomeTRIC is an optional legacy dependency used chiefly for advanced constraints beyond native frozen distances.

Legacy .inp files are INI-like text files. Options are grouped by section:

[input]
runtype=energy
method=hf
basis=6-31g*

[scf]
type=rhf

Lines beginning with # are comments. Keyword names are case-insensitive in normal use, but this manual uses lower-case names to match the Python schema.

Geometry

Inline coordinates are written under [input] system with indented atom lines:

[input]
system=
   O   0.000000000   0.000000000  -0.041061554
   H  -0.533194329   0.533194329  -0.614469223
   H   0.533194329  -0.533194329  -0.614469223

An external XYZ file can be used instead:

[input]
system=h2o.xyz

Some workflows, such as NACME, also use [input] system2 for the displaced or previous geometry.

Core Sections

Section	Purpose
[input]	Charge, basis, method, run type, geometry, AO convention, threading.
[d4]	Complete explicit DFT-D4 rational-damping parameter set.
[guess]	Initial orbitals and restart data.
[scf]	RHF/ROHF/UHF reference and SCF convergence controls.
[mp2]	Standalone MP2 spin-scaling controls.
[dftgrid]	DFT functional/grid controls.
[tdhf]	TDHF, TDDFT, SF-TDDFT, MRSF-TDDFT, and UMRSF settings.
[dftb]	DFTB backend, SCC, response, and MRSF-TDDFTB controls.
[md]	Nonadiabatic-dynamics controls used by runtype=namd.
[qmmm]	OpenMM QM/MM system and molecular-dynamics controls.
[properties]	Gradients, NAC, NMR, export, and property requests.
[optimize]	Geometry target and convergence controls; backend selection is retained for traditional .inp and Python compatibility.
[oqp]	Native optimizer, TS-Hessian, IRC, MEP, and NEB controls.
[geometric]	Optional legacy geomeTRIC controls for traditional .inp workflows.
[pcm]	Reference-SCF PCM/ddX energy settings.
[symmetry]	Point-group metadata and optional symmetry reductions.
[hess]	Hessian and frequency controls.
[nac]	NAC/NACME controls.
[ekt]	MRSF-EKT IP/EA channel selection.
[neb]	NEB product/image controls plus optional legacy geomeTRIC compatibility keys.
[json]	Advanced JSON/restart metadata.
[tests]	Internal regression-test expectations.

Run Types

Common [input] runtype values:

Run type	Meaning
energy	Single-point energy and requested properties.
grad	Energy plus gradient for the requested state.
hess	Hessian/frequency workflow.
nac, bp	Numerical nonadiabatic-coupling vector and branching-plane workflows.
nacme	Time/geometric derivative coupling between MRSF states.
soc	MRSF-TDDFT, MRSF-TDHF, or MRSF-TDDFTB spin-orbit coupling workflow.
ekt	MRSF-EKT ionization-potential/electron-affinity workflow.
md	Ground-state QM/MM molecular dynamics. The command-line runner dispatches this OpenMM path specially.
namd	Nonadiabatic molecular dynamics using [md] controls.
optimize	Geometry optimization.
meci, mecp	Crossing-point searches. meci_search=baeka selects the two-or-more-state adaptive MECI algorithm.
tci	Existing three-state adaptive-penalty workflow, retained for backward compatibility. It is distinct from the new general meci_search=baeka option.
ts, irc, neb, mep	Reaction-path workflows.
prop, data	Multi-state property/gradient workflows for downstream drivers.

For QM/MM MD, use [qmmm] n_steps=N. The older [qmmm] nsteps=N keyword remains available for legacy bookkeeping. Canonical qmmm(...) may accompany energy, md, or namd; md requires it, while namd may also run gas phase. Canonical QM/MM gradients and optimizations are rejected until their active backends provide the required assembled gradient.

Bare runtype=md without QM/MM remains invalid. With qmmm_flag=true, both the command-line path and programmatic Runner dispatch ground-state MD to the OpenMM QMMM_MD driver. This applies after a concise .oqp request has been lowered as well as to a traditional sectioned .inp. Run it without MPI, for example openqp file.oqp --nompi or openqp file.inp --nompi.

Standalone MP2 is selected with [input] method=mp2, uses only runtype=energy, and requires an empty [input] functional. Spin-scaled MP2 variants are controlled by the optional [mp2] section.