ANT Home Page

ANT 2020

A Molecular Dynamics Program for Performing Classical and Semiclassical Trajectory Simulations for Electronically Adiabatic and Nonadiabatic Processes for Gas-Phase and Materials Systems

Jingjing Zheng, Zhen Hua Li, Ahren W. Jasper, David A. Bonhommeau, Rosendo Valero, Rubén Meana-Pañeda, Steven L. Mielke, Linyao Zhang, Zoltan Varga, and Donald G. Truhlar

Department of Chemistry and Supercomputing Institute
University of Minnesota, Minneapolis, Minnesota 55455

ANT status

Most recent version: 2020
Date of most recent version: September 15, 2020
Date of most recent manual update: September 15, 2020


ANT ("Adiabatic and Nonadiabatic Trajectories) is a Fortran 90 molecular dynamics program. It has the following capabilities:

  • ANT can be used for dynamics governed either by a single potential energy surface (electronically adiabatic processes) or by two or more coupled potential energy surfaces (electronically nonadiabatic processes).
  • For an electronically adiabatic process, there are two options: (1) the user can supply a potential energy surface as a subroutine or (2) the code can calculate direct dynamics in which energies and gradients are obtained directly from electronic structure calculations carried out with the Gaussian09, Molpro, or MOPAC-mn electronic structure package (which must be obtained separately).
  • For a nonadiabatic process the user must supply two or more surfaces and their couplings in analytic form as subroutines or may employ adiabatic or diabatic input for direct dynamics. Electronically nonadiabatic processes can be treated in either the adiabatic or diabatic representation by a variety of methods including surface hopping by the fewest switches with time uncertainty (FSTU) algorithm, FSTU with stochastic decoherence (FSTU/SD), the Ehrenfest method, and coherent switches with decay of mixing (CSDM). When one uses the electronically adiabatic representation, the user may either provide the adiabatic surfaces and nonadiabatic couplings by direct dynamics, or the program may calculate them from the diabatic surfaces and diabatic couplings, which may either be analytic or direct.
  • The army ants tunneling algorithm is implemented for both electronically adiabatic and electronically nonadiabatic trajectories on unimolecular reactions or any other unimolecular process. For electronically nonadiabatic processes, the army ant tunneling algorithm is only implemented for mean-field methods, e.g., CSDM and SE methods. The tunneling path can be along any of the valence internal coordinates or a combination of two stretch coordinates.
  • ANT can handle reactive trajectories, inelastic collisions, and unimolecular processes. It can calculate cross sections and rate constants.
  • ANT can be run at fixed energy or for thermal ensembles. Collision processes between an atom and a diatom can be carried out by special algorithms with a more advanced treatment of the initial conditions. Another option is that one may begin trajectories at a dividing surface passing through a saddle point. A limited set of final-state analysis options is available.
  • Three methods (TRAPZ, mTRAPZ, and mTRAPZ* methods) are available to ensure zero-point energy maintenance in classical trajectory simulations.
  • The program can handle periodic conditions if a periodic potential is given.
  • The program can also calculate steepest-descents paths.
  • ANT can use general initial conditions for polyatomic bimolecular or unimolceular processes, or it can use accurate WKB initial conditions for atom-diatom or diatom-diatom collisions.

The current version of ANT has a large suite of test runs to assist the user in setting up the program and understanding the input options.

ANT features

One can run ensembles of classical trajectories (also called molecular dynamics) on a single user-supplied potential energy surface or on coupled surfaces; the latter is called electronically nonadiabatic dynamics or non-Born-Oppenheimer dynamics. Several non-Born-Oppenheimer (multi-surface) trajectory methods are available, including:

  • Mean-field methods
    1. Coherent Switches with Decay of Mixing (CSDM) [ J. Chem. Phys, 121 (04), 7658 (2004), J. Chem. Theor. Comput., 1 (05), 527 (2005) ].
    2. Self Consistent Decay of Mixing (SCDM) [ J. Chem. Phys, 120 (04), 5543 (2004) ].
    3. Semiclassical Ehrenfest (SE, aka TDSCF).
  • Surface hopping methods
    1. FSTU with Stochastic Decoherence (FSTU/SD) [ J. Chem. Phys, 127, 194306 (2007)]
    2. Fewest Switches with Time Uncertainty (FSTU) [ J. Chem. Phys, 116 (02), 5424 (2002), Chem. Phys. Lett., 369 (03), 60 (2003) ].
    3. Tully's Fewest Switches (TFS) [ J. Chem. Phys, 93, 1061 (1990) ].
Several prescriptions for preparing the initial conditions are also available:
  • Bimolecular collisions
    1. Quasiclassical state selection.
    2. Quasiclassical thermal ensemble.
    3. Classical thermal ensemble.
    4. User-supplied initial conditions.
  • Unimolecular processes
    1. Thermal ensemble.
    2. User-supplied initial conditions.

Thermal ensembles may be controlled by various thermostats (Berendsen, Anderson, or two-chain Nosé-Hoover) and/or a Berendsen barostat.

There are options for geometry optimization, normal mode analysis, and simulated annealing.

For single-surface dynamics, the user must supply a subroutine that calculates the potential energy surface and its gradient.

For multi-surface dynamics, the user must supply a subroutine that calculates the coupled diabatic potential energy surfaces, their scalar couplings, and the gradients of the surfaces and couplings. However the coupled-surface dynamics may be carried out in either the diabatic or adiabatic representation; in the latter case the program calculates the adiabatic surfaces and their coupling vector due to nuclear momentum by starting with the diabatic information in the user-supplied subroutine. This option is available for an arbitrary number of states.

Sample potential energy surface routines, including gradients, are available in the POTLIB library.

ANT platform compatibility

The ANT code has been tested successfully on the following platforms:




Itasca and Mesabi Linux Clusters

CentOS Linux

Intel FORTRAN Compiler 2018.0.128


Users' Manual

To obtain ANT

We are distributing ANT. If you wish to obtain the program please fill out the online license form at the link below.

Links to other pages of interest

Don Truhlar's Home Page
Computational Chemistry at the University of Minnesota
Department of Chemistry at the University of Minnesota

This document last modified on September 15, 2020
Updated by:  Software Manager