Donald L. Thompson's Theoretical Chemistry Group
Department of Chemistry, Schlundt Hall, University of Missouri - Columbia


Potential Energy Surface Fitting
Codes coming soon...

  • We have finished code development for a series of IMLS fitting methods that we plan to offer as free downloads.
  • We have recently developed a high-order local version of IMLS in which stored basis set expansions allow efficient use of high-order fits.1,2
  • Two main types of codes will be available
  • Spectroscopy/ dynamics (coming soon!!): we have automatic surface generators that will generate global potentials for 3, 4, and 5+ atom systems to a predefined accuracy in a user specified energy range.  The automatic PES generator automatically calls a prefered electronic structure code (e.g. Gaussian) at a minimal number of configurations, making use of 1) value, 2) value and gradient, or 3) value gradient and Hessian data depending on the ab initio method chosen. In a recent test we were able to reproduce a spectroscopic quality surface for methane (9-D) at the CCSD(T) level in less than 1 week on a single processor. We have also developed a parallel code using MPI. 
  • Trajectories (still in development): We are developing a PES interface for classical trajectories codes that will construct an accurate and efficient ab initio PES "on the fly".  Our IMLS PES fit with automatic surface generation will dynamically determine whether or not the trajectory is in a region that has been fit to sufficient accuracy.  Ab initio data will be added where needed and IMLS fitted energies will be determined where suitable.
Automatic PES generation for HOOH (6-D) required only 1400 points with value and gradient to acheive 0.1 kcal/mol RMS accuracy over a 100 kcal energy range. Numbers in parentheses refer to the HDMR basis set described in ref 1.

 

(1)  Dawes R, Thompson D.L, Guo Y, Wagner A.F, Minkoff M.
Interpolating moving least-squares methods for fitting potential energy surfaces: Computing high-density potential energy surface data from low-density ab initio data points. J. Chem. Phys. 126:184108 (2007)

(2)
Dawes R, Thompson D.L, Wagner A.F, Minkoff M.
Interpolating moving least-squares methods for fitting potential energy surfaces: A strategy for efficient optimal data point placement in high dimensions. (in prep).


Return to D.L. Thompson's Home Page