Wiki

Clone wiki

lj-fit.pc-mtp / Fragment-based parametrization

The script requires the presence of E.comb.dat and dimer.ljf from the ab-initio fit, as well as pureliquid.crd and pureliquid.psf and box.size. (see Generate liquid)

:::bash
~/DIR/src/liquid.sims/first.intra.then.dens.charmm.sh file.top file.par \
  file.pdb file.lpun TEMP 'atom_type_sele' M eps_num Rmin_num [processID]
Input arguments
:::text
-file.top: CHARMM-compatible topology file
-file.par: CHARMM-compatible parameter file
-file.pdb: single-molecule pdb
-file.lpun: lpun file (for MTP electrostatics); otherwise must provide "pc"
-TEMP: temperature [K]
-'atom_type_sele': one or two (use quotes) varying atom type(s).
-M: molar mass of one molecule [g/mol]
-eps_num: number of eps incremental steps between simulations 
-Rmin_num: number of Rmin incremental steps between simulations
-processID: analyze run with process ID 'processID'.

The script will read the parameter file file.par and vary the epsilon and Rmin/2 values of the atom types atom_type_sele within a range 0.1 kcal/mol and 1.0 Angstrom of the initial values. The number of parameters that will be scanned is defined by eps_num and Rmin_num. The temperature must be set in accordance to the experimental data available for the density and the heat of vaporization.

For each set of parameters scanned, the root-mean-squared error (RMSE) will be computed against the ab initio energies. Only the parameter sets within 40% of the minimum RMSE will be kept.

CHARMM simulations are generated and submitted to the remote cluster (see Config file). After all gas and NPT simulations have completed, the script computes the density and heat of vaporization for each set.

In case a run has crashed, the existing data can be recovered by reading the processID of the previous run (the process ID can be read from the directories created for the trial simulations).

Once completed, the best parameters can be identified using experimental data, see Extracting sets close to experimental data.

Updated