lj2 documentation - pgcmc.x

pgcmc.x

This page describes the use and functions of the pgcmc.x parallel Grand Canonical Monte Carlo program.

This program has just been rewritten to function with N-component mixtures and new input format files. Not all of the bugs have been chased down yet. Refer to the gcmc.x page for use instructions, the two programs are very similar.

Caveats:
1. Although there is code here for the use of long-ranged corrections to the truncated Lennard-Jones potential, it has never been used for production work and is almost certainly wrong.

2. This program only really works for relatively big systems (20 diameters across, at least) and will frequently crash for smaller ones. Of course, there's no point in parallelizing the algorithm for small simulation cells.

Control file:

p-gcmc.x reads in a file called pgcmc.control which contains run-length and simulation-control information, the name of the starting configuration, etc. An example of this file is shown here: 

'output/pgcmc.output'      output filename beginnings
'conf/l20.empty.config'   input configuration filename
100  2000  10             output freq., dump freq., stfreq
F  F                      Lrecords?, Longformat?
0                         Widom insertion test freq.
3000                      total number of steps
00                        length of "equilibration"
1			  PMfreq (passmass counter)
1			  Iperf (performance, 0=none)
.false.                   Lsequen  (sequential move choices)
1000    0.10              RF(displacement), delta-max
0.50    150000            TAratio (target acc. ratio), nadjust
0   2   0.030             RF(HMC move), L(int), dtime
0                         RFKawasaki
2			  Nspecies (number of species)
1000   1000               RFcd(1), RFcd(2), .... (create/destroy)
0.915 -4.00  4.00         T, mu(1), mu(2)....
0.915 -3.50  3.50         T, mu(1), mu(2)....
This file contains the following information:
  1. character outfile -- the prefix used for the output filenames.
  2. character configfile -- the starting configuration filename.
  3. integer integer integer Ofreq Dfreq STfreq -- parameters which control the frequency of statistics output. See the md.x program description for these.
  4. logical logical Lrecords Llongrec -- Whether or not to output unformatted record files, and if so, whether or not to use the "long" format.
  5. integer mufreq -- Widom insertion test frequency, for chemical potential calculation in the canonical ensemble.
  6. integer Nsteps -- The total number of simulation cycles to run.
  7. integer Lequil -- The number of cycles in the "equilibration" phase.
  8. integer PMfreq -- The frequency (in cycles) with which to call "passmass", which re-shuffles particles between neighboring processors.
  9. integer Iperf -- The quantity of performance timing data to gather.
  10. logical Lsequen -- Whether or not to make moves in a sequential manner.
  11. integer real RFdisp deltamax -- The number of displacement moves to make each cycle, and the maximum displacement (to start with.)
  12. real integer TAratio nadjust -- The target acceptance ratio, and the number of moves between adjustments.
  13. integer integer real RFhmc Linteg dtime -- The relative frequency of HMC-type moves, the number of integration steps, and the time-step for these moves. This does not work at the present time.
  14. integer RFkawa -- The relative frequency of species-exchange moves.
  15. integer Nspecies -- the number of different species in the simulation.
  16. integer integer... RFcdX(1) RFcdX(2)... -- The number of create or destroy moves to make in each cycle of the simulation, for each species.
  17. Lastly, the file contains one or more lines of the form:
    Temperature   mu1   mu2   mu3....
    which control the temperature and chemical potentials at each stage of the run.


Last modified on Fri Aug 17 16:24:29 2007
Email: gelb@wustl.edu