msi2lmp.exe

This code has several known limitations listed below under "LIMITATIONS" (and possibly some unknown ones, too) and is no longer under active development. Only the occasional bugfix is applied.

Please send any inquiries about msi2lmp to the lammps-users mailing list and not to individual people.

OVERVIEW

This is the third version of a program that generates a LAMMPS data file based on the information in MSI .car (atom coordinates), .mdf (molecular topology) and .frc (forcefield) files. The .car and .mdf files are specific to a molecular system while the .frc file is specific to a forcefield version. The only coherency needed between .frc and .car/.mdf files are the atom types.

The first version was written by Steve Lustig at Dupont, but required using Discover to derive internal coordinates and forcefield parameters

The second version was written by Michael Peachey while an intern in the Cray Chemistry Applications Group managed by John Carpenter. This version derived internal coordinates from the mdf file and looked up parameters in the frc file thus eliminating the need for Discover.

The third version was written by John Carpenter to optimize the performance of the program for large molecular systems (the original code for deriving atom numbers was quadratic in time) and to make the program fully dynamic. The second version used fixed dimension arrays for the internal coordinates.

The third version was revised in Fall 2011 by Stephanie Teich-McGoldrick to add support non-orthogonal cells.

The next revision was started in Summer/Fall 2013 by Axel Kohlmeyer to improve portability to Windows compilers, clean up command line parsing and improve compatibility with the then current LAMMPS versions. This revision removes compatibility with the obsolete LAMMPS version written in Fortran 90.

INSTALLATION & USAGE

This program uses the .car and .mdf files from MSI/Biosyms's INSIGHT program to produce a LAMMPS data file.

1. Building msi2lmp
   
   Use the Makefile in the src directory. It is currently set up for gcc. You will have to modify it to use a different compiler.
2. Testing the program
   
   There are several pairs of input test files in the format generated by materials studio or compatible programs (one .car and one .mdf file each) in the test directory. There is also a LAMMPS input to run a minimization for each and write out the resulting system as a data file. With the runtests.sh script all of those inputs are converted via msi2lmp, then the minimization with LAMMPS is run and the generated data files are compared with the corresponding files in the reference folder. This script assumes you are on a unix/linux system and that you have compile a serial LAMMPS executable called lmp_serial with make serial. The tests are groups by the force fields they use.
3. To run the program
   
   The program is started by supplying information at the command prompt according to the usage described below.
   
   USAGE: msi2lmp.exe <ROOTNAME> {-print #} {-class #} {-frc FRC_FILE} {-ignore} {-nocenter} {-shift # # #}
   • msi2lmp.exe is the name of the executable
   • <ROOTNAME> is the base name of the .car and .mdf files
   • -2001 Output lammps files for LAMMPS version 2001 (F90 version) Default is to write output for the C++ version of LAMMPS
   • -print (or -p)
   • is the print level 0 - silent except for error messages

	                       1 - minimal (default)
                               2 - verbose (usual for developing and
                                   checking new data files for consistency)
                               3 - even more verbose (additional debug info)

   -- -ignore (or -i)     ignore errors about missing force field parameters
                          and treat them as warnings instead.

   -- -nocenter (or -n)   do not recenter the simulation box around the
                          geometrical center of the provided geometry but
                          rather around the origin

   -- -oldstyle (or -o)   write out a data file without style hints
                          (to be compatible with older LAMMPS versions)

   -- -shift (or -s)      translate the entire system (box and coordinates)
                          by a vector (default: 0.0 0.0 0.0)

   -- -class  (or -c)
        # is the class of forcefield to use (I  or 1 = Class I e.g., CVFF)
                                            (O  or 0 = OPLS-AA)
                                            (II or 2 = Class II e.g., CFFx)
        default is -class I

   -- -frc    (or -f) specifies name of the forcefield file (e.g., cff91)
 
     If the file name includes a directory component (or drive letter 
     on Windows), then the name is used as is. Otherwise, the program
     looks for the forcefield file in $MSI2LMP_LIBRARY (or %MSI2LMP_LIBRARY%
     on Windows). If $MSI2LMP_LIBRARY is not set, ../frc_files is used
     (for testing). If the file name does not end in .frc, then .frc
     is appended to the name.

     For example,  -frc cvff (assumes cvff.frc is in $MSI2LMP_LIBRARY
                              or ../frc_files)

                   -frc cff/cff91 (assumes cff91.frc is in cff)

                   -frc /usr/local/forcefields/cff95
                       (assumes cff95.frc is in /usr/local/forcefields/)

     By default, the program uses $MSI2LMP_LIBRARY/cvff.frc  or
      ../frc_files/cvff.frc depending on whether MSI2LMP_LIBRARY is set.

  -- the LAMMPS data file is written to <ROOTNAME>.data
     protocol and error information is written to the screen.

LIMITATIONS

msi2lmp has the following known limitations:

• there is no support to select morse bonds over harmonic bonds
• there is no support for auto-equivalences to supplement fully parameterized interactions with heuristic ones
• there is no support for bond increments
• there is no support for coordinates defined by symmetry operations, i.e. the .mdf file has to be set up for space group P1

CHANGELOG

06 Oct 2016 Axel Kohlmeyer <akohlmey@gmail.com>

Improved whitespace handling in parsing topology and force field files to avoid bogus warnings about type name truncation.

24 Oct 2015 Axel Kohlmeyer <akohlmey@gmail.com>

Added check to make certain that force field files are consistent with the notation of non-bonded parameters that the msi2lmp code expects. For Class 1 and OPLS-AA the A-B notation with geometric mixing is expected and for Class 2 the r-eps notation with sixthpower mixing.

11 Sep 2014 Axel Kohlmeyer <akohlmey@gmail.com>

Refactored ReadMdfFile.c so it more consistently honors the MAX_NAME and MAX_STRING string length defines and potentially handles inputs with long names better.

27 May 2014 Axel Kohlmeyer <akohlmey@gmail.com>

Added TopoTools style type hints as comments to all Mass, PairCoeff, BondCoeff, AngleCoeff, DihedralCoeff, ImproperCoeff entries. This should make it easier to identify force field entries with the structure and force field map in the data file later.

06 Mar 2014 Axel Kohlmeyer <akohlmey@gmail.com>

Fixed a bug in handling of triclinic cells, where the matrices to convert to and from fractional coordinates were incorrectly built.

26 Oct 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Implemented writing out force field style hints in generated data files for improved consistency checking when reading those files. Also added writing out CGCMM style comments to identify atom types.

08 Oct 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Fixed a memory access violation with Class 2 force fields. Free all allocated memory to better detection of memory errors. Print out version number and data with all print levels > 0. Added valgrind checks to the regression tests.

02 Aug 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Added rudimentary support for OPLS-AA based on input provided by jeff greathouse.

18 Jul 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Added support for writing out image flags. Improved accuracy of atom masses. Added flag for shifting the entire system. Fixed some minor logic bugs and prepared for supporting other force fields and morse style bonds.

12 Jul 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Fixed the bug that caused improper coefficients to be wrong. Cleaned up the handling of box parameters and center the box by default around the system/molecule. Added a flag to make this step optional and center the box around the origin instead. Added a regression test script with examples.

1 Jul 2013 Axel Kohlmeyer <akohlmey@gmail.com>

Cleanup and improved port to windows. Removed some more static string limits. Added print level 3 for additional output. Make code stop at missing force field parameters and added -i flag to override this. Safer argument checking. Provide short versions for all flags.

23 Sep 2011

added support for triclinic boxes

V3.6 outputs to LAMMPS 2005 (C++ version).

Contact: Kelly L. Anderson, kelly.anderson@cantab.net

V3.4 and V3.5 are minor upgrades to fix bugs associated mostly with .car and .mdf files written by Accelys' Materials Studio GUI. April 2005

V3.3 was generated in response to a strange error reading a MDF file generated by Accelys' Materials Studio GUI. Simply rewriting the input part of ReadMdfFile.c seems to have fixed the problem.

V3.2 corresponds to adding code to MakeLists.c to gracefully deal with systems that may only be molecules of 1 to 3 atoms. In V3.1, the values for number_of_dihedrals, etc. could be unpredictable in these systems.

msi2lmp v3.9.8 6/10/2016