CHANGES TO THE GROMOS PACKAGE. B.P. van Eijck, March 1993.

The program PROMD(L) and the subroutines RUNMD, SHAKE, FORCE(L),
   NBPML(C/S), NONBML(C/S), COBOND, ANGLE and DIHANG were modified.
The modified programs carry an E, e.g. ENONBMLC.
   L denotes a large version, S a regular Fortran version and 
   C a special Convex version. The Fortran sources are collected in 
   the file Egromos.Z, the link procedure in Elink.Z, and two 
   examples (kris and sol) in Etest.Z. Files can be recovered by:
          uncompress *Z
          tar xvf E*
   A small version, working on a PC, is also available (the Z files
   carry the prefix PC and in the filenames the prefix E is omitted).

________________________________________________________________
The algorithm in DIHANG was modified.
Regardless of NTB, the center-of-mass movement is stopped 
   once in every NSCM steps.
If NTPR<0, a configuration file is saved every -NTPR steps.
_______________________________________________________________


I. Ewald-corrections to Coulomb and attractive LJ energy.


Added are two subroutines, RWALD(C/S) and KWALD, which calculate the Ewald 
corrections to the energy, forces and virial in real space and in
reciprocal space, respectively. They are called by FORCE. 
Some additional input in PROMD is necessary, of course; see comments
in the program. Data are transferred by the common block EWALD.


II. If abs(NTB)=3 the pressure is calculated on an atomic basis.
This is not (yet) possible for the long-range part calculated in NBPML.


III. If NTP=3 the angle beta is adjusted too (monoclinic case).
Since beta can change, the output at the end of tape12 (trajectory)
and tape31 (final coordinates and velocities) is modified to contain
always alpha, beta, gamma after the box dimensions (alpha and gamma
being always 90 degrees until I find the need to implement the triclinic 
case). This requires slight modifications in all programs that read 
coordinate files (EPROAHB, EPROAVQ, EPROAVX available).
In tape15 the cell angles and pressure values are also written: there
are now 73 rather than 61 values per frame. This could be organized
more efficiently: diagonal pressure values are given twice.


IV. A tape99 is created which tells the user in batch mode how many steps 
have been done. Changing the number in this file changes NSTLIM during
the run, if desired.


V. Selected structural parameters can be put on file directly during 
simulation: see comments in PROMD and common block NOUT.


VI. If the sign of TAUTP and/or TAUTS is negative, the kinetic energy is 
calculated by averaging the temperatures at t+dt/2 and t-dt/2 rather than
the velocities (I think that it is better to average v**2 than v).