22 Dec 2004

Added the capability for using wild-card asteriks in the following input script commands: mass, dipole, pair_coeff, bond_coeff, angle_coeff, dihedral_coeff, improper_coeff. The pair_coeff command already allowed for wild-cards; this is a richer syntax.

With the new syntax, a range of atom types (or bond types, angle types, etc) can be specified in the following ways:

mass 3 1.0 (set mass of type 3 to 1.0)
mass * 1.0 (set mass of all types to 1.0)
mass 4* 1.0 (set mass of types 4-N)
mass *2 1.0 (set types 1-2)
mass 2*3 1.0 (set types 2-3) 

This feature will be part of the next LAMMPS release (Jan 2005).

17 Dec 2004

Fixed a bug whereby deleting fixes did not causes lists of other fixes to be updated correctly.

Thanks to Paul Crozier (Sandia) for stumbling into this bug.

16 Dec 2004

Fixed a small bug with how the delete_bonds command computes the cutoff distance it needs to acquire nearby atoms.

14 Dec 2004

Added a pre-computed table option for pair potentials that couple to long-range Coulombic solvers (Ewald, PPPM). This speeds up the pairwise computation of forces and energies by as much as a factor of 2 for the lj/charmm/coul/long and lj/cut/coul/long pair styles.

This option is controlled by the new "table" keyword of the pair_modify command. The default is to use a table length of 4096 (N = 12) which gives good speed-up and an accuracy as good or better than the no-table option (since it estimates erfc() by a polynomial). Setting N = 0 uses no tables, which was the default prior to this patch.

At some point, the benchmark timings for the rhodopsin problem will be updated to reflect the speed-up due to this change.

This feature was implemented with Paul Crozier (Sandia).

1 Dec 2004

Fixed a small bug in the reading of dihedral and improper coefficients. Some coeffs were expected to be integer values. If they were listed in the input script or data file as floating point values (1.0 instead of 1), then LAMMPS would read the next value(s) incorrectly.

30 Nov 2004

Added a targeted molecular dynamics (TMD) capability. This allows a group of atoms to be guided to a new configuration over the course of a simulation. See the fix tmd command for more information.

This feature was implemented by Paul Crozier (Sandia) and Christian Burisch (Bochum Univeristy, Germany).

29 Oct 2004

Added FFT support for the SGI SCSL library used on SGI Altix machines. Also a Makefile.altix.

These additions were implemented by Jim Shepherd (Ga Tech).

28 Oct 2004

Fixed a small bug with the variable proc command.

Thanks to Lei Guo (U Illinois) for pointing out the problem.

13 Oct 2004

Fixed a bug with the neigh_modify exclude molecule command, which didn't work when molecules straddled processors. Requires the molecule ID to be communicated when neighbor lists are built.

Thanks to Rob Hoy (JHU) for pointing out the problem.

12 Oct 2004

Added a PDF version of the LAMMPS manual to the documentation as a single downloadable file. Produced via the htmldoc tool.

11 Oct 2004

Ara Kooser and Jeff Greathouse (Sandia) wrote some post-processing analysis tools for LAMMPS. They will be included in the tools directory of the next LAMMPS release. Jeff updated lmp2arc to the current LAMMPS version. It creates Accelrys Insight files from LAMMPS dump files. Ara wrote a lmp2cfg tool that converts LAMMPS dump files to AtomEye visualizer cfg files. He also wrote a lmp2traj tool that can create contour maps of atom positions as well as density and dipole moment profiles. See the tools section of the LAMMPS documentation for details.

8 Oct 2004

Fixed two-line bug with the kspace_modify command. The slab keyword was inconsistently named in the code vs the documentation. The correct keyword slab is now used in the code.

Thanks to Jeff Greathouse (Sandia) for pointing this out.

2 Oct 2004

Fixed one-line bug with the fix volume/rescale command. Expansion/compression in the z dimension would grow/shrink the simulation box correctly, but move atoms incorrectly.

Thanks to Pieter in 't Veld (Sandia) for pointing out the problem.

1 Oct 2004

Fixed bug with respa and granular pairwise potentials that track shear history. The bug has to do with how a fix associated with these commands was indexed. If other fixes were deleted, subsequent runs could create an invalid pointer to these fixes and cause LAMMPS to crash.

Thanks to Paul Crozier (Sandia) for pointing out the problem.

30 Sept 2004

Made error checking for the fix shake commmand more precise so that bonds deleted and removed by the delete_bonds command do not cause needless errors.

13 Sept 2004

Added a hybrid style to the bond_style command. This allows potentials for multiple bond styles to be used in a single simulation. For example, bonds in a polymer flow (of bond type 1) could be computed with a fene potential and bonds in the wall boundary (of bond type 2) could be computed with a harmonic potential. The bond_coeff command is where the assignment of style to type is made. For example, these commands would setup the polymer flow simulation just described:

bond_style hybrid
bond_coeff 1 fene 30.0 1.5 1.0 1.0
bond_coeff 2 harmonic 80.0 1.2 

10 Sept 2004

Fixed a one-line bug in neigh_pair_gran.cpp in Nsq newton-off neighbor list construction for granular systems.

9 Sept 2004

Fixed a bug in LAMMPS 2001 in force_many.f in dihedral_charmm() routine that affected energy computation when called from the energy minimizer (OK from dynamics). The 1-4 contribution to vdwl energy was not being added in.

Thanks to Kevin Leung (Sandia) for pointing this out.

8 Sept 2004

Fixed a small memory leak in force.cpp with respect to force style strings.

7 Sept 2004

Made the set_type command more general so it can be used to change the types of atoms, bonds, angles, etc within a group. This changes the syntax from

set_type	group-ID 3 

to

set_type	group-ID atom 3
set_type	group-ID bond 4 

Several of the input scripts in the examples dir were tweaked to reflect the new syntax.