The tar file of all Linux executables and input files needed to play around with event-driven MD and analyze the resulting packings is PackLSD.tar.gz. For now I have only documented the sphere packing options and hidden the rest, in order to avoid additional complexities for ellipsoids. I have also hidden options related to neighbour lists and other optimizations that require close familiarity with the algorithm and are hard to explain in detail. Contact me if you need more. If you use this to produce published results, please reference the following paper:

"Neighbor List Collision-Driven Molecular Dynamics Simulation for Nonspherical Particles. I. Algorithmic Details II. Applications to Ellipses and Ellipsoids", by A. Donev, F. H. Stillinger, and S. Torquato, J. Comp. Phys, 202(2): 737-764 (part I) and 765-793 (part II),  2005

Please read the following instructions carefully before asking questions, and also take a look at the examples I provided!

Producing Packings using MD

The binaries for generating jammed packing using hard-particle MD are PackLSD.static.x for packings which consist of a small number of particle species (monodisperse, bidisperse, tridisperse, etc.), or PackLSD.poly.static.x for truly polydisperse packings where each particle has a different shape/size. The executables only link in standard (OpenGL, libtiff, libjpeg, etc.) Linux libraries, and link openglut statically, which may cause failures on some systems. The programs are compiled at medium optimization, for a generic IA32 processor, on a Red Hat EL4 system. They should run on Red Hat (>=9) and Fedora (>=3) systems just fine, and hopefully other similar distributions as well. If something fails to run, I can try to help prepare a version for you with the correct libraries.

The runtime input file PackLSD.input or PackLSD.poly.input is to be piped into the executable. The main input file is a collection of Fortran 90 namelists as in the examples Input/PackLSD.nml and Input/PackLSD.poly.nml. The name of the packing can be changed inside PackLSD.input and is used as the base of all input/output filenames (as in the filename PackLSD.nml). The program usually prints when files are opened or closed. The executable needs to be run in a directory where there will be sub-directories called Input, Output, Stats, Data, Scenes, Plots, and VRML, and the files are placed or read from these subdirectories. The nml file must be in the Input directory, and the output packing files will be saved to Data. I have documented the options you will need in the nml file and removed most others (some need to be left in since the defaults are not appropriate for generic uses). Most options are the same for both programs PackLSD.static.x and PackLSD.poly.static.x, and the minor differences can be seen in the example Input/PackLSD.poly.nml. Please ask me if some option is not clear.

The simulation processes a certain number of MD events (collisions), called a stage. At the end of every stage certain statistics are output, possibly a temporary file saved, the OpenGL view is refreshed, particle velocities rescaled, etc. Many stages can be run until jamming is reached, as defined by the options in the namelist file. The executable can use OpenGL to render to screen the packing process so you can see what is happening. You can use the mouse to zoom in, rotate, etc. There are also options to generate snapshot images that can be used to make animations (I highly recommend the MNG file format!)

Analyzing Packings

The produced packings can later be viewed/analyzed by using the programs ProcessPacking.static.x or ProcessPacking.poly.static.x. In particular, you can render using OpenGL, make plots, or compute the pair correlation function g2(r) or the structure factor S(k), as well as construct and render NNLs (near-neighbor lists) based on proximity or Delaunay/Voronoi neighbors. The input file is the namelist Input/Process_HPP.nml in the Input subdirectory. It contains comments describing the various input fields. In 2D one can generate nice EPS figures to include in papers, and in 3D I provide some VRML models. These are not documented in detail (learn by looking at examples, although knowledge of VRML is required). The (anisotropic, vector k) structure factor in 3D is saved as a VTK file that can be looked at using the VisIt rendering program.

The input/output packing files use a format which should be simple to understand, see the example dat file Data/LSD.HS.mono.3D.N=1000.1.dat. For packings with the PackLSD.static.x program, this format is an ASCII text file as follows:

______________________
n_dims ! Euclidian space dimensionality (2 or 3)
N dispersity ! Total number of particles and number of species
N_1 N_2 ... ! Number of particles of each specie
D_1 D_2... ! Sphere diameters for each specie
Lambda ! Lattice vectors for unit cell
T T T ! Periodic or hard wall boundaries
x1 y1 z1 ! Coordinates of first sphere center
x2 y2 z2 ! second, etc...
...
______________________

The format is slightly different for truly polydisperse packings, the particle-specific diameter is after each particle's coordinates:

______________________
n_dims
N dispersity
N_1 N_2 ... ! One for each specie (each specie is a distribution for polydisperse packings, rather than a fixed size)
Lambda ! Unit cell
T T T ! Periodic or hard wall
x1 y1 z1 D1
x2 y2 z2 D2
...
______________________