Hi, all. A paper just appeared in the Journal of Chemical Theory and Computation (D. S. Cerutti, R. Duke, P. L. Freddolino, H. Fan, T. P. Lybrand, ASAP 2008) laying out problems arising from reuse of seeds for random number generators (e. g. involved in thermostatting). It would appear that LAMMPS (used improperly) is not immune to these problems, which the authors show to be large for simulations of proteins. I guess the message is to avoid reusing seeds when restarting simulations from restart files. Just posting this in case anyone on the list is doing such simulations.