temper/grem N M lambda fix-ID thermostat-ID seed1 seed2 index
N = total # of timesteps to run
M = attempt a tempering swap every this many steps
lambda = initial lambda for this ensemble
fix-ID = ID of fix_grem
thermostat-ID = ID of the thermostat that controls kinetic temperature
seed1 = random # seed used to decide on adjacent temperature to partner with
seed2 = random # seed for Boltzmann factor in Metropolis swap
index = which temperature (0 to N-1) I am simulating (optional)
temper/grem 100000 1000 $lambda fxgREM fxnvt 0 58728 temper/grem 40000 100 $lambda fxgREM fxnpt 0 32285 $walkers
Run a parallel tempering or replica exchange simulation in LAMMPS partition mode using multiple generalized replicas (ensembles) of a system defined by fix grem, which stands for the generalized replica exchange method (gREM) originally developed by (Kim). It uses non-Boltzmann ensembles to sample over first order phase transitions. The is done by defining replicas with an enthalpy dependent effective temperature
Two or more replicas must be used. See the temper command for an explanation of how to run replicas on multiple partitions of one or more processors.
This command is a modification of the temper command and has the same dependencies, restraints, and input variables which are discussed there in greater detail.
Instead of temperature, this command performs replica exchanges in lambda as per the generalized ensemble enforced by fix grem. The desired lambda is specified by lambda, which is typically a variable previously set in the input script, so that each partition is assigned a different temperature. See the variable command for more details. For example:
variable lambda world 400 420 440 460 fix fxnvt all nvt temp 300.0 300.0 100.0 fix fxgREM all grem $lambda -0.05 -50000 fxnvt temper 100000 100 $lambda fxgREM fxnvt 3847 58382
would define 4 lambdas with constant kinetic temperature but unique generalized temperature, and assign one of them to fix grem used by each replica, and to the grem command.
As the gREM simulation runs for N timesteps, a swap between adjacent ensembles will be attempted every M timesteps. If seed1 is 0, then the swap attempts will alternate between odd and even pairings. If seed1 is non-zero then it is used as a seed in a random number generator to randomly choose an odd or even pairing each time. Each attempted swap of temperatures is either accepted or rejected based on a Metropolis criterion, derived for gREM by (Kim), which uses seed2 in the random number generator.
File management works identical to the temper command. Dump files created by this fix contain continuous trajectories and require post-processing to obtain per-replica information.
The last argument index in the grem command is optional and is used when restarting a run from a set of restart files (one for each replica) which had previously swapped to new lambda. This is done using a variable. For example if the log file listed the following for a simulation with 5 replicas:
500000 2 4 0 1 3
then a setting of
variable walkers world 2 4 0 1 3
would be used to restart the run with a grem command like the example above with $walkers as the last argument. This functionality is identical to temper.