From 56598fcd0bcd59a5d40328c6414e27da40b75c23 Mon Sep 17 00:00:00 2001 From: Steve Plimpton Date: Thu, 11 Oct 2018 08:29:25 -0600 Subject: [PATCH] changes to prd command doc page --- doc/src/prd.txt | 50 +++++++++++++++++++++++++++++++------------------ 1 file changed, 32 insertions(+), 18 deletions(-) diff --git a/doc/src/prd.txt b/doc/src/prd.txt index f71f285336..43c0d309c8 100644 --- a/doc/src/prd.txt +++ b/doc/src/prd.txt @@ -48,11 +48,12 @@ replicas of a system. One or more replicas can be used. The total number of steps {N} to run can be interpreted in one of two ways; see discussion of the {time} keyword below. -PRD is described in "this paper"_#Voter1998 by Art Voter. It is a method -for performing accelerated dynamics that is suitable for -infrequent-event systems that obey first-order kinetics. A good -overview of accelerated dynamics methods for such systems in given in -"this review paper"_#Voter2002prd from the same group. To quote from the +PRD is described in "(Voter1998)"_#Voter1998 by Art Voter. Similar to +global or local hyperdynamics (HD), PRD is a method for performing +accelerated dynamics that is suitable for infrequent-event systems +that obey first-order kinetics. A good overview of accelerated +dynamics methods for such systems in given in this review paper +"(Voter2002)"_#Voter2002prd from Art's group. To quote from the paper: "The dynamical evolution is characterized by vibrational excursions within a potential basin, punctuated by occasional transitions between basins." The transition probability is @@ -61,15 +62,27 @@ Running multiple replicas gives an effective enhancement in the timescale spanned by the multiple simulations, while waiting for an event to occur. -Each replica runs on a partition of one or more processors. Processor -partitions are defined at run-time using the "-partition command-line -switch"_Run_options.html. Note that if you have MPI installed, you -can run a multi-replica simulation with more replicas (partitions) -than you have physical processors, e.g you can run a 10-replica -simulation on one or two processors. However for PRD, this makes -little sense, since running a replica on virtual instead of physical -processors,offers no effective parallel speed-up in searching for -infrequent events. See the "Howto replica"_Howto_replica.html doc +Both PRD and HD produce a time-accurate trajectory that effectively +extends the timescale over which a system can be simulated, but they +do it differently. PRD creates Nr replicas of the system and runs +dynamics on each independently with a normal unbiased potential until +an event occurs in one of the replicas. The time between events is +reduced by a factor of Nr replicas. HD uses a single replica of the +system and accelerates time by biasing the interaction potential in a +manner such that each timestep is effectively longer. For both +methods, per wall-clock second, more physical time elapses and more +events occur. See the "hyper"_hyper.html doc page for more info about +HD. + +In PRD, each replica runs on a partition of one or more processors. +Processor partitions are defined at run-time using the "-partition +command-line switch"_Run_options.html. Note that if you have MPI +installed, you can run a multi-replica simulation with more replicas +(partitions) than you have physical processors, e.g you can run a +10-replica simulation on one or two processors. However for PRD, this +makes little sense, since running a replica on virtual instead of +physical processors,offers no effective parallel speed-up in searching +for infrequent events. See the "Howto replica"_Howto_replica.html doc page for further discussion. When a PRD simulation is performed, it is assumed that each replica is @@ -78,8 +91,8 @@ I.e. the simulation domain, the number of atoms, the interaction potentials, etc should be the same for every replica. A PRD run has several stages, which are repeated each time an "event" -occurs in one of the replicas, as defined below. The logic for a PRD -run is as follows: +occurs in one of the replicas, as explained below. The logic for a +PRD run is as follows: while (time remains): dephase for n_dephase*t_dephase steps @@ -129,7 +142,8 @@ Minimization parameters may be set via the PRD command. The latter are the settings that would be used with the "minimize"_minimize.html command. Note that typically, you do not need to perform a highly-converged minimization to detect a transition -event. +event, though you may need to in order to prevent a set of atoms in +the system from relaxing to a saddle point. The event check is performed by a compute with the specified {compute-ID}. Currently there is only one compute that works with the @@ -307,7 +321,7 @@ deposit"_fix_deposit.html. "min_modify"_min_modify.html, "min_style"_min_style.html, "run_style"_run_style.html, "minimize"_minimize.html, "velocity"_velocity.html, "temper"_temper.html, "neb"_neb.html, -"tad"_tad.html +"tad"_tad.html, "hyper"_hyper.html [Default:]