ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

make ramp more flexible by defining lambda through a variable

Browse Source
This commit is contained in:
Axel Kohlmeyer
2023-02-27 16:01:17 -05:00
parent 83f936f31b
commit 446913f2f3
9 changed files with 122 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
doc/src/fix_alchemy.rst
View File

@ -8,17 +8,18 @@ Syntax
.. parsed-literal::
fix ID group-ID alchemy
fix ID group-ID alchemy v_name
* ID, group-ID are documented in :doc:`fix <fix>` command
* alchemy = style name of this fix command
* v_name = variable with name that determines the :math:`\lambda_p` value
Examples
""""""""
.. code-block:: LAMMPS
fix trans all alchemy
fix trans all alchemy v_ramp
Description
"""""""""""
@ -30,14 +31,17 @@ simulation between two topologies (i.e. the same number and positions of
atoms, but differences in atom parameters like type, charge, bonds,
angles and so on). For this a :ref:`multi-partition run <partition>` is
required with exactly two partitions. During the MD run, the fix will
will determine a factor :math:`\lambda_p` that will be linearly ramped
*down* from 1.0 to 0.0 for the *first* partition (*p=0*) and ramped *up*
from 0.0 to 1.0 for the *second* partition (*p=1*). The forces used for
the propagation of the atoms will be the sum of the forces of the two
systems combined and scaled with their respective :math:`\lambda_p`
factor. This allows to perform transformations that are not easily
possible with :doc:`pair style hybrid/scaled <pair_hybrid>`, :doc:`fix
adapt <fix_adapt>` or :doc:`fix adapt/fep <fix_adapt_fep>`.
will determine a factor, :math:`\lambda_p`, for each partition *p* that
will be taken from an equal style or equivalent :doc:`variable
<variable>`. Typically, this variable would be chose to linearly ramp
*down* from 1.0 to 0.0 for the *first* partition (*p=0*) and linearly
ramp *up* from 0.0 to 1.0 for the *second* partition (*p=1*). The
forces used for the propagation of the atoms will be the sum of the
forces of the two systems combined and scaled with their respective
:math:`\lambda_p` factor. This allows to perform transformations that
are not easily possible with :doc:`pair style hybrid/scaled
<pair_hybrid>`, :doc:`fix adapt <fix_adapt>` or :doc:`fix adapt/fep
<fix_adapt_fep>`.
Due to the specifics of the implementation, the initial geometry and
dimensions of the system must be exactly the same and the fix will
@ -56,6 +60,8 @@ copper/aluminum bronze.
.. code-block:: LAMMPS
variable name world pure alloy
create_box 2 box
create_atoms 1 box
pair_style eam/alloy
@ -66,6 +72,15 @@ copper/aluminum bronze.
if "${name} == alloy" then &
"set type 1 type/fraction 2 0.05 6745234"
# define ramp variable to combine the two different partitions
if "${name} == pure" then &
"variable ramp equal ramp(1.0,0.0)" &
else &
"variable ramp equal ramp(0.0,1.0)"
fix 2 all alchemy v_ramp
The ``examples/PACKAGES/alchemy`` folder contains complete example
inputs for this command.
@ -78,7 +93,7 @@ No information about this fix is written to :doc:`binary restart files <restart>
None of the :doc:`fix_modify <fix_modify>` options are relevant to this fix.
This fix stores a global scalar (the current value of :math:`\lambda_p`)
and a global vector or length 3 which contains the potential energy of
and a global vector of length 3 which contains the potential energy of
the first partition, the second partition and the combined value,
respectively. The global scalar is unitless and "intensive", the vector
is in :doc:`energy units <units>` and "extensive". These values can be
@ -86,14 +101,6 @@ used by any command that uses a global value from a fix as input. See
the :doc:`Howto output <Howto_output>` doc page for an overview of
LAMMPS output options.
The value of :math:`\lambda_p` is influenced by the *start/stop* keywords
of the :doc:`run <run>` command. Without them it will be ramped
linearly from 1.0 to 0.0 (partition 1) and 0.0 to 1.0 (partition 2)
during the steps of a run, with
*start/stop* keywords the ramp is from the *start* time step to the
*stop* timestep. This allows to break down a simulation over multiple
*run* commands or to continue transparently from a restart.
This fix is not invoked during :doc:`energy minimization <minimize>`.
Restrictions

8
examples/PACKAGES/alchemy/in.alloy
View File

@ -19,9 +19,15 @@ pair_coeff * * AlCu.eam.alloy Cu Al
if "${name} == alloy" then &
"set type 1 type/fraction 2 0.05 6745234"
# define ramp variable to combine the two different partitions
if "${name} == pure" then &
"variable ramp equal ramp(1.0,0.0)" &
else &
"variable ramp equal ramp(0.0,1.0)"
velocity all create 5000.0 6567345
fix 1 all nvt temp 2500.0 500.0 0.002
fix 2 all alchemy
fix 2 all alchemy v_ramp
compute pressure all pressure/alchemy 2

8
examples/PACKAGES/alchemy/in.twowater
View File

@ -73,11 +73,17 @@ else &
velocity all create 300.0 5463576
timestep 0.2
# define ramp variable to combine the two different partitions
if "${name} == twowater" then &
"variable ramp equal ramp(1.0,0.0)" &
else &
"variable ramp equal ramp(0.0,1.0)"
# since the trajectory and forces are kept identical through fix alchemy,
# we can do fix npt simulations, but we must use the "mixed" pressure
fix integrate all npt temp 300 300 1.0 iso 1.0 1.0 10.0
fix transform all alchemy
fix transform all alchemy v_ramp
compute pressure all pressure/alchemy transform
fix_modify integrate press pressure

12
examples/PACKAGES/alchemy/log.25Feb23.alloy.g++.4.0
View File

@ -19,7 +19,7 @@ timestep 0.002
create_atoms 1 box
Created 864 atoms
using lattice units in orthogonal box = (0 0 0) to (22.5 22.5 22.5)
create_atoms CPU = 0.000 seconds
create_atoms CPU = 0.001 seconds
displace_atoms all random 0.3 0.3 0.3 57845645
Displacing atoms ...
pair_style eam/alloy
@ -29,9 +29,13 @@ Reading eam/alloy potential file AlCu.eam.alloy with DATE: 2008-10-01
# replace 5% of copper with aluminium on the second partition only
if "${name} == alloy" then "set type 1 type/fraction 2 0.05 6745234"
# define ramp variable to combine the two different partitions
if "${name} == pure" then "variable ramp equal ramp(1.0,0.0)" else "variable ramp equal ramp(0.0,1.0)"
variable ramp equal ramp(1.0,0.0)
velocity all create 5000.0 6567345
fix 1 all nvt temp 2500.0 500.0 0.002
fix 2 all alchemy
fix 2 all alchemy v_ramp
compute pressure all pressure/alchemy 2
@ -159,7 +163,7 @@ Per MPI rank memory allocation (min/avg/max) = 3.474 | 3.474 | 3.474 Mbytes
9800 520.37758 -31000.6 -2828.0252 8.0039334 -2886.0741 58.048903 0.02 -2918.4921
9900 506.49368 -32223.744 -2830.1705 8.0039334 -2886.6706 56.500132 0.01 -2920.4741
10000 507.13597 -33930.476 -2834.3236 8.0039334 -2890.8954 56.571781 0 -2924.5423
Loop time of 19.2208 on 2 procs for 10000 steps with 864 atoms
Loop time of 17.7279 on 2 procs for 10000 steps with 864 atoms
Total wall time: 0:00:19
Total wall time: 0:00:17

10
examples/PACKAGES/alchemy/log.25Feb23.alloy.g++.4.1
View File

@ -32,9 +32,13 @@ set type 1 type/fraction 2 0.05 6745234
Setting atom values ...
37 settings made for type/fraction
# define ramp variable to combine the two different partitions
if "${name} == pure" then "variable ramp equal ramp(1.0,0.0)" else "variable ramp equal ramp(0.0,1.0)"
variable ramp equal ramp(0.0,1.0)
velocity all create 5000.0 6567345
fix 1 all nvt temp 2500.0 500.0 0.002
fix 2 all alchemy
fix 2 all alchemy v_ramp
compute pressure all pressure/alchemy 2
@ -162,7 +166,7 @@ Per MPI rank memory allocation (min/avg/max) = 3.474 | 3.474 | 3.474 Mbytes
9800 524.81105 -31000.6 -2860.6103 7.8067107 -2919.1537 58.543462 0.98 -2918.4921
9900 507.60008 -32223.744 -2864.192 7.8067107 -2920.8155 56.623552 0.99 -2920.4741
10000 521.28866 -33930.476 -2866.3918 7.8067107 -2924.5423 58.150534 1 -2924.5423
Loop time of 19.2208 on 2 procs for 10000 steps with 864 atoms
Loop time of 17.7277 on 2 procs for 10000 steps with 864 atoms
Total wall time: 0:00:19
Total wall time: 0:00:17

36
examples/PACKAGES/alchemy/log.25Feb23.twowater.g++.4.0
View File

@ -1,9 +1,9 @@
LAMMPS (8 Feb 2023)
Processor partition = 0
using 1 OpenMP thread(s) per MPI task
# Example for alchemical transformation of two water molecules into a hydronium and hydroxyl ion
# WARNING: This input is intended for demonstrating the method, only
# The force field parameters are made up and NOT suitable for production simulations.
# Example for an alchemical transformation of two water molecules into a hydronium and hydroxyl ion
# WARNING: This input is intended for demonstrating the method only,
# the force field parameters are mostly made up and NOT suitable for production simulations.
# set up different names for two partitions
variable name world twowater twoions
@ -59,7 +59,7 @@ group transform id 1:6
create_atoms 0 random 32 34564 NULL mol water 25367 overlap 1.33
Created 96 atoms
using lattice units in orthogonal box = (-5 -5 -5) to (5 5 5)
create_atoms CPU = 0.001 seconds
create_atoms CPU = 0.000 seconds
# change topology and settings for the two states
# we cannot simply create a different topology directly or
@ -156,11 +156,15 @@ Finding 1-2 1-3 1-4 neighbors ...
velocity all create 300.0 5463576
timestep 0.2
# define ramp variable to combine the two different partitions
if "${name} == twowater" then "variable ramp equal ramp(1.0,0.0)" else "variable ramp equal ramp(0.0,1.0)"
variable ramp equal ramp(1.0,0.0)
# since the trajectory and forces are kept identical through fix alchemy,
# we can do fix npt simulations, but we must use the "mixed" pressure
fix integrate all npt temp 300 300 1.0 iso 1.0 1.0 10.0
fix transform all alchemy
fix transform all alchemy v_ramp
compute pressure all pressure/alchemy transform
fix_modify integrate press pressure
@ -377,21 +381,21 @@ Per MPI rank memory allocation (min/avg/max) = 7.535 | 7.535 | 7.535 Mbytes
19800 295.68946 81474.342 2308.3249 1.0240743 2219.3041 89.020804 0.01 1910.5545
19900 306.4947 43488.052 2330.4958 1.0222707 2238.222 92.27385 0.005 1957.4368
20000 313.31679 -25133.284 -1161.6979 1.0163289 -1256.0256 94.327722 0 -1649.7551
Loop time of 11.7903 on 2 procs for 20000 steps with 102 atoms
Loop time of 8.92653 on 2 procs for 20000 steps with 102 atoms
Performance: 29.312 ns/day, 0.819 hours/ns, 1696.311 timesteps/s, 173.024 katom-step/s
99.3% CPU use with 2 MPI tasks x 1 OpenMP threads
Performance: 38.716 ns/day, 0.620 hours/ns, 2240.513 timesteps/s, 228.532 katom-step/s
96.3% CPU use with 2 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 6.8605 | 8.6071 | 10.354 | 59.5 | 73.00
Bond | 0.044148 | 0.044761 | 0.045375 | 0.3 | 0.38
Neigh | 0.10568 | 0.10573 | 0.10578 | 0.0 | 0.90
Comm | 0.74919 | 2.4939 | 4.2386 | 110.5 | 21.15
Output | 0.0033838 | 0.0036358 | 0.0038879 | 0.4 | 0.03
Modify | 0.46825 | 0.47526 | 0.48227 | 1.0 | 4.03
Other | | 0.05993 | | | 0.51
Pair | 4.7448 | 5.9908 | 7.2367 | 50.9 | 67.11
Bond | 0.036126 | 0.03641 | 0.036695 | 0.1 | 0.41
Neigh | 0.082247 | 0.082294 | 0.082342 | 0.0 | 0.92
Comm | 0.68467 | 1.9303 | 3.1759 | 89.7 | 21.62
Output | 0.0023377 | 0.0052347 | 0.0081317 | 4.0 | 0.06
Modify | 0.8194 | 0.82122 | 0.82303 | 0.2 | 9.20
Other | | 0.06029 | | | 0.68
Nlocal: 51 ave 51 max 51 min
Histogram: 2 0 0 0 0 0 0 0 0 0
@ -405,4 +409,4 @@ Ave neighs/atom = 359.21569
Ave special neighs/atom = 2
Neighbor list builds = 181
Dangerous builds = 0
Total wall time: 0:00:11
Total wall time: 0:00:08

36
examples/PACKAGES/alchemy/log.25Feb23.twowater.g++.4.1
View File

@ -1,9 +1,9 @@
LAMMPS (8 Feb 2023)
Processor partition = 1
using 1 OpenMP thread(s) per MPI task
# Example for alchemical transformation of two water molecules into a hydronium and hydroxyl ion
# WARNING: This input is intended for demonstrating the method, only
# The force field parameters are made up and NOT suitable for production simulations.
# Example for an alchemical transformation of two water molecules into a hydronium and hydroxyl ion
# WARNING: This input is intended for demonstrating the method only,
# the force field parameters are mostly made up and NOT suitable for production simulations.
# set up different names for two partitions
variable name world twowater twoions
@ -106,7 +106,7 @@ Finding 1-2 1-3 1-4 neighbors ...
1 = max # of 1-3 neighbors
1 = max # of 1-4 neighbors
7 = max # of special neighbors
special bonds CPU = 0.000 seconds
special bonds CPU = 0.002 seconds
create_bonds single/bond 2 3 4
Finding 1-2 1-3 1-4 neighbors ...
special bond factors lj: 0 0 0
@ -180,11 +180,15 @@ Setting atom values ...
velocity all create 300.0 5463576
timestep 0.2
# define ramp variable to combine the two different partitions
if "${name} == twowater" then "variable ramp equal ramp(1.0,0.0)" else "variable ramp equal ramp(0.0,1.0)"
variable ramp equal ramp(0.0,1.0)
# since the trajectory and forces are kept identical through fix alchemy,
# we can do fix npt simulations, but we must use the "mixed" pressure
fix integrate all npt temp 300 300 1.0 iso 1.0 1.0 10.0
fix transform all alchemy
fix transform all alchemy v_ramp
compute pressure all pressure/alchemy transform
fix_modify integrate press pressure
@ -402,21 +406,21 @@ Per MPI rank memory allocation (min/avg/max) = 7.535 | 7.535 | 7.535 Mbytes
19800 295.68946 81474.342 1996.4567 1.0240743 1907.4359 89.020804 0.99 1910.5545
19900 306.4947 43488.052 2048.2997 1.0222707 1956.0258 92.27385 0.995 1957.4368
20000 313.31679 -25133.284 -1555.4274 1.0163289 -1649.7551 94.327722 1 -1649.7551
Loop time of 11.7903 on 2 procs for 20000 steps with 102 atoms
Loop time of 8.92592 on 2 procs for 20000 steps with 102 atoms
Performance: 29.312 ns/day, 0.819 hours/ns, 1696.310 timesteps/s, 173.024 katom-step/s
99.3% CPU use with 2 MPI tasks x 1 OpenMP threads
Performance: 38.719 ns/day, 0.620 hours/ns, 2240.664 timesteps/s, 228.548 katom-step/s
96.2% CPU use with 2 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 6.8748 | 8.631 | 10.387 | 59.8 | 73.20
Bond | 0.04667 | 0.047105 | 0.04754 | 0.2 | 0.40
Neigh | 0.10858 | 0.10862 | 0.10866 | 0.0 | 0.92
Comm | 0.74367 | 2.4997 | 4.2558 | 111.1 | 21.20
Output | 0.0034405 | 0.0037033 | 0.003966 | 0.4 | 0.03
Modify | 0.4338 | 0.4388 | 0.4438 | 0.8 | 3.72
Other | | 0.06136 | | | 0.52
Pair | 4.7753 | 6.0676 | 7.3599 | 52.5 | 67.98
Bond | 0.036352 | 0.036909 | 0.037466 | 0.3 | 0.41
Neigh | 0.085317 | 0.085383 | 0.085449 | 0.0 | 0.96
Comm | 0.72794 | 2.022 | 3.3161 | 91.0 | 22.65
Output | 0.0023255 | 0.005156 | 0.0079865 | 3.9 | 0.06
Modify | 0.64567 | 0.64689 | 0.64812 | 0.2 | 7.25
Other | | 0.06196 | | | 0.69
Nlocal: 51 ave 51 max 51 min
Histogram: 2 0 0 0 0 0 0 0 0 0
@ -430,4 +434,4 @@ Ave neighs/atom = 359.20588
Ave special neighs/atom = 2.0196078
Neighbor list builds = 181
Dangerous builds = 0
Total wall time: 0:00:11
Total wall time: 0:00:08

38
src/REPLICA/fix_alchemy.cpp
View File

@ -18,11 +18,13 @@
#include "compute.h"
#include "domain.h"
#include "error.h"
#include "input.h"
#include "memory.h"
#include "modify.h"
#include "respa.h"
#include "universe.h"
#include "update.h"
#include "variable.h"
#include <cstring>
@ -33,8 +35,12 @@ using namespace FixConst;
FixAlchemy::FixAlchemy(LAMMPS *lmp, int narg, char **arg) : Fix(lmp, narg, arg), commbuf(nullptr)
{
if (narg != 3) error->all(FLERR, "Incorrect number of arguments for fix alchemy");
if (narg != 4) error->all(FLERR, "Incorrect number of arguments for fix alchemy");
if (universe->nworlds != 2) error->all(FLERR, "Must use exactly two partitions");
if (utils::strmatch(arg[3], "^v_"))
id_lambda = arg[3] + 2;
else
error->all(FLERR, "Must use variable as lambda argument to fix alchemy");
lambda = epot[0] = epot[1] = epot[2] = 0.0;
progress = 0;
@ -129,6 +135,14 @@ void FixAlchemy::init()
if (modify->get_fix_by_style("^alchemy").size() > 1)
error->all(FLERR, "There may only one fix alchemy at a time");
ivar = input->variable->find(id_lambda.c_str());
if (ivar < 0)
error->universe_one(FLERR, fmt::format("Variable {} for fix alchemy does not exist", id_lambda));
if (!input->variable->equalstyle(ivar))
error->universe_one(FLERR,
fmt::format("Variable {} for fix alchemy is invalid style", id_lambda));
lambda = input->variable->compute_equal(ivar);
// synchronize box dimensions, determine if resync during run will be needed.
synchronize_box(domain, samerank);
@ -152,7 +166,10 @@ void FixAlchemy::setup(int vflag)
}
if (universe->me == 0) {
progress = static_cast<int>(100.0 - lambda * 100.0);
double delta = update->ntimestep - update->beginstep;
if ((delta != 0.0) && (update->beginstep != update->endstep))
delta /= update->endstep - update->beginstep;
progress = static_cast<int>(delta*100.0);
auto msg = fmt::format("Starting alchemical transformation at {:>3d}%\n", progress);
if (universe->uscreen) fmt::print(universe->uscreen, msg);
if (universe->ulogfile) fmt::print(universe->ulogfile, msg);
@ -175,16 +192,6 @@ void FixAlchemy::post_integrate()
/* ---------------------------------------------------------------------- */
static double get_lambda(const bigint &step, const bigint &begin, const bigint &end, int iworld)
{
double lambda = step - begin;
if (lambda != 0.0) lambda /= end - begin;
if (iworld == 0) lambda = 1.0 - lambda;
return lambda;
}
/* ---------------------------------------------------------------------- */
void FixAlchemy::post_force(int /*vflag*/)
{
if (3 * atom->nmax > nmax) {
@ -194,7 +201,7 @@ void FixAlchemy::post_force(int /*vflag*/)
const int nall = 3 * atom->nlocal;
double *f = &atom->f[0][0];
lambda = get_lambda(update->ntimestep, update->beginstep, update->endstep, universe->iworld);
lambda = input->variable->compute_equal(ivar);
for (int i = 0; i < nall; ++i) commbuf[i] = f[i] * lambda;
MPI_Allreduce(commbuf, f, nall, MPI_DOUBLE, MPI_SUM, samerank);
@ -218,7 +225,10 @@ void FixAlchemy::post_force(int /*vflag*/)
// print progress info
if (universe->me == 0) {
int status = static_cast<int>(100.0 - lambda * 100.0);
double delta = update->ntimestep - update->beginstep;
if ((delta != 0.0) && (update->beginstep != update->endstep))
delta /= update->endstep - update->beginstep;
int status = static_cast<int>(delta*100.0);
if ((status / 10) > (progress / 10)) {
progress = status;
auto msg = fmt::format(" Alchemical transformation progress: {:>3d}%\n", progress);

5
src/REPLICA/fix_alchemy.h
View File

@ -42,14 +42,15 @@ class FixAlchemy : public Fix {
MPI_Comm samerank;
double *commbuf;
class Compute *pe, *temp, *press;
std::string id_pe, id_temp, id_press;
double lambda; // changes from 0 to 1 during run
std::string id_pe, id_temp, id_press, id_lambda;
double lambda; // scaling prefactor for combining the partitions
double epot[3]; // last (unscaled) potential energy from each replica and combined energy
double pressure[6]; // joined pressure
int progress; // for progress indicator
int sync_box; // 1 of box dimensions need to be synchronized
int ilevel_respa;
int nmax;
int ivar;
};
} // namespace LAMMPS_NS