tx258/lammps-gran-kokkos
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'develop' into feature/batch-snap-ilp

Browse Source
This commit is contained in:
Evan Weinberg
2025-07-14 21:40:47 -04:00
committed by GitHub
parent 929bc35133 dcdd57c130
commit 8d5c50db1f
14 changed files with 2182 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
doc/src/Commands_fix.rst
View File

@ -117,6 +117,7 @@ OPT.
* :doc:`mvv/tdpd <fix_mvv_dpd>`
* :doc:`neb <fix_neb>`
* :doc:`neb/spin <fix_neb_spin>`
* :doc:`neighbor/swap <fix_neighbor_swap>`
* :doc:`nonaffine/displacement <fix_nonaffine_displacement>`
* :doc:`nph (ko) <fix_nh>`
* :doc:`nph/asphere (o) <fix_nph_asphere>`

32
doc/src/compute_stress_mop.rst
View File

@ -45,13 +45,13 @@ Examples
Description
"""""""""""
Compute *stress/mop* and compute *stress/mop/profile* define
computations that calculate components of the local stress tensor using
the method of planes :ref:`(Todd) <mop-todd>`. Specifically in compute
*stress/mop* calculates 3 components are computed in directions *dir*,\
*x*\ ; *dir*,\ *y*\ ; and *dir*,\ *z*\ ; where *dir* is the direction
normal to the plane, while in compute *stress/mop/profile* the profile
of the stress is computed.
Compute *stress/mop* and compute *stress/mop/profile*
calculate components of the local stress tensor using
the method of planes :ref:`(Todd) <mop-todd>`. Specifically, compute
*stress/mop* calculates 3 components in directions :math:`ix`,
:math:`iy`, and :math:`iz` where :math:`i` is the direction
normal to the plane, while compute *stress/mop/profile* calculates the profile
of the local stress along the :math:`i` direction.
Contrary to methods based on histograms of atomic stress (i.e., using
:doc:`compute stress/atom <compute_stress_atom>`), the method of planes
@ -103,14 +103,15 @@ Output info
Compute *stress/mop* calculates a global vector (indices starting at 1),
with 3 values for each declared keyword (in the order the keywords have
been declared). For each keyword, the stress tensor components are
ordered as follows: stress_dir,x, stress_dir,y, and stress_dir,z.
ordered as follows: :math:`P_{ix}`, :math:`P_{iy}`, and :math:`P_{iz}`,
where :math:`i` is the direction normal to the plane.
Compute *stress/mop/profile* instead calculates a global array, with 1
column giving the position of the planes where the stress tensor was
computed, and with 3 columns of values for each declared keyword (in the
order the keywords have been declared). For each keyword, the profiles
of stress tensor components are ordered as follows: stress_dir,x;
stress_dir,y; and stress_dir,z.
of stress tensor components are ordered as follows: :math:`P_{ix}`,
:math:`P_{iy}`, and :math:`P_{iz}`.
The values are in pressure :doc:`units <units>`.
@ -129,12 +130,11 @@ package <Build_package>` doc page on for more info.
The method is implemented for orthogonal simulation boxes whose
size does not change in time, and axis-aligned planes.
The method only works with two-body pair interactions, because it
requires the class method ``Pair::single()`` to be implemented, which is
not possible for manybody potentials. In particular, compute
*stress/mop/profile* and *stress/mop* do not work with more than two-body
pair interactions, long range (kspace) interactions and
improper intramolecular interactions.
Compute *stress/mop* and *stress/mop/profile* do not work with manybody
non-bonded interactions, long range (kspace) interactions and
improper intramolecular interactions. The reason is that the current
implementation requires the class method ``Pair::single()`` to be implemented,
which is not possible for manybody potentials.
The impact of fixes that affect the stress (e.g. fix langevin) is
also not included in the stress computed here.

1
doc/src/fix.rst
View File

@ -296,6 +296,7 @@ accelerated styles exist.
* :doc:`mvv/tdpd <fix_mvv_dpd>` - constant temperature DPD using the modified velocity-Verlet algorithm
* :doc:`neb <fix_neb>` - nudged elastic band (NEB) spring forces
* :doc:`neb/spin <fix_neb_spin>` - nudged elastic band (NEB) spring forces for spins
* :doc:`neighbor/swap <fix_neighbor_swap>` - kinetic Monte Carlo (kMC) atom swapping
* :doc:`nonaffine/displacement <fix_nonaffine_displacement>` - calculate nonaffine displacement of atoms
* :doc:`nph <fix_nh>` - constant NPH time integration via Nose/Hoover
* :doc:`nph/asphere <fix_nph_asphere>` - NPH for aspherical particles

264
doc/src/fix_neighbor_swap.rst Normal file
View File

@ -0,0 +1,264 @@
.. index:: fix neighbor/swap
fix neighbor/swap command
=========================
Syntax
""""""
.. code-block:: LAMMPS
fix ID group-ID neighbor/swap N X seed T R0 voro-ID keyword values ...
* ID, group-ID are documented in :doc:`fix <fix>` command
* neighbor/swap = style name of this fix command
* N = invoke this fix every N steps
* X = number of swaps to attempt every N steps
* seed = random # seed (positive integer)
* T = scaling temperature of the MC swaps (temperature units)
* R0 = scaling swap probability of the MC swaps (distance units)
* voro-ID = valid voronoi compute id (compute voronoi/atom)
* one or more keyword/value pairs may be appended to args
* keywords *types* and *diff* are mutually exclusive, but one must be specified
* keyword = *types* or *diff* or *ke* or *region* or *rates*
.. parsed-literal::
*types* values = two or more atom types (Integers in range [1,Ntypes] or type labels)
*diff* values = one atom type
*ke* value = *yes* or *no*
*yes* = kinetic energy is conserved after atom swaps
*no* = no conservation of kinetic energy after atom swaps
*region* value = region-ID
region-ID = ID of region to use as an exchange/move volume
*rates* values = V1 V2 . . . Vntypes values to conduct variable diffusion for different atom types (unitless)
Examples
""""""""
.. code-block:: LAMMPS
compute voroN all voronoi/atom neighbors yes
fix mc all neighbor/swap 10 160 15238 1000.0 3.0 voroN diff 2
fix myFix all neighbor/swap 100 1 12345 298.0 3.0 voroN region my_swap_region types 5 6
fix kmc all neighbor/swap 1 100 345 1.0 3.0 voroN diff 3 rates 3 1 6
Description
"""""""""""
.. versionadded:: TBD
This fix performs Monte-Carlo (MC) evaluations to enable kinetic
Monte Carlo (kMC)-type behavior during MD simulation by allowing
neighboring atoms to swap their positions. In contrast to the :doc:`fix
atom/swap <fix_atom_swap>` command which swaps pairs of atoms anywhere
in the simulation domain, the restriction of the MC swapping to
neighbors enables a hybrid MD/kMC-like simulation.
Neighboring atoms are defined by using a Voronoi tesselation performed
by the :doc:`compute voronoi/atom <compute_voronoi_atom>` command.
Two atoms are neighbors if their Voronoi cells share a common face
(3d) or edge (2d).
The selection of a swap neighbor is made using a distance-based
criterion for weighting the selection probability of each swap, in the
same manner as kMC selects a next event using relative probabilities.
The acceptance or rejection of each swap is determined via the
Metropolis criterion after evaluating the change in system energy due
to the swap.
A detailed explanation of the original implementation of this
algorithm can be found in :ref:`(Tavenner 2023) <TavennerMDkMC>`
where it was used to simulated accelerated diffusion in an MD context.
Simulating inherently kinetically-limited behaviors which rely on rare
events (such as atomic diffusion in a solid) is challenging for
traditional MD since its relatively short timescale will not naturally
sample many events. This fix addresses this challenge by allowing rare
neighbor hopping events to be sampled in a kMC-like fashion at a much
faster rate (set by the specified *N* and *X* parameters). This enables
the processes of atomic diffusion to be approximated during an MD
simulation, effectively decoupling the MD atomic vibrational timescale
and the atomic hopping (kMC event) timescale.
The algorithm implemented by this fix is as follows:
- The MD simulation is paused every *N* steps
- A Voronoi tesselation is performed for the current atom configuration.
- Then *X* atom swaps are attempted, one after the other.
- For each swap, an atom *I* is selected randomly from the list of
atom types specified by either the *types* or *diff* keywords.
- One of *I*'s Voronoi neighbors *J* is selected using the
distance-weighted probability for each neighbor detailed below.
- The *I,J* atom IDs are communicated to all processors so that a
global energy evaluation can be performed for the post-swap state
of the system.
- The swap is accepted or rejected based on the Metropolis criterion
using the energy change of the system and the specified temperature
*T*.
Here are a few comments on the computational cost of the swapping
algorithm.
1. The cost of a global energy evaluation is similar to that of an MD
timestep.
2. Similar to other MC algorithms in LAMMPS, improved parallel
efficiency is achieved with a smaller number of atoms per
processor than would typically be used in an standard MD
simulation. This is because the per-energy evaluation cost
increases relative to the balance of MD/MC steps as indicated by
1., but the communication cost remains relatively constant for a
given number of MD steps.
3. The MC portion of the simulation will run dramatically slower if
the pair style uses different cutoffs for different atom types (or
type pairs). This is because each atom swap then requires a
rebuild of the neighbor list to ensure the post-swap global energy
can be computed correctly.
Limitations are imposed on selection of *I,J* atom pairs to avoid
swapping of atoms which are outside of a reasonable cutoff (e.g. due to
a Voronoi tesselation near free surfaces) though the use of a
distance-weighted probability scaling.
----------
This section gives more details on other arguments and keywords.
The random number generator (RNG) used by all the processors for MC
operations is initialized with the specified *seed*.
The distance-based probability is weighted by the specified *R0* which
sets the radius :math:`r_0` in this formula
.. math::
p_{ij} = e^{(\frac{r_{ij}}{r_0})^2}
where :math:`p_{ij}` is the probability of selecting atom :math:`j` to
swap with atom :math:`i`. Typically, a value for *R0* around the
average nearest-neighbor spacing is appropriate. Since this is simply a
probability weighting, the swapping behavior is not very sensitive to
the exact value of *R0*.
The required *voro-ID* value is the compute-ID of a
:doc:`compute voronoi/atom <compute_voronoi_atom>` command like
this:
.. code-block:: LAMMPS
compute compute-ID group-ID voronoi/atom neighbors yes
It must return per-atom list of valid neighbor IDs as in the
:doc:`compute voronoi/atom <compute_voronoi_atom>` command.
The keyword *types* takes two or more atom types as its values. Only
atoms *I* of the first atom type will be selected. Only atoms *J* of the
remaining atom types will be considered as potential swap partners.
The keyword *diff* take a single atom type as its value. Only atoms
*I* of the that atom type will be selected. Atoms *J* of all
remaining atom types will be considered as potential swap partners.
This includes the atom type specified with the *diff* keyword to
account for self-diffusive hops between two atoms of the same type.
Note that the *neighbors yes* option must be enabled for use with this
fix. The group-ID should include all the atoms which this fix will
potentially select. I.e. the group-ID used in the voronoi compute should
include the same atoms as that indicated by the *types* keyword. If the
*diff* keyword is used, the group-ID should include atoms of all types
in the simulation.
The keyword *ke* takes *yes* (default) or *no* as its value. It two
atoms are swapped with different masses, then a value of *yes* will
rescale their respective velocities to conserve the kinetic energy of
the system. A value of *no* will perform no rescaling, so that
kinetic energy is not conserved. See the restriction on this keyword
below.
The *region* keyword takes a *region-ID* as its value. If specified,
then only atoms *I* and *J* within the geometric region will be
considered as swap partners. See the :doc:`region <region>` command
for details. This means the group-ID for the :doc:`compute
voronoi/atom <compute_voronoi_atom>` command also need only contain
atoms within the region.
The keyword *rates* can modify the swap rate based on the type of atom
*J*. Ntype values must be specified, where Ntype = the number of atom
types in the system. Each value is used to scale the probability
weighting given by the equation above. In the third example command
above, a simulation has 3 atoms types. Atom *I*s of type 1 are
eligible for swapping. Swaps may occur with atom *J*s of all 3 types.
Assuming all *J* atoms are equidistant from an atom *I*, *J* atoms of
type 1 will be 3x more likely to be selected as a swap partner than
atoms of type 2. And *J* atoms of type 3 will be 6.5x more likely to
be selected than atoms of type 2. If the *rates* keyword is not used,
all atom types will be treated with the same probability during selection
of swap attempts.
Restart, fix_modify, output, run start/stop, minimize info
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
This fix writes the state of the fix to :doc:`binary restart files
<restart>`. This includes information about the random number generator
seed, the next timestep for MC exchanges, and the number of exchange
attempts and successes. See the :doc:`read_restart <read_restart>`
command for info on how to re-specify a fix in an input script that
reads a restart file, so that the operation of the fix continues in an
uninterrupted fashion.
None of the :doc:`fix_modify <fix_modify>` options are relevant to this
fix.
This fix computes a global vector of length 2, which can be accessed
by various :doc:`output commands <Howto_output>`. The vector values are
the following global cumulative quantities:
#. swap attempts
#. swap accepts
The vector values calculated by this fix are "intensive".
No parameter of this fix can be used with the *start/stop* keywords of
the :doc:`run <run>` command. This fix is not invoked during
:doc:`energy minimization <minimize>`.
Restrictions
""""""""""""
This fix is part of the MC package. It is only enabled if LAMMPS was
built with that package. See the :doc:`Build package <Build_package>`
doc page for more info. Also this fix requires that the :ref:`VORONOI
package <PKG-VORONOI>` is installed, otherwise the fix will not be
compiled.
The :doc:`compute voronoi/atom <compute_voronoi_atom>` command
referenced by the required voro-ID must return neighboring atoms as
illustrated in the examples above.
If this fix is used with systems that do not have per-type masses
(e.g. atom style sphere), the *ke* keyword must be set to *off* since
the implemented algorithm will not be able to re-scale velocities
properly.
Related commands
""""""""""""""""
:doc:`fix nvt <fix_nh>`, :doc:`compute voronoi/atom <compute_voronoi_atom>`
:doc:`delete_atoms <delete_atoms>`, :doc:`fix gcmc <fix_gcmc>`,
:doc:`fix atom/swap <fix_atom_swap>`, :doc:`fix mol/swap <fix_mol_swap>`,
:doc:`fix sgcmc <fix_sgcmc>`
Default
"""""""
The option defaults are *ke* = yes and *rates* = 1 for all atom types.
----------
.. _TavennerMDkMC:
**(Tavenner 2023)** J Tavenner, M Mendelev, J Lawson, Computational
Materials Science, 218, 111929 (2023).

1
examples/PACKAGES/neighbor-swap/MoCoNiVFeAlCr_2nn.meam Symbolic link
View File

@ -0,0 +1 @@
../../../potentials/MoCoNiVFeAlCr_2nn.meam

46
examples/PACKAGES/neighbor-swap/in.KMC_pulse_center Normal file
View File

@ -0,0 +1,46 @@
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
create_atoms 1 region whole
replicate 6 16 6
region puck block INF INF 7 9 INF INF
set region puck type 2
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
#write_data pulse_center.data

47
examples/PACKAGES/neighbor-swap/in.KMC_pulse_edge Normal file
View File

@ -0,0 +1,47 @@
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
create_atoms 1 region whole
replicate 6 16 6
region puck block INF INF INF 2 INF INF
set region puck type 2
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
#write_data pulse_end.data

1
examples/PACKAGES/neighbor-swap/library_2nn.meam Symbolic link
View File

@ -0,0 +1 @@
../../../potentials/library_2nn.meam

174
examples/PACKAGES/neighbor-swap/log.11Jul25.KMC_pulse_center.g++.1 Normal file
View File

@ -0,0 +1,174 @@
LAMMPS (12 Jun 2025 - Development - patch_12Jun2025-605-g5eb61491f0-modified)
OMP_NUM_THREADS environment is not set. Defaulting to 1 thread.
using 1 OpenMP thread(s) per MPI task
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
Lattice spacing in x,y,z = 3.762 3.762 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
Created orthogonal box = (0 0 0) to (3.762 3.762 3.762)
1 by 1 by 1 MPI processor grid
create_atoms 1 region whole
Created 4 atoms
using lattice units in orthogonal box = (0 0 0) to (3.762 3.762 3.762)
create_atoms CPU = 0.000 seconds
replicate 6 16 6
Replication is creating a 6x16x6 = 576 times larger system...
orthogonal box = (0 0 0) to (22.572 60.192 22.572)
1 by 1 by 1 MPI processor grid
2304 atoms
replicate CPU = 0.000 seconds
region puck block INF INF 7 9 INF INF
set region puck type 2
Setting atom values ...
360 settings made for type
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
Reading MEAM library file library_2nn.meam with DATE: 2024-08-08
Reading MEAM potential file MoCoNiVFeAlCr_2nn.meam with DATE: 2024-08-08
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
WARNING: No fixes with time integration, atoms won't move
For more information see https://docs.lammps.org/err0028 (src/verlet.cpp:60)
Neighbor list info ...
update: every = 1 steps, delay = 0 steps, check = yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 6.8
ghost atom cutoff = 6.8
binsize = 3.4, bins = 7 18 7
2 neighbor lists, perpetual/occasional/extra = 2 0 0
(1) pair meam, perpetual
attributes: full, newton on
pair build: full/bin/atomonly
stencil: full/bin/3d
bin: standard
(2) pair meam, perpetual, half/full from (1)
attributes: half, newton on
pair build: halffull/newton
stencil: none
bin: none
Per MPI rank memory allocation (min/avg/max) = 13.32 | 13.32 | 13.32 Mbytes
Step Temp E_pair E_mol TotEng Press
0 0 -9674.3728 0 -9674.3728 -212400.94
Loop time of 8.62e-07 on 1 procs for 0 steps with 2304 atoms
0.0% CPU use with 1 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 8.62e-07 | | |100.00
Nlocal: 2304 ave 2304 max 2304 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 4735 ave 4735 max 4735 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 99072 ave 99072 max 99072 min
Histogram: 1 0 0 0 0 0 0 0 0 0
FullNghs: 198144 ave 198144 max 198144 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 198144
Ave neighs/atom = 86
Neighbor list builds = 0
Dangerous builds = 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Your simulation uses code contributions which should be cited:
- fix neighbor/swap command: doi:10.1016/j.commatsci.2022.111929
@Article{Tavenner2023111929,
author = {Jacob P. Tavenner and Mikhail I. Mendelev and John W. Lawson},
title = {Molecular dynamics based kinetic Monte Carlo simulation for accelerated diffusion},
journal = {Computational Materials Science},
year = {2023},
volume = {218},
pages = {111929}
url = {https://dx.doi.org/10.1016/j.commatsci.2022.111929}
}
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Per MPI rank memory allocation (min/avg/max) = 13.32 | 13.32 | 13.32 Mbytes
Step Temp Press Pxx Pyy Pzz Lx Ly Lz Volume PotEng f_mc[1] f_mc[2]
0 2400 -187517.52 -187464.28 -188202.86 -186885.42 22.572 60.192 22.572 30667.534 -9674.3728 0 0
100 1687.1584 13681.454 14112.794 14026.902 12904.667 21.637683 57.728817 21.63703 27027.2 -9600.6424 24 23
200 1578.9879 -6074.2872 -7170.074 -4456.4001 -6596.3875 21.71812 58.13626 21.712735 27414.722 -9569.7632 48 44
300 1586.0502 1921.5002 2871.8396 1449.0001 1443.661 21.677891 58.13829 21.643595 27277.759 -9572.2349 72 67
400 1599.8151 459.13309 -890.62191 638.40686 1629.6143 21.662475 58.208733 21.64368 27291.495 -9576.3711 96 88
500 1622.7406 -980.83045 -358.66202 -2883.0275 299.19813 21.627475 58.244052 21.683194 27313.708 -9583.2244 120 110
600 1558.3072 661.01491 2134.4817 -1069.4315 917.99455 21.675158 58.08722 21.705367 27328.136 -9564.1953 144 132
700 1585.7908 2896.8915 2712.2516 5131.3298 847.09324 21.703761 57.89964 21.704281 27274.466 -9572.077 168 154
800 1593.7707 -754.08728 -529.51328 807.24585 -2539.9944 21.682927 58.129291 21.67599 27320.704 -9574.7174 192 175
900 1594.1026 -504.80889 147.79949 -1421.9393 -240.28691 21.678023 58.172814 21.655846 27309.572 -9574.7989 216 198
1000 1604.8161 1017.6914 707.59987 2068.9348 276.53957 21.696786 57.985179 21.676799 27271.408 -9578.0587 240 221
Loop time of 30.6062 on 1 procs for 1000 steps with 2304 atoms
Performance: 5.646 ns/day, 4.251 hours/ns, 32.673 timesteps/s, 75.279 katom-step/s
99.6% CPU use with 1 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 27.514 | 27.514 | 27.514 | 0.0 | 89.90
Neigh | 0.22106 | 0.22106 | 0.22106 | 0.0 | 0.72
Comm | 0.011786 | 0.011786 | 0.011786 | 0.0 | 0.04
Output | 0.00030095 | 0.00030095 | 0.00030095 | 0.0 | 0.00
Modify | 2.854 | 2.854 | 2.854 | 0.0 | 9.32
Other | | 0.004794 | | | 0.02
Nlocal: 2304 ave 2304 max 2304 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 4744 ave 4744 max 4744 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 130039 ave 130039 max 130039 min
Histogram: 1 0 0 0 0 0 0 0 0 0
FullNghs: 260078 ave 260078 max 260078 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 260078
Ave neighs/atom = 112.88108
Neighbor list builds = 62
Dangerous builds = 0
#write_data pulse_center.data
Total wall time: 0:00:30

174
examples/PACKAGES/neighbor-swap/log.11Jul25.KMC_pulse_center.g++.4 Normal file
View File

@ -0,0 +1,174 @@
LAMMPS (12 Jun 2025 - Development - patch_12Jun2025-605-g5eb61491f0-modified)
OMP_NUM_THREADS environment is not set. Defaulting to 1 thread.
using 1 OpenMP thread(s) per MPI task
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
Lattice spacing in x,y,z = 3.762 3.762 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
Created orthogonal box = (0 0 0) to (3.762 3.762 3.762)
1 by 2 by 2 MPI processor grid
create_atoms 1 region whole
Created 4 atoms
using lattice units in orthogonal box = (0 0 0) to (3.762 3.762 3.762)
create_atoms CPU = 0.000 seconds
replicate 6 16 6
Replication is creating a 6x16x6 = 576 times larger system...
orthogonal box = (0 0 0) to (22.572 60.192 22.572)
1 by 4 by 1 MPI processor grid
2304 atoms
replicate CPU = 0.000 seconds
region puck block INF INF 7 9 INF INF
set region puck type 2
Setting atom values ...
360 settings made for type
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
Reading MEAM library file library_2nn.meam with DATE: 2024-08-08
Reading MEAM potential file MoCoNiVFeAlCr_2nn.meam with DATE: 2024-08-08
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
WARNING: No fixes with time integration, atoms won't move
For more information see https://docs.lammps.org/err0028 (src/verlet.cpp:60)
Neighbor list info ...
update: every = 1 steps, delay = 0 steps, check = yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 6.8
ghost atom cutoff = 6.8
binsize = 3.4, bins = 7 18 7
2 neighbor lists, perpetual/occasional/extra = 2 0 0
(1) pair meam, perpetual
attributes: full, newton on
pair build: full/bin/atomonly
stencil: full/bin/3d
bin: standard
(2) pair meam, perpetual, half/full from (1)
attributes: half, newton on
pair build: halffull/newton
stencil: none
bin: none
Per MPI rank memory allocation (min/avg/max) = 9.636 | 9.636 | 9.636 Mbytes
Step Temp E_pair E_mol TotEng Press
0 0 -9674.3728 0 -9674.3728 -212400.94
Loop time of 4.3735e-06 on 4 procs for 0 steps with 2304 atoms
0.0% CPU use with 4 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 4.374e-06 | | |100.00
Nlocal: 576 ave 576 max 576 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Nghost: 2131 ave 2131 max 2131 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Neighs: 24768 ave 24768 max 24768 min
Histogram: 4 0 0 0 0 0 0 0 0 0
FullNghs: 49536 ave 49536 max 49536 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Total # of neighbors = 198144
Ave neighs/atom = 86
Neighbor list builds = 0
Dangerous builds = 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Your simulation uses code contributions which should be cited:
- fix neighbor/swap command: doi:10.1016/j.commatsci.2022.111929
@Article{Tavenner2023111929,
author = {Jacob P. Tavenner and Mikhail I. Mendelev and John W. Lawson},
title = {Molecular dynamics based kinetic Monte Carlo simulation for accelerated diffusion},
journal = {Computational Materials Science},
year = {2023},
volume = {218},
pages = {111929}
url = {https://dx.doi.org/10.1016/j.commatsci.2022.111929}
}
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Per MPI rank memory allocation (min/avg/max) = 9.636 | 9.636 | 9.636 Mbytes
Step Temp Press Pxx Pyy Pzz Lx Ly Lz Volume PotEng f_mc[1] f_mc[2]
0 2400 -187517.52 -187464.28 -188202.86 -186885.42 22.572 60.192 22.572 30667.534 -9674.3728 0 0
100 1688.7809 13746.894 14218.282 14190.983 12831.417 21.636615 57.726346 21.637269 27025.008 -9601.4799 24 24
200 1585.0272 -6497.2882 -7280.7893 -5297.3951 -6913.6802 21.716781 58.142867 21.714094 27417.864 -9571.8204 48 45
300 1592.737 1747.6515 2738.4114 1924.1359 580.40727 21.676942 58.130592 21.647155 27277.439 -9574.5368 72 68
400 1594.154 852.39952 -369.56275 1867.5224 1059.2389 21.659314 58.220677 21.640495 27289.096 -9574.1977 96 89
500 1616.8485 -1466.6184 -1965.5854 -2117.4507 -316.81912 21.622512 58.328889 21.660903 27319.101 -9580.0369 120 112
600 1560.0956 1162.0193 2410.2796 -925.45844 2001.2369 21.637185 58.168753 21.700261 27312.124 -9564.0088 144 136
700 1583.4354 1031.9177 243.6789 5034.5413 -2182.467 21.682521 57.96166 21.718651 27295.022 -9570.2301 168 158
800 1618.7711 -2148.2603 -4066.0779 -863.23022 -1515.4726 21.694024 58.144275 21.659535 27320.975 -9581.5884 192 181
900 1598.9132 -764.84277 413.68636 -2395.8324 -312.3823 21.650745 58.249779 21.651506 27305.821 -9575.9563 216 204
1000 1603.2739 240.94022 606.79481 644.45464 -528.42878 21.661235 58.138319 21.674995 27296.357 -9577.7462 240 228
Loop time of 8.90243 on 4 procs for 1000 steps with 2304 atoms
Performance: 19.410 ns/day, 1.236 hours/ns, 112.329 timesteps/s, 258.806 katom-step/s
98.9% CPU use with 4 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 7.5993 | 7.7187 | 7.8278 | 3.8 | 86.70
Neigh | 0.056181 | 0.059129 | 0.062043 | 1.1 | 0.66
Comm | 0.052658 | 0.14746 | 0.25458 | 24.3 | 1.66
Output | 0.00018588 | 0.00019825 | 0.00023195 | 0.0 | 0.00
Modify | 0.95669 | 0.97388 | 0.99047 | 1.6 | 10.94
Other | | 0.00311 | | | 0.03
Nlocal: 576 ave 619 max 531 min
Histogram: 1 0 1 0 0 0 0 0 1 1
Nghost: 2159.25 ave 2181 max 2126 min
Histogram: 1 0 0 0 0 0 1 0 1 1
Neighs: 32404.8 ave 36204 max 28467 min
Histogram: 1 0 1 0 0 0 0 0 1 1
FullNghs: 64809.5 ave 72517 max 56929 min
Histogram: 1 1 0 0 0 0 0 0 1 1
Total # of neighbors = 259238
Ave neighs/atom = 112.51649
Neighbor list builds = 62
Dangerous builds = 0
#write_data pulse_center.data
Total wall time: 0:00:08

175
examples/PACKAGES/neighbor-swap/log.11Jul25.KMC_pulse_edge.g++.1 Normal file
View File

@ -0,0 +1,175 @@
LAMMPS (12 Jun 2025 - Development - patch_12Jun2025-605-g5eb61491f0-modified)
OMP_NUM_THREADS environment is not set. Defaulting to 1 thread.
using 1 OpenMP thread(s) per MPI task
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
Lattice spacing in x,y,z = 3.762 3.762 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
Created orthogonal box = (0 0 0) to (3.762 3.762 3.762)
1 by 1 by 1 MPI processor grid
create_atoms 1 region whole
Created 4 atoms
using lattice units in orthogonal box = (0 0 0) to (3.762 3.762 3.762)
create_atoms CPU = 0.000 seconds
replicate 6 16 6
Replication is creating a 6x16x6 = 576 times larger system...
orthogonal box = (0 0 0) to (22.572 60.192 22.572)
1 by 1 by 1 MPI processor grid
2304 atoms
replicate CPU = 0.000 seconds
region puck block INF INF INF 2 INF INF
set region puck type 2
Setting atom values ...
360 settings made for type
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
Reading MEAM library file library_2nn.meam with DATE: 2024-08-08
Reading MEAM potential file MoCoNiVFeAlCr_2nn.meam with DATE: 2024-08-08
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
WARNING: No fixes with time integration, atoms won't move
For more information see https://docs.lammps.org/err0028 (src/verlet.cpp:60)
Neighbor list info ...
update: every = 1 steps, delay = 0 steps, check = yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 6.8
ghost atom cutoff = 6.8
binsize = 3.4, bins = 7 18 7
2 neighbor lists, perpetual/occasional/extra = 2 0 0
(1) pair meam, perpetual
attributes: full, newton on
pair build: full/bin/atomonly
stencil: full/bin/3d
bin: standard
(2) pair meam, perpetual, half/full from (1)
attributes: half, newton on
pair build: halffull/newton
stencil: none
bin: none
Per MPI rank memory allocation (min/avg/max) = 13.32 | 13.32 | 13.32 Mbytes
Step Temp E_pair E_mol TotEng Press
0 0 -9674.3728 0 -9674.3728 -212400.94
Loop time of 7.92e-07 on 1 procs for 0 steps with 2304 atoms
126.3% CPU use with 1 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 7.92e-07 | | |100.00
Nlocal: 2304 ave 2304 max 2304 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 4735 ave 4735 max 4735 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 99072 ave 99072 max 99072 min
Histogram: 1 0 0 0 0 0 0 0 0 0
FullNghs: 198144 ave 198144 max 198144 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 198144
Ave neighs/atom = 86
Neighbor list builds = 0
Dangerous builds = 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Your simulation uses code contributions which should be cited:
- fix neighbor/swap command: doi:10.1016/j.commatsci.2022.111929
@Article{Tavenner2023111929,
author = {Jacob P. Tavenner and Mikhail I. Mendelev and John W. Lawson},
title = {Molecular dynamics based kinetic Monte Carlo simulation for accelerated diffusion},
journal = {Computational Materials Science},
year = {2023},
volume = {218},
pages = {111929}
url = {https://dx.doi.org/10.1016/j.commatsci.2022.111929}
}
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Per MPI rank memory allocation (min/avg/max) = 13.32 | 13.32 | 13.32 Mbytes
Step Temp Press Pxx Pyy Pzz Lx Ly Lz Volume PotEng f_mc[1] f_mc[2]
0 2400 -187517.52 -187464.47 -188202.62 -186885.48 22.572 60.192 22.572 30667.534 -9674.3728 0 0
100 1665.6154 14281.316 14426.547 14555.867 13861.534 21.637238 57.719793 21.637281 27022.733 -9594.4303 24 24
200 1603.3309 -7325.7341 -8878.1524 -5333.0485 -7766.0015 21.710246 58.122827 21.725933 27415.106 -9577.4545 48 48
300 1603.2974 207.19165 1983.4565 -1841.9518 480.07024 21.678227 58.079126 21.674033 27288.745 -9577.6391 72 69
400 1600.1515 810.95054 1087.969 802.04946 542.83316 21.683731 58.045848 21.678505 27285.662 -9576.6508 96 92
500 1629.8313 -2808.1005 -3197.9357 310.89931 -5537.265 21.683924 58.090375 21.697076 27330.229 -9585.5435 120 113
600 1598.8232 -67.845623 -1573.0718 -1526.7607 2896.2957 21.70213 58.12191 21.653853 27313.504 -9576.4147 144 137
700 1607.2185 154.66718 -1777.2469 2566.4705 -325.22208 21.712408 57.971553 21.678708 27287.033 -9579.1772 168 158
800 1582.559 -891.23631 -632.46037 -636.88203 -1404.3665 21.671936 58.127004 21.678224 27308.594 -9571.6663 192 180
900 1586.7172 -617.17083 -2495.5378 -2302.8766 2946.9018 21.658489 58.181921 21.668968 27305.771 -9572.9641 216 204
1000 1607.563 -389.8113 810.4908 298.84287 -2278.7676 21.624573 58.076745 21.724272 27283.183 -9579.5034 240 227
Loop time of 31.1311 on 1 procs for 1000 steps with 2304 atoms
Performance: 5.551 ns/day, 4.324 hours/ns, 32.122 timesteps/s, 74.010 katom-step/s
99.6% CPU use with 1 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 27.859 | 27.859 | 27.859 | 0.0 | 89.49
Neigh | 0.22271 | 0.22271 | 0.22271 | 0.0 | 0.72
Comm | 0.012597 | 0.012597 | 0.012597 | 0.0 | 0.04
Output | 0.00031219 | 0.00031219 | 0.00031219 | 0.0 | 0.00
Modify | 3.0301 | 3.0301 | 3.0301 | 0.0 | 9.73
Other | | 0.006178 | | | 0.02
Nlocal: 2304 ave 2304 max 2304 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 4748 ave 4748 max 4748 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 130301 ave 130301 max 130301 min
Histogram: 1 0 0 0 0 0 0 0 0 0
FullNghs: 260602 ave 260602 max 260602 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 260602
Ave neighs/atom = 113.10851
Neighbor list builds = 62
Dangerous builds = 0
#write_data pulse_end.data
Total wall time: 0:00:31

175
examples/PACKAGES/neighbor-swap/log.11Jul25.KMC_pulse_edge.g++.4 Normal file
View File

@ -0,0 +1,175 @@
LAMMPS (12 Jun 2025 - Development - patch_12Jun2025-605-g5eb61491f0-modified)
OMP_NUM_THREADS environment is not set. Defaulting to 1 thread.
using 1 OpenMP thread(s) per MPI task
# May 2025
# Test script for MD-KMC accelerated diffusion testing in LAMMPS
# Created by Jacob Tavenner, Baylor University
# Initiation -------------------------------------
units metal
dimension 3
boundary p p p
atom_style atomic
# Atom Definition --------------------------------
lattice fcc 3.762
Lattice spacing in x,y,z = 3.762 3.762 3.762
region whole block 0 1 0 1 0 1
create_box 2 whole
Created orthogonal box = (0 0 0) to (3.762 3.762 3.762)
1 by 2 by 2 MPI processor grid
create_atoms 1 region whole
Created 4 atoms
using lattice units in orthogonal box = (0 0 0) to (3.762 3.762 3.762)
create_atoms CPU = 0.001 seconds
replicate 6 16 6
Replication is creating a 6x16x6 = 576 times larger system...
orthogonal box = (0 0 0) to (22.572 60.192 22.572)
1 by 4 by 1 MPI processor grid
2304 atoms
replicate CPU = 0.000 seconds
region puck block INF INF INF 2 INF INF
set region puck type 2
Setting atom values ...
360 settings made for type
# Force Fields -----------------------------------
pair_style meam
pair_coeff * * library_2nn.meam Mo Co Ni V Fe Al Cr MoCoNiVFeAlCr_2nn.meam Ni Cr
Reading MEAM library file library_2nn.meam with DATE: 2024-08-08
Reading MEAM potential file MoCoNiVFeAlCr_2nn.meam with DATE: 2024-08-08
# Settings ---------------------------------------
timestep 0.002
thermo 100
# Computations -----------------------------------
compute voroN all voronoi/atom neighbors yes
run 0
WARNING: No fixes with time integration, atoms won't move
For more information see https://docs.lammps.org/err0028 (src/verlet.cpp:60)
Neighbor list info ...
update: every = 1 steps, delay = 0 steps, check = yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 6.8
ghost atom cutoff = 6.8
binsize = 3.4, bins = 7 18 7
2 neighbor lists, perpetual/occasional/extra = 2 0 0
(1) pair meam, perpetual
attributes: full, newton on
pair build: full/bin/atomonly
stencil: full/bin/3d
bin: standard
(2) pair meam, perpetual, half/full from (1)
attributes: half, newton on
pair build: halffull/newton
stencil: none
bin: none
Per MPI rank memory allocation (min/avg/max) = 9.636 | 9.636 | 9.636 Mbytes
Step Temp E_pair E_mol TotEng Press
0 0 -9674.3728 0 -9674.3728 -212400.94
Loop time of 1.84625e-06 on 4 procs for 0 steps with 2304 atoms
135.4% CPU use with 4 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 1.846e-06 | | |100.00
Nlocal: 576 ave 576 max 576 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Nghost: 2131 ave 2131 max 2131 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Neighs: 24768 ave 24768 max 24768 min
Histogram: 4 0 0 0 0 0 0 0 0 0
FullNghs: 49536 ave 49536 max 49536 min
Histogram: 4 0 0 0 0 0 0 0 0 0
Total # of neighbors = 198144
Ave neighs/atom = 86
Neighbor list builds = 0
Dangerous builds = 0
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe
# Execution --------------------------------------
velocity all create 2400 908124 loop geom
fix temp all npt temp 1000 1000 1000 aniso 0 0 1
fix mc all neighbor/swap 50 12 1340723 1000 3 voroN diff 2
thermo_style custom step temp press pxx pyy pzz lx ly lz vol pe f_mc[*]
#dump dump2 all custom 5000 dump.edge-3_Ni-Cr.* id type x y z c_eng c_csym
run 1000
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Your simulation uses code contributions which should be cited:
- fix neighbor/swap command: doi:10.1016/j.commatsci.2022.111929
@Article{Tavenner2023111929,
author = {Jacob P. Tavenner and Mikhail I. Mendelev and John W. Lawson},
title = {Molecular dynamics based kinetic Monte Carlo simulation for accelerated diffusion},
journal = {Computational Materials Science},
year = {2023},
volume = {218},
pages = {111929}
url = {https://dx.doi.org/10.1016/j.commatsci.2022.111929}
}
CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE-CITE
Per MPI rank memory allocation (min/avg/max) = 9.636 | 9.636 | 9.636 Mbytes
Step Temp Press Pxx Pyy Pzz Lx Ly Lz Volume PotEng f_mc[1] f_mc[2]
0 2400 -187517.52 -187464.47 -188202.62 -186885.48 22.572 60.192 22.572 30667.534 -9674.3728 0 0
100 1662.8388 14433.559 14653.105 14684.034 13963.538 21.637114 57.721378 21.63721 27023.232 -9593.5557 24 23
200 1592.9276 -6765.4603 -8210.8697 -4995.7657 -7089.7454 21.709482 58.120737 21.726324 27413.648 -9573.9313 48 45
300 1619.7013 -391.64606 982.6812 -1910.4062 -247.21322 21.674555 58.073042 21.673346 27280.4 -9583.2581 72 69
400 1595.5295 1320.1347 726.16024 2142.0732 1092.1705 21.687154 58.033385 21.679884 27285.845 -9575.3049 96 92
500 1613.2865 -2150.2729 -2059.2665 -196.97204 -4194.58 21.675436 58.092453 21.698792 27322.668 -9581.7322 120 114
600 1587.7089 685.3534 113.39179 523.41016 1419.2583 21.69641 58.089816 21.665296 27305.65 -9572.879 144 135
700 1596.2078 232.23434 -985.47338 1818.9814 -136.80497 21.712357 57.974481 21.690493 27303.182 -9575.5985 168 157
800 1544.7314 1176.3622 -1011.314 3132.5302 1407.8703 21.681458 58.02289 21.705773 27306.315 -9560.7678 192 179
900 1579.6589 45.000317 -72.21997 -1479.7462 1686.9671 21.649984 58.218959 21.671494 27315.607 -9570.1227 216 198
1000 1609.6666 95.736886 -795.90474 641.09574 442.01966 21.651706 58.158209 21.667909 27284.76 -9579.3522 240 221
Loop time of 9.05461 on 4 procs for 1000 steps with 2304 atoms
Performance: 19.084 ns/day, 1.258 hours/ns, 110.441 timesteps/s, 254.456 katom-step/s
98.8% CPU use with 4 MPI tasks x 1 OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 7.4742 | 7.7144 | 7.8285 | 5.1 | 85.20
Neigh | 0.051898 | 0.058573 | 0.061819 | 1.6 | 0.65
Comm | 0.063429 | 0.15788 | 0.37324 | 31.7 | 1.74
Output | 0.00018368 | 0.00019631 | 0.00023307 | 0.0 | 0.00
Modify | 1.098 | 1.1206 | 1.1519 | 1.9 | 12.38
Other | | 0.002936 | | | 0.03
Nlocal: 576 ave 612 max 506 min
Histogram: 1 0 0 0 0 0 0 1 1 1
Nghost: 2165.5 ave 2211 max 2128 min
Histogram: 1 1 0 0 0 0 1 0 0 1
Neighs: 32568.8 ave 35314 max 26622 min
Histogram: 1 0 0 0 0 0 0 1 0 2
FullNghs: 65137.5 ave 70664 max 53385 min
Histogram: 1 0 0 0 0 0 0 1 0 2
Total # of neighbors = 260550
Ave neighs/atom = 113.08594
Neighbor list builds = 62
Dangerous builds = 0
#write_data pulse_end.data
Total wall time: 0:00:09

1008
src/MC/fix_neighbor_swap.cpp Normal file
View File

File diff suppressed because it is too large Load Diff

99
src/MC/fix_neighbor_swap.h Normal file
View File

@ -0,0 +1,99 @@
/* -*- c++ -*- ----------------------------------------------------------
LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
https://www.lammps.org/, Sandia National Laboratories
LAMMPS development team: developers@lammps.org
Copyright (2003) Sandia Corporation. Under the terms of Contract
DE-AC04-94AL85000 with Sandia Corporation, the U.S. Government retains
certain rights in this software. This software is distributed under
the GNU General Public License.
See the README file in the top-level LAMMPS directory.
------------------------------------------------------------------------- */
#ifdef FIX_CLASS
// clang-format off
FixStyle(neighbor/swap,FixNeighborSwap);
// clang-format on
#else
#ifndef LMP_FIX_NEIGHBOR_SWAP_H
#define LMP_FIX_NEIGHBOR_SWAP_H
#include "fix.h"
namespace LAMMPS_NS {
class FixNeighborSwap : public Fix {
public:
FixNeighborSwap(class LAMMPS *, int, char **);
~FixNeighborSwap();
int setmask();
void init();
void pre_exchange();
int pack_forward_comm(int, int *, double *, int, int *);
void unpack_forward_comm(int, int, double *);
double compute_vector(int);
double memory_usage();
void write_restart(FILE *);
void restart(char *);
private:
int nevery, seed;
int ke_flag; // yes = conserve ke, no = do not conserve ke
int diff_flag; // yes = simulate diffusion of central atom, no = swap only to certain types
int rates_flag; // yes = use modified type rates, no = swap rates are equivalent across types
int ncycles;
int niswap, njswap; // # of i,j swap atoms on all procs
int niswap_local, njswap_local; // # of swap atoms on this proc
int niswap_before, njswap_before; // # of swap atoms on procs < this proc
class Region *region; // swap region
char *idregion; // swap region id
int nswaptypes;
int jtype_selected;
tagint id_center;
double x_center;
double y_center;
double z_center;
int *type_list;
double *rate_list;
double nswap_attempts;
double nswap_successes;
bool unequal_cutoffs;
int atom_swap_nmax;
double beta, inv_r_0;
double local_probability; // Total swap probability stored on this proc
double global_probability; // Total swap probability across all proc
double prev_probability; // Swap probability on proc < this proc
double *qtype, *mtype;
double energy_stored;
double **sqrt_mass_ratio;
double **voro_neighbor_list;
int *local_swap_iatom_list;
int *local_swap_neighbor_list;
int *local_swap_type_list; // Type list index of atoms stored on this proc
double *local_swap_probability;
class RanPark *random_equal;
char *id_voro;
class Compute *c_voro, *c_pe;
void options(int, char **);
int attempt_swap();
double energy_full();
int pick_i_swap_atom();
int pick_j_swap_neighbor();
void build_i_neighbor_list(int);
void update_iswap_atoms_list();
};
} // namespace LAMMPS_NS
#endif
#endif