tweaks to src files, added examples dir
This commit is contained in:
Steve Plimpton
@ -83,6 +83,7 @@ hyper: global and local hyperdynamics of diffusion on Pt surface
|
||||
indent: spherical indenter into a 2d solid
|
||||
kim: use of potentials in Knowledge Base for Interatomic Models (KIM)
|
||||
latte: use of LATTE density-functional tight-binding quantum code
|
||||
mdi: use of the MDI package and MolSSI MDI code coupling library
|
||||
meam: MEAM test for SiC and shear (same as shear examples)
|
||||
melt: rapid melt of 3d LJ system
|
||||
message: client/server coupling of 2 codes
|
||||
|
||||
113
examples/mdi/README
Normal file
113
examples/mdi/README
Normal file
@ -0,0 +1,113 @@
|
||||
These are examples that work the MDI package in LAMMPS which uses the
|
||||
MolSSI MDI library for coupling codes together and communicating
|
||||
between them with MDI messages.
|
||||
|
||||
To use the serial_drive.py example you will need Python 3 with Numpy
|
||||
and mpi4py available in your Python. Make sure LAMMPS and Python are
|
||||
using same the same version of MPI.
|
||||
|
||||
In MDI lingo, one code is the driver and another code is the engine.
|
||||
The 2 codes can be written in any language; C++ (LAMMPS) and Python
|
||||
are illustrated here. The 2 codes can be run on different numbers of
|
||||
processors.
|
||||
|
||||
The 2 codes can communicate either via TCP (sockets) or via MPI. For
|
||||
the TCP case, the driver and engine need to be launched separately,
|
||||
e.g. in 2 windows on your desktop machine. For the MPI case, a single
|
||||
mpirun command launches both codes.
|
||||
|
||||
The example run commands below have variants for these options.
|
||||
|
||||
-------------------------------------------------
|
||||
-------------------------------------------------
|
||||
|
||||
* Example #1 = AIMD with LAMMPS as both a driver and engine
|
||||
As an engine, LAMMPS is acting as a surrogate for a quantum code.
|
||||
|
||||
Note that the 2 input scripts in.aimd.alone and in.aimd.driver
|
||||
have an option for running in NVE vs NPT mode. Comment in/out
|
||||
the appropriate line to change modes. Nothing needs to be
|
||||
changed in the 3rd input script in.aimd.engine.
|
||||
|
||||
---
|
||||
|
||||
Run the entire calculation with a single instance of LAMMPS by itself
|
||||
results should be identical to running in driver/engine mode
|
||||
|
||||
% lmp_mpi < in.aimd.alone
|
||||
|
||||
---
|
||||
|
||||
Run with TCP: 1 proc each
|
||||
|
||||
% lmp_mpi -mdi "-name driver -role DRIVER -method TCP -port 8021" -log log.aimd.driver -in in.aimd.driver
|
||||
|
||||
% lmp_mpi -mdi "-name LAMMPS -role ENGINE -method TCP -port 8021 -hostname localhost" -log log.aimd.engine -in in.aimd.engine
|
||||
|
||||
---
|
||||
|
||||
Run with TCP: 3 procs + 4 procs
|
||||
|
||||
% mpirun -np 3 lmp_mpi -mdi "-name driver -role DRIVER -method TCP -port 8021" -log log.aimd.driver -in in.aimd.driver
|
||||
|
||||
% mpirun -np 4 lmp_mpi -mdi "-name LAMMPS -role ENGINE -method TCP -port 8021 -hostname localhost" -log log.aimd.engine -in in.aimd.engine
|
||||
|
||||
---
|
||||
|
||||
Run with MPI: 1 proc each
|
||||
|
||||
% mpirun -np 1 lmp_mpi -mdi "-name driver -role DRIVER -method MPI" -log log.aimd.driver -in in.aimd.driver : -np 1 ../lammps/git/src/lmp_mpi -mdi "-name LAMMPS -role ENGINE -method MPI" -log log.aimd.engine -in in.aimd.engine
|
||||
|
||||
---
|
||||
|
||||
Run with MPI: 3 procs + 4 procs
|
||||
|
||||
% mpirun -np 3 lmp_mpi -mdi "-name driver -role DRIVER -method MPI" -log log.aimd.driver -in in.aimd.driver : -np 4 ../lammps/git/src/lmp_mpi -mdi "-name LAMMPS -role ENGINE -method MPI" -log log.aimd.engine -in in.aimd.engine
|
||||
|
||||
-------------------------------------------------
|
||||
-------------------------------------------------
|
||||
|
||||
* Example #2 = Use a Python driver code to run a series of independent
|
||||
LAMMPS calculations
|
||||
|
||||
Note that the series_driver.py code allows for optional arguments in
|
||||
addition to -mdi (required). The example run commands below just
|
||||
using the default values. The options are explained the top of the file;
|
||||
the info is copied here:
|
||||
|
||||
# -n 10
|
||||
# number of calculations to perform, default = 1
|
||||
# -mode eval/run/min
|
||||
# style of calculations: single snapshot evals, dynamics, minimization
|
||||
# default = eval
|
||||
# -size Nx Ny Nz
|
||||
# cubic lattice, default = 2 2 2
|
||||
# -rho 0.75 0.1
|
||||
# reduced density and random variation thereof, default = 0.75 0.1
|
||||
# -delta 0.1
|
||||
# randomly perturb atoms initially by this distance, default 0.0
|
||||
# -nsteps 100
|
||||
# number of timesteps in dynamics runs, default = 100
|
||||
# -temp 1.0
|
||||
# initial temperature in dynamics runs, default = 1.0
|
||||
# -tol 0.001
|
||||
# tolerance for minimizations, default = 0.001
|
||||
# -seed 12345
|
||||
# random number seed > 0, default = 12345
|
||||
|
||||
---
|
||||
|
||||
Run with TCP: 1 proc each
|
||||
|
||||
% python3 series_driver.py -mdi "-role DRIVER -name aimd -method TCP -port 8021"
|
||||
|
||||
% lmp_mpi -mdi "-role ENGINE -name LAMMPS -method TCP -port 8021 -hostname localhost" -log log.series -in in.series
|
||||
|
||||
---
|
||||
|
||||
Run with TCP: 1 proc + 4 procs
|
||||
|
||||
% python3 series_driver.py -mdi "-role DRIVER -name aimd -method TCP -port 8021"
|
||||
|
||||
% mpirun -np 4 lmp_mpi -mdi "-role ENGINE -name LAMMPS -method TCP -port 8021 -hostname localhost" -log log.series -in in.series
|
||||
|
||||
32
examples/mdi/in.aimd.alone
Normal file
32
examples/mdi/in.aimd.alone
Normal file
@ -0,0 +1,32 @@
|
||||
# 3d Lennard-Jones melt - MDI driver script
|
||||
|
||||
variable x index 5
|
||||
variable y index 5
|
||||
variable z index 5
|
||||
|
||||
units lj
|
||||
atom_style atomic
|
||||
|
||||
lattice fcc 0.8442
|
||||
region box block 0 $x 0 $y 0 $z
|
||||
create_box 1 box
|
||||
create_atoms 1 box
|
||||
mass 1 1.0
|
||||
|
||||
velocity all create 1.44 87287 loop geom
|
||||
|
||||
pair_style lj/cut 2.5
|
||||
pair_coeff 1 1 1.0 1.0 2.5
|
||||
|
||||
neighbor 0.3 bin
|
||||
neigh_modify delay 0 every 1 check yes
|
||||
|
||||
# NVE
|
||||
fix 1 all nve
|
||||
# NPT
|
||||
#fix 1 all npt temp 1.0 1.0 0.1 iso 1.0 1.0 1.0
|
||||
|
||||
thermo_style custom step temp pe etotal press vol
|
||||
thermo 1
|
||||
|
||||
run 5
|
||||
32
examples/mdi/in.aimd.driver
Normal file
32
examples/mdi/in.aimd.driver
Normal file
@ -0,0 +1,32 @@
|
||||
# 3d Lennard-Jones melt - MDI driver script
|
||||
|
||||
variable x index 5
|
||||
variable y index 5
|
||||
variable z index 5
|
||||
|
||||
units lj
|
||||
atom_style atomic
|
||||
|
||||
lattice fcc 0.8442
|
||||
region box block 0 $x 0 $y 0 $z
|
||||
create_box 1 box
|
||||
create_atoms 1 box
|
||||
mass 1 1.0
|
||||
|
||||
velocity all create 1.44 87287 loop geom
|
||||
|
||||
neighbor 0.3 bin
|
||||
neigh_modify delay 0 every 1 check yes
|
||||
|
||||
# NVE
|
||||
fix 1 all nve
|
||||
# NPT
|
||||
#fix 1 all npt temp 1.0 1.0 0.1 iso 1.0 1.0 1.0
|
||||
|
||||
fix 2 all mdi/aimd
|
||||
fix_modify 2 energy yes virial yes
|
||||
|
||||
thermo_style custom step temp pe etotal press vol
|
||||
thermo 1
|
||||
|
||||
run 5
|
||||
22
examples/mdi/in.aimd.engine
Normal file
22
examples/mdi/in.aimd.engine
Normal file
@ -0,0 +1,22 @@
|
||||
# 3d Lennard-Jones melt - MDI engine script
|
||||
|
||||
variable x index 5
|
||||
variable y index 5
|
||||
variable z index 5
|
||||
|
||||
units lj
|
||||
atom_style atomic
|
||||
|
||||
lattice fcc 0.8442
|
||||
region box block 0 $x 0 $y 0 $z
|
||||
create_box 1 box
|
||||
create_atoms 1 box
|
||||
mass 1 1.0
|
||||
|
||||
pair_style lj/cut 2.5
|
||||
pair_coeff 1 1 1.0 1.0 2.5
|
||||
|
||||
neighbor 0.3 bin
|
||||
neigh_modify delay 0 every 1 check yes
|
||||
|
||||
mdi engine
|
||||
21
examples/mdi/in.series
Normal file
21
examples/mdi/in.series
Normal file
@ -0,0 +1,21 @@
|
||||
# MDI engine script to process a series of evaulate, run, minimize commands
|
||||
|
||||
units lj
|
||||
atom_style atomic
|
||||
|
||||
lattice fcc 0.8442
|
||||
region box block 0 1 0 1 0 1
|
||||
create_box 1 box
|
||||
mass 1 1.0
|
||||
|
||||
pair_style lj/cut 2.5
|
||||
pair_coeff 1 1 1.0 1.0 2.5
|
||||
|
||||
neighbor 0.3 bin
|
||||
neigh_modify delay 0 every 1 check yes
|
||||
|
||||
fix 1 all nve
|
||||
|
||||
thermo 10
|
||||
|
||||
mdi engine
|
||||
266
examples/mdi/series_driver.py
Normal file
266
examples/mdi/series_driver.py
Normal file
@ -0,0 +1,266 @@
|
||||
# MDI driver to perform a series of independent calculations
|
||||
# using LAMMPS as an engine
|
||||
|
||||
# Syntax: python3 series_driver.py switch arg switch arg ...
|
||||
# possible switches:
|
||||
# -mdi "-role DRIVER ..."
|
||||
# required switch
|
||||
# -n 10
|
||||
# number of calculations to perform, default = 1
|
||||
# -mode eval/run/min
|
||||
# style of calculations: single snapshot evals, dynamics, minimization
|
||||
# default = eval
|
||||
# -size Nx Ny Nz
|
||||
# cubic lattice, default = 2 2 2
|
||||
# -rho 0.75 0.1
|
||||
# reduced density and random variation thereof, default = 0.75 0.1
|
||||
# -delta 0.1
|
||||
# randomly perturb atoms initially by this distance, default 0.0
|
||||
# -nsteps 100
|
||||
# number of timesteps in dynamics runs, default = 100
|
||||
# -temp 1.0
|
||||
# initial temperature in dynamics runs, default = 1.0
|
||||
# -tol 0.001
|
||||
# tolerance for minimizations, default = 0.001
|
||||
# -seed 12345
|
||||
# random number seed > 0, default = 12345
|
||||
|
||||
import sys,math,random
|
||||
import mdi
|
||||
import numpy as np
|
||||
from mpi4py import MPI
|
||||
|
||||
# error message
|
||||
|
||||
def error(txt=None):
|
||||
if txt: raise Exception(txt)
|
||||
raise Exception("Syntax: python3 series_driver.py switch arg switch arg ...")
|
||||
|
||||
# send a LAMMPS input script command to MDI engine
|
||||
|
||||
def send_command(cmd):
|
||||
mdi.MDI_Send_Command("NBYTES",mdicomm)
|
||||
mdi.MDI_Send(len(cmd),1,mdi.MDI_INT,mdicomm)
|
||||
mdi.MDI_Send_Command("COMMAND",mdicomm)
|
||||
mdi.MDI_Send(cmd,len(cmd)+1,mdi.MDI_CHAR,mdicomm)
|
||||
|
||||
# parse command-line args
|
||||
|
||||
args = sys.argv[1:]
|
||||
narg = len(args)
|
||||
|
||||
mdiarg = 0
|
||||
ncalc = 1
|
||||
mode = "eval"
|
||||
nx = ny = nz = 2
|
||||
rho = 0.75
|
||||
rhodelta = 0.1
|
||||
delta = 0.0
|
||||
nsteps = 100
|
||||
tinitial = 1.0
|
||||
tol = 0.001
|
||||
seed = 12345
|
||||
|
||||
iarg = 0
|
||||
while iarg < narg:
|
||||
if args[iarg] == "-mdi":
|
||||
if iarg+2 > narg: error()
|
||||
mdiarg = iarg + 1
|
||||
iarg += 2
|
||||
elif args[iarg] == "-n":
|
||||
if iarg+2 > narg: error()
|
||||
ncalc = int(args[iarg+1])
|
||||
iarg += 2
|
||||
elif args[iarg] == "-mode":
|
||||
if iarg+2 > narg: error()
|
||||
mode = args[iarg+1]
|
||||
if mode != "eval" and mode != "run" and mode != "min": error()
|
||||
iarg += 2
|
||||
elif args[iarg] == "-size":
|
||||
if iarg+4 > narg: error()
|
||||
nx = int(args[iarg+1])
|
||||
ny = int(args[iarg+2])
|
||||
nz = int(args[iarg+3])
|
||||
if nx <= 0 or ny <= 0 or nz <= 0: error()
|
||||
iarg += 4
|
||||
elif args[iarg] == "-rho":
|
||||
if iarg+3 > narg: error()
|
||||
rho = float(args[iarg+1])
|
||||
rhodelta = float(args[iarg+2])
|
||||
if rho-rhodelta <= 0.0: error()
|
||||
iarg += 4
|
||||
elif args[iarg] == "-delta":
|
||||
if iarg+2 > narg: error()
|
||||
delta = float(args[iarg+1])
|
||||
if delta < 0.0: error()
|
||||
iarg += 2
|
||||
elif args[iarg] == "-nsteps":
|
||||
if iarg+2 > narg: error()
|
||||
nsteps = int(args[iarg+1])
|
||||
if nsteps < 0: error()
|
||||
iarg += 2
|
||||
elif args[iarg] == "-temp":
|
||||
if iarg+2 > narg: error()
|
||||
tinitial = float(args[iarg+1])
|
||||
if tinitial < 0.0: error()
|
||||
iarg += 2
|
||||
elif args[iarg] == "-tol":
|
||||
if iarg+2 > narg: error()
|
||||
tol = float(args[iarg+1])
|
||||
if tol < 0.0: error()
|
||||
iarg += 2
|
||||
elif args[iarg] == "-seed":
|
||||
if iarg+2 > narg: error()
|
||||
seed = int(args[iarg+1])
|
||||
if seed <= 0: error()
|
||||
iarg += 2
|
||||
else: error()
|
||||
|
||||
if not mdiarg: error()
|
||||
|
||||
# initialize MDI Library
|
||||
|
||||
mdi.MDI_Init(args[mdiarg])
|
||||
|
||||
# MPI communicator for just the driver
|
||||
|
||||
world = mdi.MDI_MPI_get_world_comm()
|
||||
me = world.Get_rank()
|
||||
nprocs = world.Get_size()
|
||||
|
||||
# connect to engine
|
||||
|
||||
mdicomm = mdi.MDI_Accept_Communicator()
|
||||
|
||||
# allocate vectors for per-atom types, coords, vels, forces
|
||||
|
||||
natoms = nx * ny * nz
|
||||
atypes = np.zeros(natoms,dtype=np.int)
|
||||
coords = np.zeros(3*natoms,dtype=np.float64)
|
||||
vels = np.zeros(3*natoms,dtype=np.float64)
|
||||
forces = np.zeros(3*natoms,dtype=np.float64)
|
||||
|
||||
atypes[:] = 1
|
||||
|
||||
# initialize RN generator
|
||||
|
||||
random.seed(seed)
|
||||
|
||||
# loop over sequence of calculations
|
||||
|
||||
for icalc in range(ncalc):
|
||||
|
||||
# delete all atoms so can run a new calculation
|
||||
|
||||
send_command("delete_atoms group all")
|
||||
|
||||
# define simulation box
|
||||
|
||||
onerho = rho + (random.random()-0.5)*rhodelta;
|
||||
sigma = pow(1.0/onerho,1.0/3.0)
|
||||
|
||||
xlo = ylo = zlo = 0.0
|
||||
xhi = nx * sigma
|
||||
yhi = ny * sigma
|
||||
zhi = nz * sigma
|
||||
|
||||
# send simulation box to engine
|
||||
|
||||
vec = [xlo,ylo,zlo,xhi,yhi,zhi,0.0,0.0,0.0]
|
||||
mdi.MDI_Send_command("RESET_BOX",mdicomm)
|
||||
mdi.MDI_Send(vec,9,mdi.MDI_DOUBLE,mdicomm)
|
||||
|
||||
# create atoms on perfect lattice
|
||||
|
||||
m = 0
|
||||
for k in range(nz):
|
||||
for j in range(ny):
|
||||
for i in range(nx):
|
||||
coords[m] = i * sigma
|
||||
coords[m+1] = j * sigma
|
||||
coords[m+2] = k * sigma
|
||||
m += 3
|
||||
|
||||
# perturb lattice
|
||||
|
||||
for m in range(3*natoms):
|
||||
coords[m] += 2.0*random.random()*delta - delta
|
||||
|
||||
# define initial velocities
|
||||
|
||||
for m in range(3*natoms):
|
||||
vels[m] = random.random() - 0.5
|
||||
|
||||
tcurrent = 0.0
|
||||
for m in range(3*natoms):
|
||||
tcurrent += vels[m]*vels[m]
|
||||
tcurrent /= 3*(natoms-1)
|
||||
|
||||
factor = math.sqrt(tinitial/tcurrent)
|
||||
|
||||
for m in range(3*natoms):
|
||||
vels[m] *= factor
|
||||
|
||||
# send atoms and their properties to engine
|
||||
|
||||
mdi.MDI_Send_command("CREATE_ATOM",mdicomm)
|
||||
mdi.MDI_Send(natoms,1,mdi.MDI_INT,mdicomm)
|
||||
mdi.MDI_Send_command("CREATE_TYPE",mdicomm)
|
||||
mdi.MDI_Send(atypes,natoms,mdi.MDI_INT,mdicomm)
|
||||
mdi.MDI_Send_command("CREATE_X",mdicomm)
|
||||
mdi.MDI_Send(coords,3*natoms,mdi.MDI_DOUBLE,mdicomm)
|
||||
mdi.MDI_Send_command("CREATE_V",mdicomm)
|
||||
mdi.MDI_Send(vels,3*natoms,mdi.MDI_DOUBLE,mdicomm)
|
||||
mdi.MDI_Send_command("CREATE_GO",mdicomm)
|
||||
|
||||
# eval or run or minimize
|
||||
|
||||
if mode == "eval":
|
||||
mdi.MDI_Send_command("EVAL",mdicomm)
|
||||
elif mode == "run":
|
||||
send_command("run %d" % nsteps)
|
||||
elif mode == "min":
|
||||
send_command("minimize %g %g 1000 1000" % (tol,tol))
|
||||
|
||||
# request energy
|
||||
|
||||
mdi.MDI_Send_command("<ENERGY",mdicomm)
|
||||
energy = mdi.MDI_Recv(1,mdi.MDI_DOUBLE,mdicomm)
|
||||
energy = world.bcast(energy,root=0)
|
||||
|
||||
# request pressure tensor
|
||||
|
||||
mdi.MDI_Send_command("<PTENSOR",mdicomm)
|
||||
ptensor = mdi.MDI_Recv(6,mdi.MDI_DOUBLE,mdicomm)
|
||||
ptensor = world.bcast(ptensor,root=0)
|
||||
|
||||
# request forces
|
||||
|
||||
mdi.MDI_Send_command("<FORCES",mdicomm)
|
||||
mdi.MDI_Recv(3*natoms,mdi.MDI_DOUBLE,mdicomm,buf=forces)
|
||||
world.Bcast(forces,root=0)
|
||||
|
||||
# final output from each calculation
|
||||
|
||||
aveeng = energy/natoms
|
||||
pressure = (ptensor[0] + ptensor[1] + ptensor[2]) / 3.0
|
||||
|
||||
m = 0
|
||||
fx = fy = fz = 0.0
|
||||
for i in range(natoms):
|
||||
fx += forces[m]
|
||||
fy += forces[m+1]
|
||||
fz += forces[m+2]
|
||||
m += 3
|
||||
|
||||
fx /= natoms
|
||||
fy /= natoms
|
||||
fz /= natoms
|
||||
|
||||
line = "Calc %d: eng %7.5g press %7.5g aveForce %7.5g %7.5g %7.5g" % \
|
||||
(icalc+1,aveeng,pressure,fx,fy,fz)
|
||||
if me == 0: print(line)
|
||||
|
||||
# send EXIT command to engine
|
||||
|
||||
mdi.MDI_Send_command("EXIT",mdicomm)
|
||||
@ -42,10 +42,10 @@ FixMDIAimd::FixMDIAimd(LAMMPS *lmp, int narg, char **arg) :
|
||||
// check requirements for LAMMPS to work with MDI as an engine
|
||||
|
||||
if (atom->tag_enable == 0)
|
||||
error->all(FLERR, "Cannot use mdi engined without atom IDs");
|
||||
error->all(FLERR, "Cannot use MDI engine without atom IDs");
|
||||
|
||||
if (atom->natoms && atom->tag_consecutive() == 0)
|
||||
error->all(FLERR, "mdi engine requires consecutive atom IDs");
|
||||
error->all(FLERR, "MDI engine requires consecutive atom IDs");
|
||||
|
||||
// confirm LAMMPS is being run as a driver
|
||||
|
||||
|
||||
@ -635,7 +635,7 @@ void MDIEngine::mdi_commands()
|
||||
|
||||
// ------------------------------------
|
||||
// custom commands and nodes which LAMMPS supports
|
||||
// max length for a command is current 11 chars in MDI
|
||||
// max length for a command is currently 11 chars in MDI
|
||||
// ------------------------------------
|
||||
|
||||
MDI_Register_command("@DEFAULT", "NBYTES");
|
||||
@ -1377,8 +1377,8 @@ void MDIEngine::evaluate()
|
||||
|
||||
/* ----------------------------------------------------------------------
|
||||
RESET_BOX command
|
||||
9 values = boxlo, boxhi, xy, yz, xz
|
||||
wrapper on library reset_box() method
|
||||
9 values = boxlo, boxhi, xy, yz, xz
|
||||
requires no atoms exist
|
||||
allows caller to define a new simulation box
|
||||
---------------------------------------------------------------------- */
|
||||
|
||||
Reference in New Issue
Block a user