Lennard-Jones 38 cluster
In this example we search for a LJ 38 cluster and compare the effect of symmetry and minimum species-wise separation.
The baseline seed is shown below. Where atoms are placed in a sphere of 3.75 A with a minimum separation of 1.5 A.
Al.cell
%BLOCK LATTICE_CART
20 0 0
0 20 0
0 0 20
#FIX
%ENDBLOCK LATTICE_CART
%BLOCK POSITIONS_FRAC
Al 0.0 0.0 0.0 # Al1 % NUM=38
%ENDBLOCK POSITIONS_FRAC
FIX_ALL_CELL : true
#MINSEP=1.5
#CLUSTER
#POSAMP=3.75
#SYMMOPS=2-5
#NFORM=1
Al.pp
1 12 6 2.5
Al
# Epsilon
1
# Sigma
2
# Beta
1
Two more seeds are included, with one not using any symmetry, and the other also have minimum separation disabled
Al_NOSYM.cell
%BLOCK LATTICE_CART
20 0 0
0 20 0
0 0 20
#FIX
%ENDBLOCK LATTICE_CART
%BLOCK POSITIONS_FRAC
Al 0.0 0.0 0.0 # Al1 % NUM=38
%ENDBLOCK POSITIONS_FRAC
FIX_ALL_CELL : true
#MINSEP=1.5
#CLUSTER
#POSAMP=3.75
##SYMMOPS=2-5
#NFORM=1
Al_NOSYM_NOMINSEP.cell
%BLOCK LATTICE_CART
20 0 0
0 20 0
0 0 20
#FIX
%ENDBLOCK LATTICE_CART
%BLOCK POSITIONS_FRAC
Al 0.0 0.0 0.0 # Al1 % NUM=38
%ENDBLOCK POSITIONS_FRAC
FIX_ALL_CELL : true
##MINSEP=1.5
#CLUSTER
#POSAMP=3.75
##SYMMOPS=2-5
#NFORM=1
Deploy three search runs to the server.
disp deploy search --seed Al --project example/1.04 --num 200 --code pp3 --cluster
disp deploy search --seed Al_NOSYM --project example/1.04 --num 200 --code pp3 --cluster
disp deploy search --seed Al_NOSYM_NOMINSEP --project example/1.04 --num 200 --code pp3 --cluster
Now run the search with:
export DISP_DB_FILE=`pwd`/disp_db.yaml
rlaunch multi 4 --local_redirect
There will be many launch* sub directories got created, which can be deleted.
Note
If you don't have a FW_CONFIG_FILE environmental varible set, search result will also be copied to $USER/disp-base/airss-datastore/example/1.04.
$ disp db summary --project example/1.04
Structure WF count - search
RES Init COMPLETED ALL
project seed
example/1.04 Al 200 200 200 200
Al_NOSYM 200 200 200 200
Al_NOSYM_NOMINSEP 200 200 200 200
To retrieve the SHELX files:
disp db retrieve-project --project example/1.04
Standard AIRSS toolkits can be used to rank the structures:
$ ca -u 0.01 -s -cl -r -t
Al-22022*08-4d246b -173.912 38 Al Oh 2 582 ~
Number of structures : 582
Number of compositions : 1
Al-22022*08-4d246b 0.00 210.526 -4.577 38 Al Oh 2
Al-22022*05-910e48 0.00 210.526 0.018 38 Al C5 1
Al-22022*03-db3120 0.00 210.526 0.026 38 Al C5 1
Al_NOSYM*36-414ffd 0.00 210.526 0.047 38 Al C1 1
Al-22022*57-e24a2b 0.00 210.526 0.051 38 Al Cs 1
Al-22022*56-66b471 0.00 210.526 0.066 38 Al Cs 1
Al_NOSYM*14-999f3c 0.00 210.526 0.077 38 Al C1 1
Al-22022*54-625f07 0.00 210.526 0.078 38 Al D4h 1
Al_NOSYM*36-30e049 0.00 210.526 0.085 38 Al C1 1
Al-22022*53-a81769 0.00 210.526 0.091 38 Al C2 1
Using DISP python interface for analysis
SHELX entries stored in DISP can be pulled and analysed. The python code below generates histograms of the energy distribution from these three different seeds.
from disp.database.api import SearchDB
from disp.analysis.airssutils import ResFile, collect_results_in_df
sdb = SearchDB.from_db_file("disp_db.yaml")
df = collect_results_in_df(project_name='example/1.04')
yrange = (df.enthalpy.min(), df.enthalpy.min() + 40)
for key, sub in df.groupby('seed_name'):
sub.enthalpy.hist(range=yrange, bins=30, alpha=0.3, label=key)
plt.legend()
plt.xlabel('Energy')
plt.ylabel('Count')
plt.savefig('disp-lj38.png', dpi=150)
Output plot:
It can be seen that the symmetry and minimum separations constraints have great impact on the distribution of the relaxed structure. The ground state structure with \(O_h\) point group symmetry can be found easily with 200 trials, but it can be a lot most difficult to obtain by a naive random placement of atoms inside a sphere.