"cutoff": 5.0,
"radial_basis": "chebyshev",
"cutoff_hyps": [],
"sigma_e": 0.009,
"sigma_f": 0.005,
"sigma_s": 0.0006,
"hpo_max_iterations": 50,
"freeze_hyps": 0,
}
bi, final_cluster = cluster_GA(
nPool,
eleNames,
eleNums,
eleRadii,
generations,
calc,
filename,
log_file,
CXPB,
singleTypeCluster,
use_dask,
use_vasp,
al_method,
learner_params,
train_config,
optimizer,
use_vasp_inter,
)
CXPB = 0.5
eleRadii = [covalent_radii[atomic_numbers[ele]] for ele in eleNames]
filename = "clus_Cu4" # For saving the best cluster at every generation
log_file = "clus_Cu4.log"
singleTypeCluster = False
calc = EMT()
use_vasp = False
if use_dask == True:
# Set up the dask run using the worker-spec file based on the computing cluster
cluster = KubeCluster.from_yaml("worker-cpu-spec.yml")
client = Client(cluster)
# cluster.adapt(minimum=0, maximum=10)
cluster.scale(10) # Since 10 clusters in the pool
bi, final_cluster = cluster_GA(
nPool,
eleNames,
eleNums,
eleRadii,
generations,
calc,
filename,
log_file,
CXPB,
singleTypeCluster,
use_dask,
use_vasp,
optimizer=BFGS, # Set ase optimizer
)
lcharg=False,
lwave=False,
lreal=False,
ispin=2,
isym=0,
)
if cluster_use_dask == True:
# Set up the dask run using the worker-spec file based on the computing cluster
cluster = KubeCluster.from_yaml("worker-cpu-spec.yml")
client = Client(cluster)
# cluster.adapt(minimum=0, maximum=10)
cluster.scale(10) # Since 10 clusters in the pool
cluster.scale(10)
bi, final_cluster = cluster_GA(
nPool,
eleNames,
eleNums,
eleRadii,
generations,
calc,
filename,
log_file,
CXPB,
singleTypeCluster,
cluster_use_dask,
use_vasp,
)