Artificial bee Colony algorithm for global geometry optimisation
artificial_bee_colony_algorithm.py runs in parallel using 'mpi4py'.
The directory /config contains a config file for each of
the algorithms
mpiexec -n P python3 artificial_bee_colony_algorithm.py
where P is the number of processors you want to run it with.
cluster_config:
# Radius of the sphere the initial atoms configuration
# is uniformly distributed in
cluster_radius: 2.5
# the size of the cluster
cluster_size: 20
# number of clusters in the population
pop_size: 24
employed_bee_config:
# Enable the employed bee
enable: 1
# monte carlo search is x_i +f*(x_k1+x_k2-x_k3-x_k3)
# where k1 ~ k4 is random cluster from the population
monte_carlo_search_f: 0.035
# there is two search method which is mutation search (0)
# and monte carlo search (1)
search_method: 1
# for monte carlo search size is total number of clusters positions
# subtracted and added and for mutation search 3 search size means
# trigonometric search
search_size: 4
onlooker_bee_config:
enable: 1
run_config:
# eg potential energy is compared with current loop and index of loop devided by the auto_stop value
# if no meaningful improvement or change is discoverd the loop will stop. To disable -1.
auto_stop: 1.4
# significant values considered when two potential energy is compared
auto_stop_sf: 2
# algorithm can run in parallel (1) or only on single processor (0)
is_parallel: 1
# maximum cycle ignoring other factors such as timeout, or auto stop
maximum_cycle: 999999999
# minimum cycle ignoring other factors such as timeout, or auto stop
minimum_cycle: 10
# run id, increments on each loop
run_id: 19
# Stopping the loop after set time. To disable -1.
time_out: 30
# view trajectory at the end
view_traj: 0
scout_bee_config:
# the scout bee removes a cluster when a cluster is not updated for x loops on other bees and replaces
# it with a new cluster to look for other local minima
check_energies_every_x_loops: 6
# enable/disable cluster renewal when cluster is not updated for x loops
update_energies: 1
# specifies after how many iterations an unchanged energy from a cluster should be considered a local minima
count: 3
# enable/disable the scout bee
enable: 1
# scout bee scans through the list of current clusters and finds the cluster with the lowest energy. It only
# continues with the clusters that are close to the lowest energy found. How close the energies should be is specified
# by this value. If the lowest energy is -10 and energy_abnormal = 0.6, then it only continues with clusters that have
# energies between -10 - (-10 * 0.6) = -4 and -10.
energy_abnormal: 0.6
# if two clusters are (almost) similar in energies, it only keeps the lowest. Similarity is based on this value. For
# example: there are two clusters with energies -74 and -75 and this value is set to 1, it will only keep -75.
energy_difference: 0.06