def run(settings, Iterations):
size = settings['size']
nSensors = settings['nSensors']
nMotors = settings['nMotors']
# LOAD ISING CORRELATIONS
# filename = 'correlations-ising2D-size400.npy'
filename2 = 'correlations-ising-generalized-size83.npy'
settings['Cdist'] = np.load(filename2)
# --- POPULATE THE ENVIRONMENT WITH FOOD ---------------+
foods = []
for i in range(0, settings['food_num']):
foods.append(food(settings))
# Food is only created uniformly distributed at the very beginning.
# For a new iteration the placement of the food is kept.
# --- POPULATE THE ENVIRONMENT WITH ORGANISMS ----------+
startstr = 'Starting simulation: (' + str(settings['TimeSteps']) + \
' timesteps) x (' + str(Iterations) + ' iterations)'
print(startstr)
isings = []
for i in range(0, settings['pop_size']):
isings.append(
ising(settings,
size,
nSensors,
nMotors,
name='gen[0]-org[' + str(i) + ']'))
# ------- Reinitialize isings --------
for i, I in enumerate(isings):
if i < 25:
I.species = 0
I.Beta = 1
if settings['isolated_populations']:
I.isolated_population = 0
else:
I.species = 1
I.Beta = 10
if settings['isolated_populations']:
I.isolated_population = 1
#No critical learning:
# CriticalLearning(isings, foods, settings, Iterations)
sim_name, not_used_isings = EvolutionLearning(isings, foods, settings,
Iterations)
return sim_name
N = 64
Nsensors = 4
Nmotors = 2
size = N + Nsensors + Nmotors
ind = 9
beta = 1
Iterations = 1000
T = 3600
visualize = True
visualize = False
I = ising(size, Nsensors, Nmotors)
I.Beta = beta
filename = 'files/network-size_' + str(size) + '-sensors_' + str(
Nsensors) + '-motors_' + str(Nmotors) + '-T_' + str(
T) + '-Iterations_' + str(Iterations) + '-ind_' + str(ind) + '.npz'
directorio = '_' + str(size) + '_' + str(Nsensors) + '_' + str(
Nmotors) + '_' + str(T) + '_' + str(Iterations) + '_' + str(ind)
data = np.load(filename)
I.h = data['h']
I.J = data['J']
p = np.zeros(T)
s = np.zeros(T)
m = np.zeros(T)
def run(settings, Iterations):
size = settings['size']
nSensors = settings['nSensors']
nMotors = settings['nMotors']
# LOAD ISING CORRELATIONS
# filename = 'correlations-ising2D-size400.npy'
filename2 = 'correlations-ising-generalized-size83.npy'
# This adds 13 MB to settings.pickle and we don't need it
# settings['Cdist'] = np.load(filename2)
# --- POPULATE THE ENVIRONMENT WITH FOOD ---------------+
foods = []
for i in range(0, settings['food_num']):
foods.append(food(settings))
# Food is only created uniformly distributed at the very beginning.
# For a new iteration the placement of the food is kept.
if settings['beta_linspace'] is not None:
settings['init_beta'] = beta_linspace(settings)
# --- POPULATE THE ENVIRONMENT WITH ORGANISMS ----------+
try:
set_isings = settings['set_isings']
except KeyError:
set_isings = None
try:
change_beta_loaded_simulation = settings[
'change_beta_loaded_simulation']
except KeyError:
change_beta_loaded_simulation = None
if set_isings is not None:
isings = settings['set_isings']
elif settings['LoadIsings']:
if not settings['switch_seasons_repeat_pipeline']:
loadfile = 'save/' + settings['loadfile'] + '/isings/gen[' + str(
settings['iter']) + ']-isings.pickle'
else:
loadfile = 'save/' + settings['loadfile'] + '/isings/gen[' + str(
settings['iter']) + ']-isings.pickle'
startstr = 'Loading simulation:' + loadfile + ' (' + str(settings['TimeSteps']) + \
' timesteps) x (' + str(Iterations) + ' iterations)'
if not settings['switch_seasons_repeat_pipeline']:
prev_settings = load_settings(settings['loadfile'])
#pop size of current simulation is taken from loaded simulation
settings['pop_size'] = prev_settings['pop_size']
print(startstr)
try:
file = open(loadfile, 'rb')
isings = pickle.load(file)
file.close()
except FileNotFoundError:
# Looking for compressed ising file in case normal pickle file is not found
isings = decompress_pickle(loadfile)
if settings['speciation']:
for I in isings:
I.species = 0
I.shared_fitness = 0
if change_beta_loaded_simulation is not None:
for I in isings:
I.Beta = change_beta_loaded_simulation
else:
startstr = 'Starting simulation: (' + str(settings['TimeSteps']) + \
' timesteps) x (' + str(Iterations) + ' iterations)'
print(startstr)
isings = []
for i in range(0, settings['pop_size']):
isings.append(
ising(settings,
size,
nSensors,
nMotors,
name='gen[0]-org[' + str(i) + ']'))
if settings['beta_linspace_within_sim']:
beta_linspace_within_sim(isings, settings)
# --- CYCLE THROUGH EACH GENERATION --------------------+
# Choose between CriticalLearning (which has both inverse-ising and GA with toggle)
# or EvolutionLearning which is only GA. The functions are fairly similar, should find a
# better way to call them than this.
# ------------------------------------------------------+
#No critical learning:
# CriticalLearning(isings, foods, settings, Iterations)
sim_name, not_used_isings = EvolutionLearning(isings, foods, settings,
Iterations)
return sim_name
