class ForagingModel(Model):
"""A environemnt to model foraging environment."""
def __init__(self,
N,
width,
height,
grid=10,
iter=100000,
seed=None,
name='SForaging',
viewer=False,
parent=None,
ratio=1.0):
"""Initialize the attributes."""
if seed is None:
super(ForagingModel, self).__init__(seed=None)
else:
super(ForagingModel, self).__init__(seed)
# Create a unique experiment id
self.runid = datetime.datetime.now().timestamp()
self.runid = str(self.runid).replace('.', '')
# Create the experiment folder
# If parent folder exits create inside it
if parent is not None and Path(parent).is_dir():
self.pname = parent + '/' + self.runid + '-' + str(ratio)
else:
self.pname = '/'.join(os.getcwd().split(
'/')[:-2]) + '/results/' + self.runid + '-' + str(iter) + name
# Define some parameters to count the step
self.stepcnt = 1
self.iter = iter
# UI
self.viewer = viewer
# Create db connection
connect = Connect('swarm', 'swarm', 'swarm', 'localhost')
self.connect = connect.tns_connect()
# Fill out the experiment table
self.experiment = Experiment(self.connect, self.runid, N, seed, name,
iter, width, height, grid)
self.experiment.insert_experiment()
# Get the primary key of the experiment table for future use
self.sn = self.experiment.sn
# Create a folder to store results
os.mkdir(self.pname)
# Number of agents
self.num_agents = N
# Environmental grid size
self.grid = Grid(width, height, grid)
# Schedular to active agents
self.schedule = SimultaneousActivation(self)
# Empty list of hold the agents
self.agents = []
def create_agents(self, random_init=True, phenotypes=None):
"""Initialize agents in the environment."""
# This is abstract class. Each class inherting this
# must define this on its own
pass
def create_environment_object(self, jsondata, obj):
"""Create env from jsondata."""
name = obj.__name__.lower()
temp_list = []
i = 0
for json_object in jsondata[name]:
location = (json_object["x"], json_object["y"])
if "q_value" in json_object:
temp_obj = obj(i,
location,
json_object["radius"],
q_value=json_object["q_value"])
else:
temp_obj = obj(i, location, json_object["radius"])
self.grid.add_object_to_grid(location, temp_obj)
temp_list.append(temp_obj)
i += 1
return temp_list
def build_environment_from_json(self):
"""Build env from jsondata."""
jsondata = JsonData.load_json_file(filename)
# Create a instance of JsonData to store object that
# needs to be sent to UI
self.render = JsonData()
self.render.objects = {}
# First create the agents in the environment
for name in jsondata.keys():
obj = eval(name.capitalize())
self.render.objects[name] = self.create_environment_object(
jsondata, obj)
self.hub = self.render.objects['hub'][0]
self.total_food_units = 0
self.foods = []
try:
self.site = self.render.objects['sites'][0]
for i in range(self.num_agents * 1):
f = Food(i,
location=self.site.location,
radius=self.site.radius)
f.agent_name = None
self.grid.add_object_to_grid(f.location, f)
self.total_food_units += f.weight
f.phenotype = dict()
self.foods.append(f)
except KeyError:
pass
def step(self):
"""Step through the environment."""
# Next step
self.schedule.step()
# Increment the step count
self.stepcnt += 1
def gather_info(self):
"""Gather information from all the agents."""
# diversity = np.ones(len(self.agents))
exploration = np.ones(len(self.agents))
foraging = np.ones(len(self.agents))
fittest = np.ones(len(self.agents))
prospective = np.ones(len(self.agents))
for id in range(len(self.agents)):
# diversity[id] = self.agents[id].diversity_fitness
# exploration[id] = self.agents[id].exploration_fitness()
# foraging[id] = self.agents[id].food_collected
# fittest[id] = self.agents[id].individual[0].fitness
# prospective[id] = self.agents[id].carrying_fitness()
exploration[id] = self.agents[id].exploration_fitness()
foraging[id] = self.agents[id].food_collected
fittest[id] = self.agents[id].individual[0].fitness
prospective[id] = self.agents[id].carrying_fitness()
beta = self.agents[-1].beta
mean = Best(self.pname, self.connect, self.sn, 1, 'MEAN', self.stepcnt,
beta, np.mean(fittest), np.mean(prospective),
np.mean(exploration), np.mean(foraging), "None")
mean.save()
std = Best(self.pname, self.connect, self.sn, 1, 'STD', self.stepcnt,
beta, np.std(fittest), np.std(prospective),
np.std(exploration), np.std(foraging), "None")
std.save()
# Compute best agent for each fitness
self.best_agents(prospective, beta, "PROSPE")
self.best_agents(exploration, beta, "EXPLORE")
self.best_agents(foraging, beta, "FORGE")
self.best_agents(fittest, beta, "OVERALL")
return np.argmax(foraging)
def best_agents(self, data, beta, header):
"""Find the best agents in each category."""
idx = np.argmax(data)
# dfitness = self.agents[idx].diversity_fitness
ofitness = self.agents[idx].individual[0].fitness
ffitness = self.agents[idx].food_collected
efitness = self.agents[idx].exploration_fitness()
pfitness = self.agents[idx].carrying_fitness()
phenotype = self.agents[idx].individual[0].phenotype
best_agent = Best(self.pname, self.connect, self.sn, idx, header,
self.stepcnt, beta, ofitness, pfitness, efitness,
ffitness, phenotype)
best_agent.save()
def find_higest_performer(self):
"""Find the best agent."""
fitness = self.agents[0].individual[0].fitness
fittest = self.agents[0]
for agent in self.agents:
if agent.individual[0].fitness > fitness:
fittest = agent
return fittest
def find_higest_food_collector(self):
"""Find the best agent to collect food."""
fitness = self.agents[0].food_collected
fittest = self.agents[0]
for agent in self.agents:
if agent.food_collected > fitness:
fittest = agent
return fittest
def detect_food_moved(self):
"""Detect food moved."""
grid = self.grid
food_loc = self.site.location
neighbours = grid.get_neighborhood(food_loc, 10)
food_objects = grid.get_objects_from_list_of_grid('Food', neighbours)
# print (food_objects)
return food_objects
def foraging_percent(self):
"""Compute the percent of the total food in the hub."""
grid = self.grid
hub_loc = self.hub.location
neighbours = grid.get_neighborhood(hub_loc, self.hub.radius)
food_objects = grid.get_objects_from_list_of_grid('Food', neighbours)
_, hub_grid = grid.find_grid(hub_loc)
for food in self.foods:
_, food_grid = grid.find_grid(food.location)
if food_grid == hub_grid:
food_objects += [food]
food_objects = set(food_objects)
total_food_weights = sum([food.weight for food in food_objects])
return ((total_food_weights * 1.0) / self.total_food_units) * 100
class EvolModel(Model):
"""A environemnt to model swarms."""
def __init__(self,
N,
width,
height,
grid=10,
iter=100000,
seed=None,
expname='COT',
agent='EvolAgent',
parm='swarm_mcarry.txt'):
"""Initialize the attributes."""
if seed is None:
super(EvolModel, self).__init__(seed=None)
else:
super(EvolModel, self).__init__(seed)
self.runid = datetime.datetime.now().strftime("%s") + str(
self.random.randint(1, 1000, 1)[0])
self.pname = '/'.join(os.getcwd().split('/')[:-2]) + '/results/' \
+ self.runid + expname
self.stepcnt = 1
self.iter = iter
self.top = None
# Create db connection
connect = Connect('swarm', 'swarm', 'swarm', 'localhost')
self.connect = connect.tns_connect()
# Fill out the experiment table
self.experiment = Experiment(self.connect, self.runid, N, seed,
expname, iter, width, height, grid)
self.experiment.insert_experiment()
self.sn = self.experiment.sn
# Create a folder to store results
os.mkdir(self.pname)
self.num_agents = N
self.grid = Grid(width, height, grid)
self.schedule = SimultaneousActivation(self)
self.agents = []
self.parm = parm
# Create agents
for i in range(self.num_agents):
a = eval(agent)(i, self)
self.schedule.add(a)
# Add the agent to a random grid cell
x = self.random.randint(-self.grid.width / 2, self.grid.width / 2)
# x = 0
y = self.random.randint(-self.grid.height / 2,
self.grid.height / 2)
# y = 0
a.location = (x, y)
self.grid.add_object_to_grid((x, y), a)
a.operation_threshold = 2 # self.num_agents // 10
self.agents.append(a)
def create_environment_object(self, jsondata, obj):
"""Create env from jsondata."""
name = obj.__name__.lower()
temp_list = []
i = 0
for json_object in jsondata[name]:
location = (json_object["x"], json_object["y"])
if "q_value" in json_object:
temp_obj = obj(i,
location,
json_object["radius"],
q_value=json_object["q_value"])
else:
temp_obj = obj(i, location, json_object["radius"])
self.grid.add_object_to_grid(location, temp_obj)
temp_list.append(temp_obj)
i += 1
return temp_list
def build_environment_from_json(self):
"""Build env from jsondata."""
jsondata = JsonData.load_json_file(filename)
# Create a instance of JsonData to store object that
# needs to be sent to UI
self.render = JsonData()
self.render.objects = {}
for name in jsondata.keys():
obj = eval(name.capitalize())
self.render.objects[name] = self.create_environment_object(
jsondata, obj)
self.hub = self.render.objects['hub'][0]
try:
self.foods = []
self.site = self.render.objects['sites'][0]
food_radius = self.random.randint(20, 30)
for i in range(self.num_agents):
f = Food(i, location=self.site.location, radius=food_radius)
f.agent_name = None
f.phenotype = {}
self.grid.add_object_to_grid(f.location, f)
self.foods.append(f)
except KeyError:
pass
def step(self):
"""Step through the environment."""
# Gather info from all the agents
self.top = self.gather_info()
# Next step
self.schedule.step()
# Increment the step count
self.stepcnt += 1
def gather_info(self):
"""Gather information from all the agents."""
diversity = np.ones(len(self.agents))
exploration = np.ones(len(self.agents))
foraging = np.ones(len(self.agents))
fittest = np.ones(len(self.agents))
for id in range(len(self.agents)):
diversity[id] = self.agents[id].diversity_fitness
exploration[id] = self.agents[id].exploration_fitness()
foraging[id] = self.agents[id].food_collected
fittest[id] = self.agents[id].individual[0].fitness
beta = self.agents[-1].beta
"""
mean = Best(
self.pname, self.connect, self.sn, 1, 'MEAN', self.stepcnt,
beta, np.mean(fittest), np.mean(diversity), np.mean(exploration),
np.mean(foraging), "None"
)
mean.save()
std = Best(
self.pname, self.connect, self.sn, 1, 'STD', self.stepcnt, beta,
np.std(fittest), np.std(diversity), np.std(exploration),
np.std(foraging), "None"
)
std.save()
"""
# Compute best agent for each fitness
self.best_agents(diversity, beta, "DIVERSE")
self.best_agents(exploration, beta, "EXPLORE")
self.best_agents(foraging, beta, "FORGE")
self.best_agents(fittest, beta, "OVERALL")
return np.argmax(foraging)
def best_agents(self, data, beta, header):
"""Find the best agents in each category."""
idx = np.argmax(data)
dfitness = self.agents[idx].diversity_fitness
ofitness = self.agents[idx].individual[0].fitness
ffitness = self.agents[idx].food_collected
efitness = self.agents[idx].exploration_fitness()
phenotype = self.agents[idx].individual[0].phenotype
best_agent = Best(self.pname, self.connect, self.sn, idx, header,
self.stepcnt, beta, ofitness, dfitness, efitness,
ffitness, phenotype)
best_agent.save()
def find_higest_performer(self):
"""Find the best agent."""
fitness = self.agents[0].individual[0].fitness
fittest = self.agents[0]
for agent in self.agents:
if agent.individual[0].fitness > fitness:
fittest = agent
return fittest
def find_higest_food_collector(self):
"""Find the best agent to collect food."""
fitness = self.agents[0].food_collected
fittest = self.agents[0]
for agent in self.agents:
if agent.food_collected > fitness:
fittest = agent
return fittest
def detect_food_moved(self):
"""Detect food moved."""
grid = self.grid
food_loc = self.site.location
neighbours = grid.get_neighborhood(food_loc, 10)
food_objects = grid.get_objects_from_list_of_grid('Food', neighbours)
# print (food_objects)
return food_objects
def food_in_loc(self, loc):
"""Find amount of food in hub."""
grid = self.grid
neighbours = grid.get_neighborhood(loc, 10)
food_objects = grid.get_objects_from_list_of_grid('Food', neighbours)
return food_objects