def sync(self): """ Overlaps the current grid world with the updated one """ self.world = Grid(Environment_s.width, Environment_s.height) for pos in self.food.food: self.world.set_value(pos[0], pos[1], Environment_s.grid_food) for pos in self.population.keys(): self.world.set_value(pos[0], pos[1], Environment_s.grid_crit)
def __init__(self):
"""
Initializes a simulation creating the world (which is a grid),
creating a dictionary of critters, the keys are tuples representing the positions
and the values are the critters, spawning food and then syncing everything with the
world
"""
self.world = Grid(Environment_s.width, Environment_s.height)
self.world_age = 0
self.population = {}
for i in range(Simulation_s.n_crit):
c = Critter(Critter_s.randpos())
self.population[c.pos] = c
self.food = Food(Simulation_s.n_food)
self.sync()
# TO DO: Make it work with (3,4) q table size
# Remove redundant initialization of params
'''
Agent starts at bottom left and needs to make its
way to the top right
'''
import numpy as np
from environment import Grid
from agent import Agent
import random
import matplotlib.pyplot as plt
if __name__ == '__main__':
grid = Grid(5, 5)
agent = Agent()
num_episodes = 5000
num_steps = 99
gamma = 0.7
alpha = 0.610
epsilon = 1
max_epsilon = 1
min_epsilon = 0.01
decay_rate = 0.01
def step(action, current_state):
previous_state_position = grid.state_to_position(current_state)
class Simulation:
def __init__(self):
"""
Initializes a simulation creating the world (which is a grid),
creating a dictionary of critters, the keys are tuples representing the positions
and the values are the critters, spawning food and then syncing everything with the
world
"""
self.world = Grid(Environment_s.width, Environment_s.height)
self.world_age = 0
self.population = {}
for i in range(Simulation_s.n_crit):
c = Critter(Critter_s.randpos())
self.population[c.pos] = c
self.food = Food(Simulation_s.n_food)
self.sync()
def sync(self):
"""
Overlaps the current grid world with the updated one
"""
self.world = Grid(Environment_s.width, Environment_s.height)
for pos in self.food.food:
self.world.set_value(pos[0], pos[1], Environment_s.grid_food)
for pos in self.population.keys():
self.world.set_value(pos[0], pos[1], Environment_s.grid_crit)
def step(self):
"""
Makes the population move by one step, reproduce according to the constraints and
will, increase world's age, spawn new food and update the world
"""
children = []
critters = self.population.values()
old_pos = self.population.keys()
for crit in critters:
self.world.set_value(crit.pos[0], crit.pos[1], Environment_s.grid_void) # clear the space on the world
sub = self.world.sub_grid(crit.pos) # critter's view
partners_pos = sub.get_type(Environment_s.grid_crit) # relative position of the potential partners
partners = [] # potential partners
for pos in partners_pos:
pos = (crit.pos[0]+pos[0] - Critter_s.sight, crit.pos[1]+pos[1] - Critter_s.sight) # absolute position of the partner
partners.append(self.population[pos])
old_pos = crit.pos
children += crit.process(self.world, partners)
if self.food.food.has_key(crit.pos):
crit.eat(self.food.food.pop(crit.pos)) # removes the food from the food list and increments the energy of the critter
self.world.set_value(crit.pos[0], crit.pos[1], Environment_s.grid_void) # removes the food from the world
self.population.pop(old_pos) # removes critter from old position
if not crit.is_dead():
if self.population.has_key(crit.pos):
crit.pos = old_pos # if the new position is occupied by another critter don't move
self.population[crit.pos] = crit # set critter in the new (or old) position
self.world.set_value(crit.pos[0], crit.pos[1], Environment_s.grid_crit) # updates the world
self.add_children(children) # add children in the population list
self.world_age += 1 # increments world's age
self.food.add(1, None, self.world_age) # spawns randomly 1 food
self.sync()
def add_children(self, children):
"""
Adds the given children on the population list of the simulation
Parameter children is a list of critters
"""
for child in children:
if not self.population.has_key(child.pos):
self.population[child.pos] = child
def min_stat(self, field):
"""
Returns the critter with the lowest value of field
or None if the population is empty
Parameter field is a string indicating the attribute to compare
"""
l = [(getattr(c, field), c) for c in self.population.values()] # tuples with (value, critter)
return min((l, [(None, None)])[l == []])[1]
def max_stat(self, field):
"""
Returns the critter with the highest value of field
or None if the population is empty
Parameter field is a string indicating the attribute to compare
"""
l = [(getattr(c, field), c) for c in self.population.values()] # tuples with (value, critter)
return max((l, [(None, None)])[l == []])[1]
def sum_stat(self, field):
"""
Returns the sum of the values in field of the whole population
or 0 if the population is empty
Parameter field is a string indicating the attribute to sum
"""
l = [getattr(c, field) for c in self.population.values()] # list of values
return sum((l, [0])[l == []])
def average_stat(self, field):
"""
Returns the average of the values in field of the whole population
or 0 if the population is empty
Parameter field is a string indicating the attribute to average
"""
return float(self.sum_stat(field))/(len(self.population), 1)[len(self.population) == 0]
#########temporary until new gui#############
def stats(self):
"""
Statistics sample
Returns a string with some informations about the world and about the population
"""
max_energy = self.max_stat("energy")
min_energy = self.min_stat("energy")
oldest = self.max_stat("age")
average_age = self.average_stat("age")
average_energy = self.average_stat("energy")
if max_energy == None or min_energy == None or oldest == None:
return ["World age: " + str(self.world_age),
"Food quantity: " + str(len(self.food.food))]
stats = [str(len(self.population)) + " critters.",
"World age: " + str(self.world_age),
"Food quantity: " + str(len(self.food.food)),
"Max energy critter[" + str(max_energy) + "]",
"Min energy critter[" + str(min_energy) + "]",
"Oldest critter[" + str(oldest) + "]",
"Average age: " + str(average_age),
"Average energy: " + str(average_energy)
]
return stats
