# The best nieche

genome = p1.get_best_genome()

print("Fitness:", genome.get_fitness())

DrawNN(genome)

# input-output function of the genome

nn = NeuralNetwork(genome)

x_axis = []

neuron_outputs = []

targets = []

errors = []

for data in training_data:

nn.add_input(data[0])

outputs = nn.get_output()

errors.append(abs(outputs[0] - data[1][0]) )

x_axis.append(data[0][0])

neuron_outputs.append(outputs[0])

targets.append(data[1][0])

plt.plot(x_axis, neuron_outputs, 'b.', label='Output')

plt.plot(x_axis, targets, 'r.', label='Target')

plt.plot(x_axis, errors, 'g.', label='Error')

plt.legend()

plt.savefig('Fitness/map'+str(genome.get_genome_id())+'.png')

plt.close()

# Of the histogram of everyone

# fitnessList = []

# for gkey in p1.get_genomes():

# fitness = p1.genomes[gkey].get_fitness()

# fitnessList.append(fitness)

# plt.hist(fitnessList, bins=50, color='g', density=True)

# plt.savefig('Fitness/fitness'+ str(i) +'.png', bbox_inches='tight')

# plt.close()

# Pie chart of nieches

# bestNieche = p1.get_best_genome().niecheID

# labels = []

# sizes = []

# explode = []

# for niecheKey in p1.nieches:

# labels.append(p1.nieches[niecheKey].identifier)

# sizes.append(p1.nieches[niecheKey].get_member_numb())

# # Explode the best nieche

# if(bestNieche == int(niecheKey)):

# explode.append(0.1)

# else:

# explode.append(0)

# figPie, axPie = plt.subplots()

# axPie.pie(sizes, explode=explode, labels=labels, autopct='%1.1f%%', startangle=90)

# axPie.axis('equal')

# plt.savefig('Fitness/nieche' + str(i) +'.png')

# plt.close()

# Scatter plot

x = []

fitness = []

index = 0

color = []

area = []

for genomeKey in p1.genomes:

x.append(index)

fitness.append(p1.genomes[genomeKey].get_fitness())

color.append(p1.genomes[genomeKey].niecheID)

area.append(5*len(p1.genomes[genomeKey].connection_genes))

index += 1

plt.scatter(x,fitness,s=area,c=color, alpha=0.5)

plt.savefig('Fitness/scatter' + str(i) +'.png')

plt.close()

# Set seed to always get the same result, for debugging

#random.seed(49843651)

# Create population of neural networks

p1 = Population(members=100, inputs=1, outputs=1)

# Create training data

training_data = []

for j in range(1,25):

training_data.append([ [pow(j,2)], [j] ])

for i in range(1,100):

print("Evolution cycle:", i)

# Test it fitness by asking them to first 100 integer root value

print("\tEvaluate...")

# Clear previous fitness values

p1.clear_fitness()

# Shuffle the data

random.shuffle(training_data)

#Feed the data

for data in training_data:

p1.evaluate_fitness(inputs=data[0], targets=data[1])

print("\tMakeing figures...")

createGraphs(p1, i, training_data)

# Doing the rest

p1.selection()

p1.populate()

p1.mutate()

p1.group_genes()

print(len(p1.genomes))

print(len(p1.nieches))

pop1 = Population(members=100, inputs=1, outputs=1)

# Create training data

training_data = []

for j in range(1,50):

training_data.append([ [j**2], [j] ]) #Input array, Target array

print(training_data)

# Evolution cycle

#for i in range(0,100):

i = 0

while(True):

print("i", i)

# Evaluate the fitness of the genomes

pop1.clear_fitness()

for data in training_data:

pop1.evaluate_fitness(inputs=data[0], targets=data[1])

# Evaluate the fitness of the nieches too

pop1.evaluate_nieche_fitness()

# Make figures..

createGraphs(pop1, i, training_data)

# Select the genomes we wish to keep

pop1.selection()

# Create new members

pop1.populate()

# Mutate the genomes

pop1.mutate()

# Put the newly created genes into species

pop1.group_genes()

i = i+1

genome = Genome(weight_mutation=0.1, input_nodes=2, output_nodes=1, genome_id = 1)

# Create input, output nodes, and connect them

genome.create_inputs()

genome.create_outputs()

#Create test data

testData = []

for x in range(0,100):

testData.append([x,x])

# Create 2 hidden layers, and fully connect them to test if it is still going to be non-linear

h1 = Node_gene(node_type=Node_type.HIDDEN, node_id=10)

h2 = Node_gene(node_type=Node_type.HIDDEN, node_id=11)

h3 = Node_gene(node_type=Node_type.HIDDEN, node_id=12)

h4 = Node_gene(node_type=Node_type.HIDDEN, node_id=13)

i1 = genome.node_genes[0]

i2 = genome.node_genes[1]

o1 = genome.node_genes[2]

b1 = Node_gene(node_type=Node_type.HIDDEN, node_id=14)

b1.set_value(50)

b2 = Node_gene(node_type=Node_type.HIDDEN, node_id=15)

b2.set_value(50)

b3 = Node_gene(node_type=Node_type.HIDDEN, node_id=16)

b3.set_value(50)

b4 = Node_gene(node_type=Node_type.HIDDEN, node_id=17)

b4.set_value(50)

genome.node_genes.append(h1)

genome.node_genes.append(h2)

genome.node_genes.append(h3)

genome.node_genes.append(h4)

genome.node_genes.append(b1)

genome.node_genes.append(b2)

genome.node_genes.append(b3)

genome.node_genes.append(b4)

#Connection i1 to h1

con_i1_h1 = Connection_gene(

in_node = i1,

out_node = h1,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 1 )

#Connection i2 to h1

con_i2_h1 = Connection_gene(

in_node = i2,

out_node = h1,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 2 )

#Connection i1 to h2

con_i1_h2 = Connection_gene(

in_node = i1,

out_node = h2,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 3 )

#Connection i2 to h2

con_i2_h2 = Connection_gene(

in_node = i2,

out_node = h2,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 4 )

#Connection h1 to h3

con_h1_h3 = Connection_gene(

in_node = h1,

out_node = h3,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 5 )

#Connection h2 to h3

con_h2_h3 = Connection_gene(

in_node = h2,

out_node = h3,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 6 )

#Connection h1 to h4

con_h1_h4 = Connection_gene(

in_node = h1,

out_node = h4,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 7 )

#Connection h2 to h4

con_h2_h4 = Connection_gene(

in_node = h2,

out_node = h4,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 8 )

#Connection h3 to o1

con_h3_o1 = Connection_gene(

in_node = h3,

out_node = o1,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 9 )

#Connection h4 to o1

con_h4_o1 = Connection_gene(

in_node = h4,

out_node = o1,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 10 )

#Conneciton b1-h1

con_b1_h1 = Connection_gene(

in_node = b1,

out_node = h1,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 11 )

#Conneciton b2-h2

con_b2_h2 = Connection_gene(

in_node = b2,

out_node = h2,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 12 )

#Conneciton b3-h3

con_b3_h3 = Connection_gene(

in_node = b3,

out_node = h3,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 13 )

#Conneciton b4-h4

con_b4_h4 = Connection_gene(

in_node = b4,

out_node = h4,

weight = uniform(-1, 1),

expressed = True,

innovation_number = 14 )

genome.connection_genes.append(con_i1_h1)

genome.connection_genes.append(con_i2_h1)

genome.connection_genes.append(con_i1_h2)

genome.connection_genes.append(con_i2_h2)

genome.connection_genes.append(con_h1_h3)

genome.connection_genes.append(con_h2_h3)

genome.connection_genes.append(con_h1_h4)

genome.connection_genes.append(con_h2_h4)

genome.connection_genes.append(con_h3_o1)

genome.connection_genes.append(con_h4_o1)

genome.connection_genes.append(con_b1_h1)

genome.connection_genes.append(con_b2_h2)

genome.connection_genes.append(con_b3_h3)

genome.connection_genes.append(con_b4_h4)

# Draw and print

DrawNN(genome)

genome.print_genome()

#Test

o1_output = []

i1_input = []

i2_input = []

h1_output = []

h2_output = []

h3_output = []

h4_output = []

for data in testData:

nn = NeuralNetwork(genome)

nn.add_input(data)

nn.get_output()

i1_input.append( genome.node_genes[0].value )

i2_input.append( genome.node_genes[1].value )

o1_output.append( genome.node_genes[2].value )

h1_output.append( genome.node_genes[3].value )

h2_output.append( genome.node_genes[4].value )

h3_output.append( genome.node_genes[5].value )

h4_output.append( genome.node_genes[6].value )

# Plot

#plt.plot(i1_input, i1_input, 'b-', label="i1_input")

#plt.plot(i1_input, i2_input, 'g-', label="i2_input")

#plt.plot(i1_input, h1_output, 'c.', label="h1_output")

#plt.plot(i1_input, h2_output, 'm.', label="h2_output")

plt.plot(i1_input, h3_output, 'y.', label="h3_output")

plt.plot(i1_input, h4_output, 'k.', label="h4_output")

plt.plot(i1_input, o1_output, 'r-', label="o1_output")

plt.legend()