def test_readMemoryByFrame(): #Req 6
"""
The Pac-Man AI shall read memory from the NES emulator
when the game produces a new frame.
"""
evolver = Evolver.Evolver(api, populationSize=2, mutationRate=0.15, maxGenerations=2, \
showResults=True, keepBestProportion=0.5, keepRandomProportion=0.0)
def frameCounter():
"""
Counts the number of frames and stops after 25,000 frames
"""
global frameCount
frameCount += 1
#If the evolver failed to read memory, the frame counts wont match!
assert frameCount == evolver.frameCount
#Test 25,000 frames of gameplay, making sure the tester's frame counter never falls
#out of sync with the evolver.
if frameCount > 25000:
api.removeFrameListener(frameCounter)
api.removeFrameListener(evolver.playGame)
api.stop()
api.addFrameListener(frameCounter)
api.addFrameListener(evolver.playGame)
api.addAccessPointListener(evolver.onLifeCountChanged, nintaco.PostWrite, 0x0067)
api.run()
def test_geneticAlgorithmParameters(): #Req 2
"""
The Pac-Man AI shall accept parameters for the
genetic algorithm's settings when the user enters them.
"""
evolver = Evolver.Evolver(api=api, populationSize=100, \
mutationRate=0.15, maxGenerations=2, showResults=False)
assert evolver.populationSize == 100 and \
evolver.mutationRate == pytest.approx(0.15) and \
evolver.maxGenerations == 2 and evolver.showResults == False
def test_cullPoorNetworks(): #Req 4
"""
The Pac-Man AI shall cull poor-performing neural networks
when it evaluates its neural network's performance
"""
evolver = Evolver.Evolver(api, populationSize=10)
#We will go through and assign a fake fitness to each network,
#because we have already determined our fitness checking
#function works. Then we will have the evolver make a new network
#and check that the poorest one was in fact removed.
for network in evolver.population:
network.fitness = 1000
evolver.population[0].fitness = 0 #Worst one.
evolver.population[0]._poorPerformerFlag = True #Set a flag to show that this one is bad.
evolver.generateNewPopulation(keepBestProportion=0.1, keepRandomProportion=0.0)
for network in evolver.population:
assert not hasattr(network, "_poorPerformerFlag")
def test_getFitness(): #Req 3
"""
The Pac-Man AI shall evaluate its neural network's
performance against the fitness function when it dies.
"""
evolver = Evolver.Evolver(api)
def testFitness():
"""
Gets the current fitness
"""
global fitness
fitness = evolver.getCurrentFitness()
api.removeActivateListener(testFitness)
api.stop()
api.addActivateListener(testFitness)
api.run()
assert fitness >= 0 #Pac-man cannot have a negative score.
def test_sendControllerInput(): #Req 1
"""
The Pac-Man AI shall begin sending output to the
NES controller when it is enabled.
"""
evolver = Evolver.Evolver(api)
def writeAndRead():
"""
Checks to make sure the enter button is pressed and
then released.
"""
evolver.writeGamepad(Evolver.GAMEPAD_START, True)
assert api.readGamepad(0, Evolver.GAMEPAD_START)
evolver.writeGamepad(Evolver.GAMEPAD_START, False)
assert not api.readGamepad(0, Evolver.GAMEPAD_START)
api.removeActivateListener(writeAndRead)
api.stop()
api.addActivateListener(writeAndRead)
api.run()
def main():
"""The main routine."""
with evolver.Evolver() as E:
n = 1000
init_bo = 0
params = np.empty([3, n])
for i in range(n):
print '-----------------\nTime Around ' + str(i)
bo = init_bo + 0.05 * i
define_values(bo)
E.open_file('dropSinusoidal.fe')
for j in range(1):
#E.refine(1)
vals = E.evolve(1)
E.run_command('car_app')
E.run_command('car')
E.run_command('')
E.run_command('dump')
E.run_command('')
E.close_file()
params[0, i] = bo
params[1, i] = find_values('dropSinusoidal.fe.dmp',
'contact_angle_right')[0]
params[2, i] = find_values('dropSinusoidal.fe.dmp',
'contact_angle_right_app')[0]
print 'To Plot'
plt.plot(params[0, :], params[1, :], label="Contact Angle")
plt.plot(params[0, :], params[2, :], label="Apparent Contact Angle")
plt.ylabel('Contact Angle')
plt.xlabel('Bond Number (Bo)')
plt.legend()
plt.show()
def test_constantOutput(): #Req 5
"""
The Pac-Man AI shall display results to the user constantly.
"""
evolver = Evolver.Evolver(api)
assert evolver.printGenerationResults(60)
drawer = ImageDraw.Draw(painting, 'RGBA')
drawer.rectangle([0, 0, w, h], (255, 255, 255, 255))
for f in dude:
drawer.ellipse([f[0], f[1], f[0] + f[2], f[1] + f[3]],
(int(f[4]), int(f[5]), int(f[6]), opacity))
# calculate error
err = 0
for i in range(img.size[0]):
for j in range(img.size[1]):
r, g, b = img.getpixel((i, j))
rp, gp, bp = painting.getpixel((i, j))
err = err + math.sqrt((r - rp)**2 + (g - gp)**2 + (b - bp)**2)
return err
def makepic(dude):
drawer = ImageDraw.Draw(painting, 'RGBA')
drawer.rectangle([0, 0, w, h], (255, 255, 255))
for f in dude:
drawer.ellipse([f[0], f[1], f[0] + f[2], f[1] + f[3]],
(int(f[4]), int(f[5]), int(f[6]), opacity))
painting.show()
filename = "my_drawing.jpg"
painting.save(filename)
ev = evolver.Evolver(population, parents, mrate, controls, figures)
for i in range(10000):
ev.evolve(fitfunc)
makepic(ev.fittest[-1])
import ising1d
import nqs
import sampler
import evolver
import observables
nsteps = 400
### INITIALIZATION
wf = nqs.NqsLocal(10, 2, 1) # Set up a translation-invariant neural network
wf.load_parameters(
'../Outputs/10_Ising05_2loc_200.npz') # Load this pre-trained ANNQS
## TIME EVOLVE
h = ising1d.Ising1d(10, 1.0)
evo = evolver.Evolver(h)
wf = evo.evolve(wf,
.01,
nsteps + 1,
symmetry="local",
file='../Outputs/10SpinEvolve/evolution_2loc_',
print_freq=25,
out_freq=1,
batch_size=1000)
### INITIALIZATION
wf = nqs.NqsLocal(10, 1, 1) # Set up a translation-invariant neural network
wf.load_parameters(
'../Outputs/10_Ising05_1loc_200.npz') # Load this pre-trained ANNQS
## TIME EVOLVE
import nintaco
import evolver as Evolver
populationSize = input("Enter the population size: ")
mutationRate = input("Enter the mutation rate (0-1): ")
maxGenerations = input("How many generations should be ran before \
urning the AI off (0 for infinite): ")
showResults = str(
raw_input("Show results report after the AI is toggled off (y/n): "))
showResults = showResults == "y" or showResults == "Y"
nintaco.initRemoteAPI("localhost", 9999)
api = nintaco.getAPI()
evolver = Evolver.Evolver(api, populationSize=populationSize, \
mutationRate = mutationRate, maxGenerations=maxGenerations, showResults=showResults)
api.addFrameListener(evolver.playGame)
api.addAccessPointListener(evolver.onLifeCountChanged, nintaco.PostWrite,
0x0067)
api.run()
def main():
"""The main method."""
with evolver.Evolver() as E:
print E.run_command('foo := 3')
print E.run_command('print foo')
