def __init__(self, container, master, genotype):
"""Constructor for slider frame"""
tk.Frame.__init__(self, container)
# must make subplots and canvas instance variables so that the making sliders
# method can access them
# create CPPN graphs to pack onto the GUI
f = Figure(figsize=(10, 10), dpi=100)
self.subp_1 = f.add_subplot(211)
self.subp_2 = f.add_subplot(212)
# create graphs of genotype and phenotype
graph_genotype_GUI(genotype, self.subp_2)
# get outputs for current genotype
norm_in = getNormalizedInputs(num_x, num_y)
outputs = []
for ins in norm_in:
outputs.append(genotype.getOutput(ins)[0])
outputs_np = np.array(outputs, copy=True)
self.subp_1.imshow(np.reshape(outputs_np, (num_x, num_y)),
cmap='Greys')
# create canvas for GUI
self.canvas = FigureCanvasTkAgg(f, self)
self.canvas.show()
self.canvas.get_tk_widget().pack()
# initialize dictionary of sliders
self.scale_dict = {}
# connect frame to the root
self.controller = master
def slider_update_CPPN(self):
"""Pulls weight values from all sliders and updates
the corresponding weights within the CPPN genotype,
then regraphs the plot of the CPPN genotype and
phenotype
"""
# get needed CPPN inputs
norm_in = getNormalizedInputs(num_x, num_y)
keys = list(self.scale_dict.keys())
# for each connection with a slider, update weight to its current value
for innov_num in keys:
innov_num_int = int(innov_num)
for con in genotype.connections:
if (con.getInnovationNumber() == innov_num_int):
con.setWeight(self.scale_dict[innov_num].get())
# replot the CPPN with new weights
# both subplots must be cleared to replace them with new ones
self.subp_1.clear()
self.subp_2.clear()
outputs = []
for ins in norm_in:
outputs.append(genotype.getOutput(ins)[0])
outputs_np = np.array(outputs, copy=True)
self.subp_1.imshow(np.reshape(outputs_np, (num_x, num_y)),
cmap='Greys')
graph_genotype_GUI(genotype, self.subp_2)
self.canvas.show()
self.canvas.get_tk_widget().pack()
def __init__(self, container, master, genotype):
"""Constructor for the main GUI frame"""
# instantiate the frame
tk.Frame.__init__(self, container)
# connect frame to the controller
self.controller = master
f = Figure(figsize=(10, 10), dpi=100)
subp_1 = f.add_subplot(211)
subp_2 = f.add_subplot(212)
# create graphs of genotype and phenotype
graph_genotype_GUI(genotype, subp_2)
# get outputs for current genotype
norm_in = getNormalizedInputs(num_x, num_y)
outputs = []
for ins in norm_in:
outputs.append(genotype.getOutput(ins)[0])
outputs_np = np.array(outputs, copy=True)
subp_1.imshow(np.reshape(outputs_np, (num_x, num_y)), cmap='Greys')
# create canvas for GUI
canvas = FigureCanvasTkAgg(f, self)
canvas.show()
canvas.get_tk_widget().pack()
# create labels for text entry and text entry
tk.Label(self, text="Innovation Number").pack(pady=(10, 0))
tk.Label(self, text="New Weight").pack()
innov_e = tk.Entry(self)
weight_e = tk.Entry(self)
innov_e.pack(pady=(10, 0))
weight_e.pack()
# create buttons for exit and entry b genotype, E1, subp_1, subp_2, norm_in, canvas
tk.Button(
self,
text="Apply",
command=(lambda: update_CPPN(genotype, innov_e, weight_e, subp_1,
subp_2, norm_in, num_x, num_y, canvas)
)).pack()
tk.Button(
self,
text="Sweep Weight",
command=(lambda: sweep_CPPN(genotype, innov_e, subp_1, subp_2,
norm_in, canvas))).pack()
tk.Button(self, text="Exit", command=self.quit).pack(pady=(20, 10))
parser.add_argument("act", type=int,
help="Activation Mutation Probability.")
parser.add_argument("cross", type=int,
help="Crossover probability.")
'''
args = parser.parse_args()
# set numpy seed number for all random numbers
SEED = args.seed
np.random.seed(SEED)
# sets global parameters for 2D structure being created by CPPN, generates inputs
NORM_IN_FILE = open("norm_in.txt", "wb")
NUM_X = 75
NUM_Y = 75
NORM_IN = getNormalizedInputs(NUM_X, NUM_Y)
pickle.dump(NORM_IN, NORM_IN_FILE)
# must get filename from parser to complete file path
FILE_PATH = './fitting_images/' + args.path
PIXELS = getBinaryPixels(FILE_PATH, NUM_X, NUM_Y)
print("Creating distance matrix...")
DIST_MAT_BLACK = get_dist_mat(np.reshape(PIXELS, (NUM_X, NUM_Y)), 1)
DIST_MAT_WHITE = get_dist_mat(np.reshape(PIXELS, (NUM_X, NUM_Y)), 0)
print("Distance matrix created...")
# determines when to save the current population
NGEN_TO_SAVE = args.ngen - 1 # save every n generations
# probability of using NSGA-II to select individuals
SELECT_PROB = 2.0