#!python
# CS 42 Exercise
# Uncomment stuff as needed
# # comments are to end-of-line

# For multi-line comments, start and end with """ , same as for docstrings

# Keep lines under 80 characters.
# To continue a line, use \ at the end of the line before

from neuron import LinearNeuron #, QuadraticNeuron
#from cancer_data import cancer_input, cancer_output

# create a 2-input LinearNeuron neuron2
neuron2 = LinearNeuron(2, 0.1)

# Preliminary stuff

print neuron2.operate([0, 0])

# neuron2.learn([1, 1], 1)
# neuron2.display()


# This vector lists a pair of binary inputs for each training set
inputs2 = [[0, 0], [0, 1], [1, 0], [1, 1]]

# For the 16 training sets below, each output corresponds to one 
# of the input pairs in the list inputs2 above

zero_outputs     = [0, 0, 0, 0]
and_outputs      = [0, 0, 0, 1]
nimplies_outputs = [0, 0, 1, 0]
id1_outputs      = [0, 0, 1, 1]

nimplied_outputs = [0, 1, 0, 0]
id2_outputs      = [0, 1, 0, 1]
xor_outputs      = [0, 1, 1, 0]
or_outputs       = [0, 1, 1, 1]

nor_outputs      = [1, 0, 0, 0]
iff_outputs      = [1, 0, 0, 1]
not2_outputs     = [1, 0, 1, 0]
implied_outputs  = [1, 0, 1, 1]

not1_outputs     = [1, 1, 0, 0]
implies_outputs  = [1, 1, 0, 1]
nand_outputs     = [1, 1, 1, 0]
one_outputs      = [1, 1, 1, 1]

# Train the same LinearNeuron neuron2 on each training set 
# and assess its performance

# limit is the limit on number of epochs
limit = 100

"""
neuron2.train_and_assess("zero:    ", inputs2, zero_outputs,     limit, False)
neuron2.train_and_assess("one:     ", inputs2, one_outputs,      limit, False)
neuron2.train_and_assess("id1:     ", inputs2, id1_outputs,      limit, False)
neuron2.train_and_assess("id12     ", inputs2, id2_outputs,      limit, False)
neuron2.train_and_assess("not1:    ", inputs2, not1_outputs,     limit, False)
neuron2.train_and_assess("not2:    ", inputs2, not2_outputs,     limit, False)
neuron2.train_and_assess("and:     ", inputs2, and_outputs,      limit, False)
neuron2.train_and_assess("or:      ", inputs2, or_outputs,       limit, False)
neuron2.train_and_assess("nand:    ", inputs2, nand_outputs,     limit, False)
neuron2.train_and_assess("nor:     ", inputs2, nor_outputs,      limit, False)
neuron2.train_and_assess("implies: ", inputs2, implies_outputs,  limit, False)
neuron2.train_and_assess("nimplies:", inputs2, nimplies_outputs, limit, False)
neuron2.train_and_assess("implied: ", inputs2, implied_outputs,  limit, False)
neuron2.train_and_assess("nimplied:", inputs2, nimplied_outputs, limit, False)
neuron2.train_and_assess("xor:     ", inputs2, xor_outputs,      limit, False)
neuron2.train_and_assess("iff:     ", inputs2, iff_outputs,      limit, False)
"""

"""
# for difficult sets, try a QuadraticNeuron

neuron22 = QuadraticNeuron(2, 0.1)

neuron4 = QuadraticNeuron(2, 0.1)
neuron22.train_and_assess("xor quadratic:", inputs2, xor_outputs, limit, False)
neuron22.train_and_assess("iff quadratic:", inputs2, iff_outputs,  limit, False)


# Try a LinearNeuron on the cancer data

neuron9 = LinearNeuron(9, 0.05)
neuron9.train_and_assess("cancer:", cancer_input, cancer_output, limit, False)


# Try a QuadraticNeuron on the cancer data. Use a larger limit and smaller rate

neuron99 = QuadraticNeuron(9, 0.01)

larger_limit = 500

neuron99.train_and_assess("quadratic cancer:", \
                          cancer_input, cancer_output, larger_limit, False)
"""