def runtrial(n, asks, bids, y, V, p, iscontrol, gensize=8):
"""
Main execution, running a single experiment.
@param n {int} The number of generations to run.
@param asks {pandas.DataFrame} The historical ask prices.
@param bids {pandas.DataFrame} The historical bid prices.
@param y {dict or pandas.Series} The current portfolio composition.
@param V {number} The portfolio value.
@param p {number} The percentage of V that should be kept as cash (upper
bound), where no more than 2p is kept in cash.
@param iscontrol {boolean} True if the transaction costs constraint should
be ingnored.
@param gensize {int, default=8} The maximum generation size as well as the
size of the starting generation.
"""
returns, rbar = get_returns(bids)
variables = list(asks.axes[1])
definition_inner = {
'lb': 0,
'ub': 10000,
'category': 'discreet'
}
definition = {key: definition_inner for key in variables}
objective = [
lambda x: (expected_returns(x, variables, y, V, rbar, bids.ix[-1]) -
expected_risk(x, variables, y, V, returns, rbar,
bids.ix[-1])
) * -1
]
if not iscontrol:
objective.append(lambda x: expected_transaction(x, variables,
asks.ix[-1],
bids.ix[-1]))
ineq = [
lambda x: constraint(x, variables, y, V, asks.ix[-1], bids.ix[-1], p),
lambda x: -constraint(x, variables, y, V, asks.ix[-1], bids.ix[-1],
2*p),
]
j = 0
for symb in variables:
ineq.append(lambda x: -(x[j] + y.get(symb, 0)))
j += 1
starting_gen = []
for i in range(gensize - 2):
curr_chromosome = []
for j in range(len(variables)):
curr_chromosome.append(random.randint(0, 2000))
starting_gen.append(curr_chromosome)
chr0 = []
chr100 = []
for i in range(len(variables)):
chr0.append(0)
chr100.append(100)
starting_gen.append(chr0)
starting_gen.append(chr100)
problem = ga(name='Test', variables=variables, definition=definition,
objective=objective, starting_gen=starting_gen,
max_gen_size=gensize, total_generations=n, ineq=ineq,
beta=2, mutation_prob=0.15, crossover_prob=0.75)
# print problem
problem.solve()
# print problem.current_generation_str()
# print problem.solution_str()
x = problem.solution['chromosome']
# print x
Er = expected_returns(x, variables, y, V, rbar, bids.ix[-1])
# print 'Expected Returns = %.5f (should be near 0.00007)' % Er
Ersk = expected_risk(x, variables, y, V, returns, rbar, bids.ix[-1])
# print 'Expected Risk = %.5f (should be near 0.00126)' % Ersk
g = constraint(x, variables, y, V, asks.ix[-1], bids.ix[-1], p)
return pd.Series(map_x_to_definition(x, definition)), problem, Er, Ersk, g
'lb': -100,
'ub': 100,
'category': 'discreet'
},
'x2': {
'lb': -100,
'ub': 100,
'category': 'discreet'
}
}
objective = [
lambda x: 10*x[0] - x[1],
lambda x: (1 + x[1]) / x[0],
]
starting_gen = [[1, 1],
[1, 8],
[7, 55],
[1, 0],
[3, 17],
[2, 11]]
problem = ga(name='Example 5-7', variables=variables,
definition=definition, objective=objective,
starting_gen=starting_gen, trim_first=False,
total_generations=10000)
print problem
print problem.current_generation_str(True)
problem.solve()
print problem.current_generation_str()
print problem.solution_str()
]
eq = []
starting_gen = [
[0.2833, 0.1408],
[0.0248, 0.0316],
[0.1384, 0.4092],
[0.3229, 0.1386],
[0.0481, 0.1625],
[0.4921, 0.2845],
]
problem = ga(
name="Two-bar Truss",
variables=variables,
definition=definition,
objective=objective,
ineq=ineq,
starting_gen=starting_gen,
trim_first=False,
total_generations=10,
max_gen_size=6,
)
print problem
print problem.current_generation_str(True)
selection_rnd = [
0.5292,
0.0436,
0.2949,
0.0411,
0.9116,
0.7869,
