root/trunk/openopt/scikits/openopt/examples/solverParams.py

"""
Modifying of some solver default parameters is performed via
kwargs for p.solve(). They can be solver or prob attributes.
"""
from numpy import *
from scikits.openopt import *
f = lambda x: (x[0]-1.5)**2 + sin(0.8 * x[1] ** 2 + 15)**4 + cos(0.8 * x[2] ** 2 + 15)**4 + (x[3]-7.5)**4
lb, ub = -ones(4), ones(4)

# example 1
p = GLP(f, lb=lb, ub=ub,  maxIter = 1e3, maxCPUTime = 3,  maxFunEvals=1e5,  fEnough = 80)
# solve() kwargs can include some prob settings (like maxTime) as well
r = p.solve('galileo', crossoverRate = 0.80, maxTime = 3,  population = 15,  mutationRate = 0.15)

# example 2
params_list = [{'h0':0.80, 'alp':2.15, 'show':False}, {'h0':0.15, 'alp':2.80, 'color':'k'}]
for i in [0,1]:
    p = NSP(f, [0]*4, lb=lb, ub=ub, legend='ralg'+str(i+1))
    r = p.solve('ralg', plot=True, **params_list[i])