OpenOpt stochastic programming demo

from FuncDesigner import *
from openopt import NLP

x, y, z = oovars('x y z', lb=-1, ub=1)
z.lb = 0

# dice distribution
s = discrete(range(1, 7), [1.0/6]*6)
#                  values           probabilities

# future plans include:
# common distributions like gauss, uniform, exponential, etc
# possibility to use quantiles, e.g. constraint1 = P(sin(a)<0.3)>0.2 (probability of event sin(a)<0.3 must less than 0.2)

f1 = -2*s * z**2
f = f1 + 2*x + 3*y

objective = f.M

startPoint = {x:0, y:0, z:0.001}
p = NLP(objective, startPoint, constraints = [f.std < 0.15])
r = p.solve('scipy_cobyla')
print(f.std(r))
'''
Solver:   Time Elapsed = 0.33   CPU Time Elapsed = 0.33
objFunValue: -5.3074085 (feasible, MaxResidual = 0)
0.149999999601
'''