I'm trying to use scipy.optimize.linprog and it's not working for me in a simple case. I must be doing something wrong -- help me out to find it please.
Here are the inputs and the expected solution and a test that the expected solution is feasible
import numpy, scipy.optimize
print "A", A_ub
print "b", b_ub
print "c", c
print "x_expected", x_expected
print "expected optimum", numpy.dot(c, x_expected)
print "feasibility test", numpy.dot(A_ub, x_expected) <= b_ub
This prints
A [[-1. -0. 0. 1.]
[-0. -1. 1. 1.]
[-1. -0. -0. -1.]
[-0. -1. -1. -1.]]
b [ 2. 1. -2. -1.]
c [ 1. 1. 0. 0.]
x_expected [ 0 0 -1 2]
expected optimum 0.0
feasibility test [ True True True True]
When I call linprog I get a feasible solution, but it seems to not be as optimal as the one I expected above.
r = scipy.optimize.linprog(c, A_ub, b_ub)
x_derived = r.x
print "x_derived", x_derived
print "derived optimum", numpy.dot(c, x_derived)
print "derived feasibility", numpy.dot(A_ub, x_derived) <= b_ub
This prints
x_derived [ 1. 0. 0. 1.]
derived optimum 1.0
derived feasibility [ True True True True]
Is there something wrong in my reasoning that I should be getting the expected solution (or a better one if there is one) instead of the one I get? If I'm not wrong, how do I make linprog get it for me? Thanks!
