1 | initial version |

This question has been sitting here for a while with no answer. I had to do something like this myself recently, so I thought I'd post an answer.

Here is some sample code that you could adapt to your situation. I've used 2 variables for illustration, but most of the code only depends on the value of `n`

set at the top. The code at the bottom displays a 3d plot of your function, the minimum, and a contour line indicating the constraint I choose.

```
n = 2
x = var(",".join(["x%d" % i for i in range(n)]))
a = [-2, 3] # length n
f = sum([a[i]*log(1/x[i]) for i in range(n)])
# constraints: either test functions (should be positive in region) or intervals
c2 = lambda u: -1*(u[0]+2.0*u[1]-1) # x1 <= (-1/2)*x0 + 1/2
c3 = lambda u: 5.0-u[0] # x0 and x1 less than 5
c4 = lambda u: 5.0-u[1]
c5 = lambda u: u[0]-0.1 # x0 and x1 bigger than 0.1
c6 = lambda u: u[1]-0.1
# I1 = (0.1, 5) # uncomment if you just want intervals
# I2 = (0.1, 5)
# initial point
p0 = (0.1,0.1)
# find a minimum and evaluate f there
# p = minimize_constrained(f, [I1, I2], p0) # uncomment if you just want intervals
p = minimize_constrained(f, [c2, c3, c4, c5, c6], p0)
eval_dict = dict([ ("x%d" % i, p[i]) for i in range(n) ])
fmin = f(**eval_dict) # evaluate f at arbitrarily many inputs
print "min occurs at p = ", p, "fmin = ", fmin
# show the minimum and constraints
P = plot3d(f, (x[0], 0.1, 5), (x[1], 0.1,5))
P += point3d((p[0],p[1],fmin), size=20, color='red')
P += parametric_plot3d([lambda s:s, lambda s: -(1/2)*s+1/2, lambda s: f(x0=s, x1=-(1/2)*s+1/2)], (0.1, 0.9),color='red')
show(P)
```

Copyright Sage, 2010. Some rights reserved under creative commons license. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.