sympy Differential Calculus Constrained Non-Linear Optimization

Example

Problem statement:

Find the minimum (over x, y) of the function f(x,y), subject to g(x,y)=0, where f(x,y) = 2 * x**2 + 3 * y**2 and g(x,y) = x**2 + y**2 - 4.

Solution: We will solve this problem by performing the following steps:

1. Specify the Lagrangian function for the problem
2. Determine the Karush-Kuhn-Tucker (KKT) conditions
3. Find the (x,y) tuples that satisfy the KKT conditions
4. Determine which of these (x,y) tuples correspond to the minimum of f(x,y)

First, define the optimization variables as well as objective and constraint functions:

import sympy as sp
x, y = sp.var('x,y',real=True);
f = 2 * x**2 + 3 * y**2
g = x**2 + y**2 - 4

Next, define the Lagrangian function which includes a Lagrange multiplier lam corresponding to the constraint

lam = sp.symbols('lambda', real = True)
L = f - lam* g

Now, we can compute the set of equations corresponding to the KKT conditions.

gradL = [sp.diff(L,c) for c in [x,y]] # gradient of Lagrangian w.r.t. (x,y)
KKT_eqs = gradL + [g]
KKT_eqs

[-2*lambda*x + 4*x, -2*lambda*y + 6*y, x**2 + y**2 - 4]

The potential minimizers of f (given g=0) are obtained by solving the KKT_eqs equations overs x, y, lam:

stationary_points = sp.solve(KKT_eqs, [x, y, lam], dict=True) # solve the KKT equations
stationary_points 

[{x: -2, y: 0, lambda: 2},
 {x: 2, y: 0, lambda: 2},
 {x: 0, y: -2, lambda: 3},
 {x: 0, y: 2, lambda: 3}]

Finally, check the objective function for each of the above points to determine the minimum

[f.subs(p) for p in stat_points]

[8, 8, 12, 12]

It follows that the constrained minimum of f equals 8 and is achieved at (x,y)=(-2,0) and (x,y)=(2,0).