scipy.optimize.minimize返回不满足问题约束的解决方案。 为什么？

Question

scipy.optimize.minimize produces solution that does not satisfy the constraints, but the report says that optimization terminated successfully.

The objective is (x[0] - 3)**2 + (x[1] - 2)**2

The constraint is x[0]+x[1]<=4

The correct solution should be [2.5, 1.5]

The answer from the function is [3,2]

I tried different optimization methods. Tried no methods. Tried variations on the syntax.

Here is the code (super simple, it seems):

import numpy as np
from scipy import optimize

def f(x):
    return (x[0] - 3)**2 + (x[1] - 2)**2

def con(x):
    return sum(x)-4 

x0 = np.array([0, 0])
res=optimize.minimize(f, x0, method="SLSQP",constraints={"fun": con, "type": "ineq"}, options={'disp':True}) 
print(res)
print(sum(res.x)-4)

Answer 1

ineq is a >= 0 inequality. I assumed (based on the examples I have seen) that it is <=0 inequality. Indeed, the code below, after correcting for this misinterpretation, produces the correct answer.

import numpy as np from scipy import optimize

def f(x): return (x[0] - 3)**2 + (x[1] - 2)**2

def con(x): return 4-sum(x)

x0 = np.array([0, 0]) res=optimize.minimize(f, x0, method="SLSQP",constraints={"fun": con, "type": "ineq"}, options={'disp':True}) print(res) print(sum(res.x)-4)