How to code "optimization" problem while using penalty function in Python

Question

Look at this simple optimization problem which had just a bit of complexity (penalty function):

I checked scipy and pulp library documentation in Python. I know that I can use "def" for my objective but it doesnt work even in a simple problem - I'm confused and don't know which solver or method should be used to solve it. Even I try the easiest form of penalty function as shown below, a problem which just should try 4 given numbers instead of the y variable to minimize penalty function as objective (the answer is clearly y=-2 and the obj function will be -4):

from scipy.optimize import minimize
import numpy as np

def f(y):
    if y>=0:
        return y
    else:
        return 2*y

y = np.array([1,-1,2,-2])

res = minimize(f(y),args = [y])

But I am faced with this error:

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Please help me to know how I should code this penalty function, and which other library or method I should use.

RobinFrcd · Accepted Answer · 2020-09-21 12:35:02Z

1

Your function here is a scalar function and minimize takes a vector function as input.

In your case, you'd prefer using minimize_scalar which takes a scalar function as input:

from scipy.optimize import minimize_scalar

def f(y):
    if y >= 0:
        return y
    else:
        return 2 * y 
    
minimize_scalar(f, bounds=(-2, 1), method='bounded')

#     fun: -3.9999930649182844
# message: 'Solution found.'
#    nfev: 28
#  status: 0
# success: True
#       x: -1.9999965324591422

answered Sep 21, 2020 at 12:35

RobinFrcd

5,7047 gold badges32 silver badges53 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

fateme krm Over a year ago

i have just tried to make my problem simpler but my real problem had inputs as vector(7 dimensional vector). i guess error was related to how i define y values and as you correct ,it is about method and mention to being scaler. it seems this way of defining function in objective works ,so it may solve my higher dimensional problem too. thanks

RobinFrcd Over a year ago

Glad to help! If it doesn't work, you can ask a new question with a function which matches your problem better :)

Collectives™ on Stack Overflow

How to code "optimization" problem while using penalty function in Python

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related