Problem-based optimization: Initial guess for only one variable

Question

b999 2023-10-25

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2038351-problem-based-optimization-initial-guess-for-only-one-variable

评论： Walter Roberson 2023-10-26

Hello,

I have a optimization problem with the problem-based approach and I am defining a few variables

A = ones(50,10);
a   = optimvar("a",size(A),"Type","integer","LowerBound",0,"UpperBound",2);     
b   = optimvar("b",size(a),"Type","integer","LowerBound",0,"UpperBound",2);             
c  = optimvar("c",size(a),"Type","integer","LowerBound",0,"UpperBound",2);

Now I want to make an initial guess for only a, because I can't guess values for b and c. I tried it like this, but it doesn't work:

initial_guess = ...;
x0 = struct('a', initial_guess, 'b', [], 'c', []);
...
solution = solve(prob, x0);

It says, that b and c can't be empty. But I don't know how to implemnt it in another way.

Is there a possibilty to make a guess for only one variable?

(If I am defining b and c e.g. as zeros, the guess is infeasable)

2 个评论
显示无隐藏无

Fabio Freschi 2023-10-25

Why don't you set your b and c variables in the middle of the range? In your case 1?

b999 2023-10-25

Because I have a lot of constraints besides the variables, which are not fulfilled then.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2023-10-25

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2038351-problem-based-optimization-initial-guess-for-only-one-variable#answer_1340421

there are only 3^3 different possibilities. Run them each in turn hardcoding one particular a, b, c possibility. Take the best of the 27 results as the solution.

2 个评论
显示无隐藏无

b999 2023-10-25

It is an easy example to show my problem. The Size of the Variables could be 50x10 or more and there are more than three variables. The a,b and c is just an example. I can't try taht many possibilities