Linprogr giving 2 different solutions when changing the way to impose upper and lower bounds

1 次查看（过去 30 天）

CT 2019-8-14

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/476113-linprogr-giving-2-different-solutions-when-changing-the-way-to-impose-upper-and-lower-bounds

评论： Torsten 2019-8-14

I am running a linear programming problem in Matlab using linprogr. The linear programming problem has many solutions. When I impose "manually" the upper and lower bounds on the unknowns together with the other inequality constraints I get a different result than when using the lb and ub inputs inside the linprogr function. Is it possible? Does it indicate an error in my code?

This is my specific case, with matrices of parameters here https://filebin.net/ppno4h7r4q1p3fbl. I could not attach them because too many. The upper and lower bounds consist of 0<=x_i<=1 for every i-th component of the vector of unknowns x.

1) With "manually" included upper and lower bounds.

clear
rng default
load Aeq_man
load Aineq_man
load beq_man
load bineq_man
f=zeros(size(Aeq_man,2),1);
sol_1=linprog(f,Aineq_man,bineq_man,Aeq_man,beq_man);

2) When using lb and ub inputs inside the linprogr function

rng default
load Aeq
load Aineq
load be
load bineq
load lb
load ub
sol_2=linprog(f,Aineq,bineq,Aeq,beq, lb, ub);

3 个评论
显示 1更早的评论隐藏 1更早的评论

CT 2019-8-14

编辑：CT 2019-8-14

Two different x's. The objective is just a vector of zeros.

Torsten 2019-8-14

That's possible. If you specify the objective to be 0, then one feasible point x is returned - and I guess your problem has more than just one of them.

请先登录，再进行评论。

请先登录，再回答此问题。