Diagnose infeasibility of linear programming
4 views (last 30 days)
Show older comments
Greetings,
I use linprog function with different method to solve a linear programming problem. But I got exitflag=-2, which means that there is no feasible solution. It seems the current problem may have conflicts in real world. I need to check which constraints can be relaxed to well-define the problem.
However, I have more thant 100 constraints in this linear programming problem. I don't know how can I diagnose the infeasibility. Can anyone give me some hints on how to diagnose the infeasibility of constraints?
By the way, I also use cplexlp function which is linear programming function of Cplex by IBM. It seems there are some ways to diagnose the infeasibility in Cplex. But I just started to use cplex and don't have idea where to start to diagnose.
Thank you very much! :)
0 Comments
Answers (1)
Bruno Luong
on 14 Dec 2018
Edited: Bruno Luong
on 14 Dec 2018
"I need to check which constraints can be relaxed to well-define the problem."
It likes asking: when a glass of water spills out which of the water molecule is a culpite. The answer is all of them.
For example if you take as simple case in R^2:
z=exp(2i*pi*[0:2]/3);
A=[real(z); imag(z)]'
b=[-1; -1; -1]
Values
A =
1.0000 0
-0.5000 0.8660
-0.5000 -0.8660
>> b
b =
-1
-1
-1
you'll get no feasible point x such that three constraints
A*x <= b
are satisisfied.
But if you remove any row of A, at let the number of constraints is 2, then it becomes well defined. Which one is a bad one? Answer: all of them.
See Also
Categories
Find more on Linear Least Squares in Help Center and File Exchange
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!