0

I have an OR model, with linear constraints and a nonlinear objective function. Is it impossible to solve it with "linprog"? If so, are there any methods to linearize objective functions, fore example by adding some constraints to the model? Thanks for your help.

Morianta
  • 21
  • 3
  • 2
    If you have Matlab at your disposition, and can use any of its functions, you should look at [fmincon](http://www.mathworks.com/help/optim/ug/fmincon.html), which minimizes nonlinear objective functions under linear and nonlinear constraints. – Etienne Pellegrini Aug 15 '14 at 23:47
  • Thank you so much; I think this toolbox solves my problem. – Morianta Aug 16 '14 at 07:42
  • You're welcome! I made it into a proper answer so you can accept it and close this question! Happy coding! – Etienne Pellegrini Aug 16 '14 at 18:46

2 Answers2

1

`linprog' solves linear programming problems, and the objective function should be linear. The nonlinear objective functions can be linearized with mathematical linearization method.

Linearization is a linear approximation of a nonlinear system that is valid in a small region around the operating point.

For example you can use Function approximation for linearization. Taylor approximation is one of the methods. The linearization of a function is the first order term of its Taylor expansion around the point of interest. There exist other methods as well.

Matlab has a document for basic methods and functions of linearization that you may check.

If you support the question with code, people at SO can help more with the problem.

NKN
  • 6,482
  • 6
  • 36
  • 55
0

If you have Matlab at your disposition, and can use any of the optimization toolbox's functions, you should look at fmincon, which minimizes nonlinear objective functions under linear and nonlinear constraints.

Etienne Pellegrini
  • 738
  • 1
  • 5
  • 11