MIQP

The NEOS Server offers MIQP_BB for the solution of Mixed Integer Quadratic Programming (MIQP) problems. The package implements a branch-and-bound solver with depth-first tree search and maximal fractional branching.

Some background information on the solver and MIQP problems can be found in Numerical Experience with lower bounds for MIQP branch--and--bound.

MIQP was developed by Roger Fletcher and Sven Leyffer

To learn more about integer optimization, click the Background link to the NEOS Guide.

Using the NEOS Server for MIQP

MIQP solve the following problem:

min 1/2x^TQx + c^Tx
s.t. l<= [x; A^Tx] <= u, some components of x are integers

The user needs to provide a data file in a dense or sparse format which basically includes the following information:

The number of variables and the number of linear constraints
An initial guess of the solution
The objective function c and the constraint matrix A (in sparse or dense formats)
Lower and upper bounds
The Hessian Q (in sparse or dense formats)
Number and indices of integer variables

Users could refer to the following sample file.

Interfaces to MIQP

You can submit MIQP jobs by any of the following interfaces:

Or just try sending an sample job:

Example Submission!!!

[Testneos Server]