Artelys distributes worldwide the nonlinear optimization
package KNITRO either as a standalone component or with AMPL
modeling environment. The result of a ten-year research
effort at Northwestern
University, KNITRO is developed and commercialized by
the US company Ziena
Optimization since 2001.
KNITRO is directly interfaced with MATLAB since release KNITRO
5.2, which allows every user of the Matlab Optimization Toolbox
to use KNITRO functions.
Nonlinear programming and KNITRO
Nonlinear optimization is at the core
of many business problems, such as:
•
pricing and revenue management;
•
parameter identification in financial and industrial models;
•
design and operation of electricity and gas transportation
networks.
KNITRO, which uses the most recent optimization techniques,
is an ideal tool for solving these problems. It can efficiently
handle nonconvex nonlinear models with thousands of variables.
KNITRO is the first nonlinear optimization package to offer
two options, "interior points" and "active
set", for the solution of inequality constrained problems
; this reinforces its efficiency on a wide range of models
with different structures.
KNITRO interfaces
KNITRO is directly interfaced with Matlab: all its features
and algorithms for continuous optimization are accessible
seamlessly via the Matlab Optimization Toolbox. The use of
this interface is particularly recommended to Matlab users
for coding prototypes or study software as well as for the
development of operational Matlab solutions.
KNITRO can be embedded in any software developed in C/C++,
Java or Fortran. It can thus be fully integrated into and
distributed with an application. In this mode of use, the
developer writes C/C++, Java functions or Fortran subroutines
that compute the objective function and constraints of the
model, as well as their first and optionally second derivatives.
For feasibility and rapid prototyping studies, KNITRO may
be used in conjunction with the AMPL modeling environment.
It is indeed remarkably easy to write a nonlinear model in
AMPL and solve it with KNITRO. Derivative functions computations
are no longer the modeler's concern, thanks to first and second
derivatives provided by AMPL's automatic differentiation engine.
The development cycle is consequently much shorter. In fact
the nature of the work itself is modified, as the user can
focus on modelling and on analysing the results without low-level
issues of communication with the KNITRO code.
Optimization
algorithms within KNITRO
Contact
For more information regarding KNITRO,
please contact us at:
•
Tel: +33 1 44 77 89 00
•
E-mail: info-knitro@artelys.com
Back to
the top
|
