| 6.0.0 - Currently available |
| |
6.0.0 - Mar 2009 - Major Release |
| |
| |
KNITRO 6.0 introduces new features for solving optimization models
(both linear and nonlinear) with binary or integer variables.
The KNITRO mixed integer programming (MIP) code offers two
algorithms for mixed-integer nonlinear programming (MINLP). The
first is a nonlinear branch and bound method and the second
implements the hybrid Quesada-Grossman method for convex MINLP.
The KNITRO MINLP code is designed for convex mixed integer
programming and is a heuristic for nonconvex problems. The MIP code
also handles mixed integer linear programs (MILP) of moderate size.
Many new user options beginning with "mip_*" have been added to
offer user control over the MIP methods. In addition, the KNITRO
multi-start generation of new start points is improved and the new
user option "ms_maxbndrange" was added.
See more details in the KNITRO 6.0 User's Manual and in the MINLP
examples provided with the distribution.
|
|
|
| 5.2 - No longer available, but still supported |
| |
5.2.0 - Mar 2008 - Minor Release |
| |
| |
KNITRO 5.2 adds new user options to improve default algorithm
performance, control multi-start behavior, control output, and
emphasize achieving feasibility on problems with difficult
nonlinear constraints. The distribution
includes the Intel Math Kernel Library for BLAS and
LAPACK functions, providing significant speedup on machines
running an Intel processor.
New user options include "bar_penaltycons",
"bar_penaltyrule", "bar_feasible", "ms_num_to_save", "ms_savetol",
"ms_terminate", "outdir" and "outappend". Options are given carefully
selected default values, but can be modified to improve
performance on difficult optimization problems.
MATLAB® users may now interface to the KNITRO 5.2.0
libraries using the mex interface provided in Optimization Toolbox
with MATLAB release R2008a.
See the 5.2 KNITRO User's Manual for more information. |
|
|
| 5.1 - No longer available, but still supported |
| |
5.1.0 - Nov 2006 - Minor Release |
| |
| |
KNITRO 5.1 adds new user options to improve algorithm
performance, control multi-start behavior, and save
iterates on large, difficult problems. The distribution
includes the Intel Math Kernel Library for BLAS and
LAPACK functions, providing significant speedup on machines
running an Intel processor. The release also includes an
object-oriented C++ example test driver.
New user options include "bar_maxrefactor",
"bar_maxbacktrack", "ms_maxbndrange", "ms_maxtime_cpu",
and "ms_maxtime_real". Options are given carefully
selected default values, but can be modified to improve
performance on difficult optimization problems.
The KNITRO interfaces for Mathematica and AMPL
add new features.
See the 5.1 KNITRO User's Manual for more information.
|
|
| |
5.1.1 - 20 Feb 2007 - New features and bug fixes |
| |
| |
Allows linking with any BLAS and LAPACK library; in
particular, the ACML from AMD. |
| |
Places the KNITRO Java API in a package named
com.ziena.knitro and distributes the class as a jar file.
|
| |
Adds CPLEX 10.1 to the list of CPLEX libraries searched
for when using option "lpsolver". |
| |
Fixes an error in the solution reported by the KNITRO AMPL
solver when the objective is linear with a constant term.
The final objective reported in the AMPL status line was
incorrect, disagreeing with the correct objective reported
by KNITRO. |
|
| |
5.1.2 - 27 Jul 2007 - New features and bug fixes |
| |
| |
KNITRO can check out licenses from the Ziena network server
and reuse them. This reduces overhead and makes the license
server more robust when an application solves thousands of
optimization problems. |
| |
Improves multithreaded performance on Windows by using
Critical Sections and linking with the MSVC LIBCMT library
(/MT). |
| |
Adds CPLEX 10.2 to the list of CPLEX libraries searched
for when using option "lpsolver". |
| |
Allows the Ziena license manager to run on Windows machines
installed with the Spanish language option. |
|
|
| 5.0 - No longer available, but still supported |
| |
5.0.0 - Feb 2006 - Major Release |
| |
| |
KNITRO 5.0 introduces a streamlined
programming interface, the capability to solve MPEC
problems (i.e., problems with complementarity constraints),
a multi-start option for trying to find the global
solution, the ability to crossover from an Interior-point
solution to the Active Set algorithm for highly accurate
solutions, and a new barrier update rule.
Other enhancements include built-in derivative checking,
option to maximize or minimize the objective, addition of
Hessian user option 5 for the KNITRO solver under AMPL,
and general improvements in algorithm efficiency and
robustness.
See the 5.0 KNITRO User's Manual for more information.
|
|
|
| 4.0 - No longer available or supported |
| |
4.0.0 - Oct 2004 - Major Release |
| |
| |
KNITRO 4.0 adds the Active Set optimization algorithm,
additional barrier update rules, automatic problem
scaling, and specializations that improve performance
for LPs and QPs.
The software is converted entirely to thread-safe
C/C++ code. |
|
|
