Software packages available to the public:
Click
DSDP5.8 for Matlab (R2008b) and Readme file
. (Posted September 3, 2009).
The following is a suite of Matlab codes for sensor network localization:
Readme File, and
Dual ESDP Matlab Code, or
ESDP Matlab Code, or
generateD Matlab Code, or
ESDPD Matlab Code, or
Dual ESDPD Matlab Code,
compare them to full SDP relaxation and other approaches
Full SDP Matlab Code (all need to use SDP solver Sedumi in order to run).
Sensor network data files:
A sample problem of 500 sensors and
A sample problem of 1000 sensors.
Also try our new sensor network localization and tracking system
ESDP Matlab Tracking Code.
This code simultaneously localizes and tracks a mass of randomly moving sensors in 2D based on their local distance information.
(Posted November 17, 2007; and updated May 2, 2009.)
Convex Programming: CVX is for modeling and solving convex
optimization problems. Click CVX
to find more information (updated on October 3, 2005).
Semidefinite Programming Solver: The program, DSDP5, solves
general semidefinite programs, and
is an implementation of the dual scaling algorithm.
The source code, written in C, with User-Guide
(postscript file) and sample problems,
can be downloaded by clicking
DSDP Site
(updated Sep. 2005).
View independent benchmarks of DSDP5 and other solvers on SDPLIB,
the DIMACS problem set , and some
large SDPs for an idea of its robustness, speed,
and efficiency.
Semidefinite Programming Solver: The program, COPL_DSDP (DSDP2.0), solves
most semidefinite programs arisen in combinatorial and global optimization,
including max-cut or min-cut, graph-partition, max-cover, stable-set and
nonconvex-qp. COPL_DSDP is a generalization and enhancement of COPL_SDP.
It reads the input data from a MPS-like format. The program
is a implementation of the dual scaling algorithm for mixed linear and
semidefinite programming problems with rank-one matrix constraints, and
of several randomized (or heuristic) rank reduction methods. It is
proven to be particularly effective in solving large scale problems with sparse
input matrices. The source code, written in C, with User-Guide
(postscript file) and sample problems (last updated April 27, 1999),
can be downloaded by clicking
copldsdp.zip,
or
copldsdp.tar.gz
(use "tar -xzf" to untar it), or the windows matlab executable package
dsdp2.zip (added October 2004).
Linear Programming Solver:
A linear programming solver COPL_LP (PC DOS, HP and Linux versions).
This package
solve linear programs with sparse data. It has options to return an optimal
basic solution and to detect infeasibility or unboundedness.
The input data files are in MPS format.
The executable codes, with User-Guide (postscript file) and
sample problems (last updated May 21, 98), can be downloaded by clicking
Dos version , or
HP version , or
Linux version (last updated March/9/2000).
Convex Quadratic Programming Solver: An experimental convex quadratic programming
solver, COPL_QP, is available. This package tries to solve linearly
constrained convex quadratic programs.
The source code, written in C, with User-Guide (postscript file) and
sample problems (last updated Jan 15, 2002; thanks to Alexandre Belloni for
modifying qpproc.c), can be downloaded by clicking
coplqp.zip
or
coplqp.tar.gz .
Linearly Constrained Convex Programming Solver: A
LCCP solver, COPL_LC, is available. This package can solve any linearly
constrained convex program with known gradient and hessian functions.
The source code, written in Fortran, with User-Guide (postscript file) and
sample problems, can be downloaded by clicking
copllc.zip
or
copllc.tar.gz .
Geometric Programming Solver: A GP
solver, COPL_GP, is available.
The executable code (Linux or HP), with User-Guide (postscript file) and
sample problems (last updated May 10, 2000), can be downloaded by clicking
coplgp.zip (Linux)
or
coplgp-hp.zip .
Semidefinite Programming Solver: The program, COPL_SDP, solves
Max-Cut, Equal-Cut,
Unequal-Cut, S-T Max-Cut, and Box Constrained Quadratic Programming Problems.
This program is an implementation of the dual scaling algorithm for SDP and
randomized (or heuristic) rank reduction, and
proven to be particularly effective in large scale problems with sparse
input matrices. The source code, written in C,
with User-Guide (postscript file) and sample problems (last updated Feb 16,
99), can be downloaded by clicking
coplsdp.zip
or
coplsdp.tar.gz.
More test problem can be found in
Gset .
General Optimization Solvers: The selected Matlab optimization programs,
LP, LCP, Indefinite QP and Nonlinear Programming,
are available. Click here for the Matlab File .
*****************************************************************
COPYRIGHT NOTIFICATION
*****************************************************************
(C) COPYRIGHT 1997, 1998, 2002 UNIVERSITIES OF STANFORD and IOWA
*****************************************************************
These software disclose material protectable under copyright laws of
the United States. Permission is hereby granted to use, reproduce,
prepare derivative works, and redistribute to others at no charge,
provided that the original copyright notice, Government license and
disclaimer are retained and any changes are clearly documented;
however, any entity desiring permission to incorporate this software
or a work based on the software into a product for sale must contact
Yinyu Ye at the Department of Management Science & Engineering, Stanford University.