WORMS Brian's Digest: Software - Quadratic Programming


volume 1997



SCI.OP-RESEARCH Digest Mon, 28 June 98 Volume 5: Issue 27

Date: 29 Jun 1998 10:30:25 GMT
From: spellucci@mathematik.tu-darmstadt.de (Peter Spellucci)
Subject: Quadratic Programming
check http://plato.la.asu.edu/guide.html

dualqp is available also in c (translated by f2c). also cfsqp (not quite public domain, but free for university use) by andre tits contains of course a qpsolver, this time generically written in c. you did not specify the dimensions of your problem. for large scale problems, there seems indeed to be no public domain code in c.

hope this helps

peter

Date: 1 Jul 1998 09:10:17 -0500
From: hennebry@plains.NoDak.edu (Michael J. Hennebry)
Subject: Quadratic Programming
Usually you can link FORTRAN with C and C++. If you really need C or C++, you can use f2c. The resulting code will be somewhat less penetrable than the original.

Mike hennebry@plains.NoDak.edu


SCI.OP-RESEARCH Digest Mon, 28 June 98 Volume 5: Issue 26

Date: 23 Jun 1998 10:02:37 GMT
From: tilo.schwarz@dbag.ulm.DaimlerBenz.COM (Tilo Schwarz)
Subject: quadratic programming
Hello anybody,

I'm looking for c/c++ - code to do quadratic programming (linear equality and inequality constraints), where the quadratic function to minimize is _not_ convex.

Does a sci.op-research - faq exist?

Thanks a lot.

Tilo

Date: 23 Jun 1998 11:26:38 GMT
From: spellucci@mathematik.tu-darmstadt.de (Peter Spellucci)
Subject: quadratic programming
regarding the question for a faq: there is a lp and a nlp - faq at http://www.mcs.anl.gov/home/otc/Guide/faq/ you may also consult michael tricks homepage http://mat.gsia.cmu.edu/ for nonconvex qp's there are several codes available, but not in the public domain; the stanford group has its qpsol (solves a nonconvex qp for a _local_ solution). fletcher and leyffer have such a code (very low priced for university users):

contact sleyffer@mcs.dundee.ac.uk for the _global_ solution of nonconvex qp_problems there exist presently research codes only which in addition are limited to either low dimension or special representation of the matrix (block diagonalized with respect to the negative eigenvalues). the books of horst and tuy on global optimization are a good source for this kind of problem.

hope this helps

peter

Date: Fri, 26 Jun 1998 10:02:06 -0400
From: Yulin Yao <yao@gs.com>
Subject: Quadratic Programming
Hi,

I am looking for C/C++ code to solve a quadratic programming problem with linear equality and inequality constraints. I have checked the FAQ and only found the fortran code (not c/c++) from public domain.

Thanks.

--Yulin


SCI.OP-RESEARCH Digest Mon, 1 Jun 98 Volume 5: Issue 22

Date: Wed, 27 May 1998 10:32:49 -0400
From: Irv Lustig <irv@dizzy.cplex.com>
Subject: Help : C/C++ routine for quadratic programming
To: Dr Montaz Ali

Dr Montaz Ali wrote:
>
> I am looking for C/C++ routine for quadratic programming with constraints.
> Any pointer/help will be highly appreciated.
The CPLEX callable library supports solving convex quadratic programs with linear constraints. Visit http://www.cplex.com/ for more information.

        -Irv Lustig                          irv@dizzy.cplex.com
        Product Manager                      http://www.cplex.com/~irv/
        Math Programming Software            http://www.ilog.com/
        ILOG CPLEX Division                  http://www.cplex.com/

SCI.OP-RESEARCH Digest Mon, 1 Jun 98 Volume 5: Issue 22

Date: 28 May 1998 23:37:23 GMT
From: mackey@Hawaii.Edu (John Mackey)
Subject: binary quadratic problem
Hi Group,

I have a set of variables x1, x2, ..., xn which can each assume a value of 0 or 1. My objective function is a some over some pairs i and j of

(1-xi)(1-xj) + xi xj.

Is there a good program available to minimize this? I don't think LOQO does this type of binary programming.

Thanks, John

Date: Thu, 28 May 1998 17:08:44 -0700
From: "Hans D. Mittelmann" <mittelmann@asu.edu>
Subject: binary quadratic problem
Hi,

OBPDP can do that. Link is on http://plato.la.asu.edu/guide.html#LP

Hans D. Mittelmann                      http://plato.la.asu.edu/
Arizona State University                Phone: (602) 965-6595
Department of Mathematics               Fax:   (602) 965-0461
Tempe, AZ 85287-1804                    email: mittelmann@asu.edu
Date: 29 May 1998 06:49:08 GMT
From: znan@math.waikato.ac.nz (NAN)
Subject: binary quadratic problem
Hi, there.

Assume the problem proposed is an unconstrainted quadratic zero-one programming problem. A paper, contained two algorithm pseudo-codes, could be useful:

P.M. Pardalos and G.P. Rodgers: A branch and bound algorithm for the maximum clique problem, Computers Ops. Res. vol.19 (1992) no.5, 363-375.

Cheers, -Nan Zhu

=================================================================
Nan Zhu, Dr
Department of Mathematics    Fax:   +64 (7) 838 4666
The University of Waikato    Tel:   +64 (7) 838 4466 ex. 8326
Private Bag 3105, Hamilton   Email: znan@waikato.ac.nz
New Zealand                  http://www.math.waikato.ac.nz/~znan/
=================================================================
Date: Sun, 31 May 1998 18:56:00 GMT
From: hwolkowi@orion.math.uwaterloo.ca (Henry Wolkowicz)
Subject: binary quadratic problem
This is an example of 'the max-cut' problem, i.e. a quadratic objective over binary variables. The max cut problem is usually phrased in terms of a graph and modelled as a homogeneous quadratic over +-1 variables. However, these problems are equivalent after a simple transformation.

Semidefinite relaxations provide very good bounds for this problem - then 'triangle inequalities' help in a cutting plane approach. A couple of references:


         author      = "C. HELMBERG and F. RENDL and R. J. VANDERBEI and H. WOLKOWICZ",
         title       = "An interior point method for
                       semidefinite programming",
         journal     =  siopt,
         note     = "
~\\URL: ftp://orion.uwaterloo.ca/pub/henry/reports/sdp.ps.gz",
         year     = "1996",
         vol   = "6",
         pages = "342-361"

or:
         author    = "C. HELMBERG",
         title     = "An interior point method for semidefinite
programming and max-cut bounds",
         school    = "Graz University of Technology",
         address   = "Austria",
         year      = "1994"}
or:
         author    = "C. HELMBERG and F. RENDL",
         title     = "A spectral bundle method for semidefinite
programming",
         school    = "Graz University of Technology",
         address   = "Austria",
         year      = "1997"}
You can try Helmberg's home page for more references and details. Also, there is some public domain code now available by Karisch et al and accessible from that home page (or my home page).

||Prof. Henry Wolkowicz          |Fax:   (519) 725-5441
||University of Waterloo         |Tel:   (519) 888-4567, 1+ext. 5589
||Dept of Comb and Opt           |email:  henry@orion.math.uwaterloo.ca
||Waterloo, Ont. CANADA N2L 3G1  |URL: http://orion.math.uwaterloo.ca/~hwolkowi

SCI.OP-RESEARCH Digest Mon, 25 May 98 Volume 5: Issue 21

Date: Mon, 25 May 1998 08:33:25 -0700
From: Hans D Mittelmann <mittelmann@asu.edu>
Subject: Help : C/C++ routine for quadratic programming

To: Dr Montaz Ali 

Dr Montaz Ali wrote:
>
> I am looking for C/C++ routine for quadratic programming with constraints.
> Any pointer/help will be highly appreciated.
>
Hi,

check out

Hans Mittelmann

Date: Mon, 25 May 1998 17:18:10 +0200
From: Dr Montaz Ali <mali@pc55.cam.wits.ac.za>
Subject: Help : C/C++ routine for quadratic programming
I am looking for C/C++ routine for quadratic programming with constraints. Any pointer/help will be highly appreciated.

With regards

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 Montaz Ali, Ph.D                      | Email : mali@pc55.cam.wits.ac.za
 Dept. of Computational & Applied      | Tel :(011)-716-3969 (W)
 Mathematics,                          | Tel :(011)-403-0250 (H)
 The University of Witwatersrand       | Fax :(011)-403-9317 (W)
 Private Bag - 3, Wits - 2050          |
 Johannesburg, Republic of South Africa|
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

SCI.OP-RESEARCH Digest Mon, 23 Feb 98 Volume 5: Issue 8

Date: Fri, 20 Feb 1998 08:01:21 -0600
From: kim.allemand@epfl.ch
Subject: Searching Beale Algorithm in C/C++
Hi,

Has someone programmed in C/C++ the Beale (E.L.M.) algorithm to solve the following program:

  min Q (quadratic convex)
  st: Ax = b
      x >= 0

SCI.OP-RESEARCH Digest Mon, 26 Jan 98 Volume 5: Issue 4

Date: Mon, 19 Jan 1998 14:46:22 -0700
From: "Hans D. Mittelmann" <mittelmann@asu.edu>
Subject: optimization
To: Juan Rufilanchas <jrufigo@redestb.es>

Juan Rufilanchas wrote:
>
> Dear all,
>
> I have to optimize the following function with some restrictions:
>
> Max:     r*x
> Subject to     a) x' *c *x < k0
>                        b) sum(xi)=1
>                        c) xi>=0
>
> where r=matrix(1;100) and is known
>             x=matrix(100;1) is unknown
>             c=matrix(100,100) is known
>              k0=scalar is known
>             xi is each element of the matrix 'x'.
>
> Is there any good algorithm to do this ?
>
> Could somebody point me to an appropriate quadratic optimization routine ?
> (I am actually working with matlab but I'am not very sure of its accuracy)
>
> Thanks in advance.
Hi,

all the programs listed in http://plato.la.asu.edu/guide.html#NLP can solve this. I would recommend you use DONLP2, available now in f90, f77, and C.

Be aware that there may be multiple solutions in case C is not symmetric and positive definite.

Hans D. Mittelmann                      http://plato.la.asu.edu/
Arizona State University                Phone: (602) 965-6595
Department of Mathematics               Fax:   (602) 965-0461
Tempe, AZ 85287-1804                    email: mittelmann@asu.edu
Date: Tue, 20 Jan 1998 11:16:51 +0100
From: frangio@di.unipi.it (Antonio Frangioni)
Subject: optimization
>I have to optimize the following function with some restrictions:
>
>Max:     r*x
>Subject to     a) x' *c *x < k0
>                       b) sum(xi)=1
>                       c) xi>=0
In Bundle methods for NDO, we have to solve the similar problem

Min:         r*x + t * ( x' *c *x )
Subject to   b) sum(xi)=1
             c) xi>=0
where t > 0 is a parameter: clearly, you can solve the original problem (in Min format) via a line search on t, since it is just the Lagrangean of your nonlinear constraint. I think this may be a reasonable approach if you have a routine for efficiently solving the above (QP) that supports reoptimization on t. This also gives a nice way of handling the case when the problem has no solutions at all.

I have a C++ module that does this: for details you can check http://www.di.unipi.it/~frangio/thesis.html (Chapter I.3).

I know of only another specialzed code around for this, I think it is written in Fortran: the references are in the source I pointed.

Of course, I'm assuming that the matrix c is positive semi-definite: if not the problem is nonconvex and things are much more difficult.

Hope this helps


                                       Antonio Frangioni
Antonio Frangioni                      Research Associate  |  I have gained ...
Dip. di Informatica - Corso Italia 40 - 56125 Pisa, Italy  |  the colour of the
Ph +039-50-887216                www.di.unipi.it/~frangio  |  corn.   [Exupery]

SCI.OP-RESEARCH Digest Mon, 19 Jan 98 Volume 5: Issue 3

Date: Wed, 14 Jan 1998 10:49:38 +0100
From: "Jean-Marc BUIS" <jmbuis@starnet.fr>
Subject: HELP ! Looking for a Quadratic Programming DLL
Hi,

I'm looking for a basic quadratic programming DLL which could solve the simple minimization of a quadratic function with linear constraints

please reply by e-mail.

Thanks in advance !

JMB

A.I.M
Paris

Date: Wed, 14 Jan 1998 10:20:59 -0500
From: Irv Lustig <irv@dizzy.cplex.com>
Subject: HELP ! Looking for a Quadratic Programming DLL
You can do this with the CPLEX callable library, which we supply in DLL form. Visit http://www.cplex.com/"> for more info.


        -Irv Lustig                          irv@dizzy.cplex.com
        Director of Numerical Optimization   http://www.cplex.com/~irv/
        ILOG CPLEX Division                  http://www.ilog.com/
                                             http://www.cplex.com/
Date: Wed, 14 Jan 1998 18:38:25 -0700
From: Hans D Mittelmann <mittelmann@asu.edu>
Subject: HELP ! Looking for a Quadratic Programming DLL
Hi,

I suggest you take BPMPD (accepts extended MPS input). Link in http://plato.la.asu.edu/guide.html#QP

Hans D. Mittelmann                      http://plato.la.asu.edu/
Arizona State University                Phone: (602) 965-6595
Department of Mathematics               Fax:   (602) 965-0461
Tempe, AZ 85287-1804                    email: mittelmann@asu.edu
Date: Thu, 15 Jan 1998 06:48:28 -0700
From: Hans D Mittelmann <mittelmann@asu.edu>
Subject: HELP ! Looking for a Quadratic Programming DLL
Hi,

I forgot to mention BPMPD is free and it's excellent. Look at the benchmarks at http://plato.la.asu.edu/bench.html

I know you asked for a "basic" QP-DLL, but why not take a state-of-the-art one. You just write your input in MPS.