DONLP2

You find information about the nonlinear optimization code donlp2 in its different variants here. In order to get the codes you must directly write to s p e l l u c c i _at_ @mathematik.tu-darmstadt.de (remove the blanks from the name). for academic use you will get the codes free of any charge . !!!For commercial use you have to sign a license agreement with TUD !!!which owns the right of commercial exploitation of this work. !!!There will be a small royalty fee to be paid once for all.
 
  • donlp2doc.pdf contains the general description of donlp2 in the f77 version. this also applies to large extent to the other language versions. each version comes with its own documentation.
  • donlp2f77.tar.gz now contains a slightly enhanced f77 version with some minor corrections. This has been tested on a large variety of systems and compilers and runs also on the new 64 bit systems under gfortran. This one has the problem format
    minimize f(x)
    subject to h(x)=0
    and g(x) >= 0
    where f, g and h can be general nonlinear. bound constraints on x within g can be characterized by a descriptive array of integers and are handled in a special way in the code.
  • donlp2f90.tar contains a f90 translation of the original f77 version done by Alan Miller, formerly at CSIRO.
  • The version donlp2_ansi_c.tar.gz contains an ANSI-C-translation of the version of donlp2 as of 6/03 (the f77 version) (done by Serge Schoeffler) which works under any ansi-c-compiler and can be maintained/modified as easily as the original f77 version. no f2c stuff !
  • The version donlp2_static_csharp.cs with the main program donlp2callingapp.cs is an adaptation of this C-code to C#, with omission of the part dealing with linearly dependent constraint gradients and user block interface, available for academic use only.
  • donlp2_ad.tar.gz is the f77-version, combined with the automatic differentiator "jakef" from netlib, and some in this case superfluous stuff like blockwise evaluation and internal numerical differentiation removed. This one generates exact gradients automatically, so that you need to code the function evaluation only. The user interface is as in the f77-version. you can try it interactively also here: NUMAWWW in the section "restringierte Minimierung, SQP-Methode".
  • donlp2_intv_dyn.tar.gz contains a completely new and enhanced version of the code: ansi_c, fully dynamic memory management, elimination of redundant linear equality constraints, and a new user interface, which is easier to use and allows subsets of nonlinear constraints to be evaluated at once. The problem format is now
    minimize f(x)
    subject to
    xu<=x<=xo
    bu<=Ax<=bo
    cu<=c(x)<=co
    where the nonlinear constraints are assembled in c(x). Equalities are obtained by setting the upper bound equal to the lower bound.
  • donlp2_intv_bb.tar.gz is the same code as donlp2_intv_dyn, with a main program driver showing how some variables may be fixed (say at a finite set of discrete values).
  • donlp2_intv.tar.gz is the same code as donlp2_intv_dyn.tar.gz, but with static memory.
  • donlp2_revcomm_c.tar.gz is a version of donlp2_ansi_c which uses the technique of "reverse communication" instead of user function calls to get user information on function and gradient values. This is especially suited for integration of the code into a larger simulation/optimization environment. This is with static memory and the first interface.
  • donlp2 theory part 1 and donlp2 theory part 2 contain the two papers with donlp2's theory.
  • donlp2 results gives a short table of results of donlp2 with the test environment and also the description of some testcases not contained in the other public testcase distributions.
  • donlp2 results detailed contains the complete testresults of the f77 version with its large testenvironment.
  • donlp2 in COPS1
    donlp2 in COPS1 part2
    donlp2 in COPS1 (Argonne report)

    show the results obtained with donlp2 on some of the large scale test cases of Argonne's COPS project (COPS1). The first two files are an extract of the report and give additional results for some larger dimensions, the pdf document is the original Argonne report on COPS1.
  • donlp2 on cute gives results of donlp2 on a part of the CUTE test set, excluding the very large scale cases since donlp2 uses no sparse linear algebra.
  • testcollection is a collection of testcases (all of Hock/Schittkowksi, Schittkowski, Dembo, Himmelblau and others) in f77 to be used with the f77 version. It is independent of donlp2 and could be used for testing other codes too.
  • donlp2 with ampl is the student binary of AMPL plus donlp2 for INTEL 32 bit with linux. The AMPL people provide an interface for donlp2 with the unrestricted version of AMPL.
  • donlp2_ampl_full contains the ampl-interface to the f77-code (old format). you need your own licensed version of (full) ampl in order to use this.
  • MT-donlp2.zip contains a multitasking safe version of the new c-version, done by the Polish student Marek Szyprowski and put under the BSD license. Hence commercial use of this one is prohibited. We give no commercial license of this.