lmbma

lmbma − CUTEr L-BFGS-B test driver

lmbma

The lmbma main program test drives L-BFGS-B on SIF problems from the CUTEr distribution.

L-BFGS-B is a nonlinear programming code for unconstrained or bound constrained problems, using the limited memory BFGS quasi-Newton update.

L-BFGS-B was written by C. Zhu, R.H. Byrd, P. Lu and J. Nocedal. It is available from J. Nocedal, Department of Electrical Engineering and Computer Sciences, Nothwestern University, Evanston IL 60201, USA (email: nocedal@venus.eecs.nwu.edu).

The object module lmbma.o is stored in $MYCUTER/precision/bin, where precision is either "single" or "double", according to your local installation.

To avoid multiply defined subroutines when LBFGS is linked with the CUTEr tools, the duplicate BLAS subroutines daxpy, dcopy, ddot and dscal must be removed from the file routines.f in the LBFGS distribution.

Compile (but do not link) the resulting L-BFGS-B source code and copy the resulting object file lbfgsb.o in the directory $MYCUTER/precision/bin. Launch using lmb(1) or sdlmb(1).

If no LBFGSB.SPC file is present in the current directory, the default version is copied from $CUTER/common/src/pkg/lbfgsb/. The default specifications are

corrections

0.0

FACTR

factor in built-in stopping test

0.00001

GTOL

termination criterion on infinity

norm of the gradient

0

IPRINT

output specifier

1

ISBMIN

minimization method in the subspace

10000

MAXIT

maximum number of iterations

10000

MAXCLL

maximum number of function calls

The reader is referred to the papers quoted below if they wishe to modify these parameters.

CUTER

Parent directory for CUTEr

MYCUTER

Home directory of the installed CUTEr distribution.

I. Bongartz, A.R. Conn, N.I.M. Gould, D. Orban and Ph.L. Toint

CUTEr (and SifDec): A Constrained and Unconstrained Testing Environment, revisited, N.I.M. Gould, D. Orban and Ph.L. Toint, ACM TOMS, 29:4, pp.373-394, 2003.

CUTE: Constrained and Unconstrained Testing Environment, I. Bongartz, A.R. Conn, N.I.M. Gould and Ph.L. Toint, TOMS, 21:1, pp.123-160, 1995.

A limited memory algorithm for bound constrained optimization, R. H. Byrd, P. Lu, J. Nocedal and C. Zhu, SIAM J. Scientific Computing 16 (1995), no. 5.

L-BFGS-B: a limited memory FORTRAN code for solving bound constrained optimization problems, C. Zhu, R.H. Byrd, P. Lu, J. Nocedal, Tech. Report, EECS Department, Northwestern University, 1994.