MINPACK-1

From NEOS

Systems of nonlinear equations and nonlinear least squares problems

The algorithms in MINPACK-1 are based on the trust region concept. A modification of Powell's hybrid algorithm is used for systems of nonlinear equations, and a version of the Levenberg-Marquardt algorithm is used for nonlinear least squares problems.

For each problem area there are algorithms that proceed from the analytic Jacobian matrix or directly from the functions themselves. Since the specification of the Jacobian matrix can be an error-prone task, MINPACK-1 also contains an algorithm to check that the Jacobian matrix is consistent with the functions. Also included in the package are machine-readable documentation and a complete set of testing aids.

Software is written in ANSI Fortran. Single- and double-precision versions of the software are available. Machine dependencies are restricted to a single subroutine that defines three machine-dependent constants.

Need more info?

Software can be obtained from netlib (send index from minpack). Other sources include:

C. Abaci, Inc.                  NAG, Inc.
208 St. Mary's St.              1400 Opus Place
Raleigh, NC  27605              Downers Grove, IL 60515
Phone: (919) 832-4847           Phone: (708) 971-2337

Reference:

J. J. More', B. S. Garbow, and K. E. Hillstrom, User Guide for MINPACK-1, Argonne National Laboratory Report ANL-80-74, Argonne, Ill., 1980.

J. J. More', D. C. Sorensen, K. E. Hillstrom, and B. S. Garbow, The MINPACK Project, in Sources and Development of Mathematical Software, W. J. Cowell, ed., Prentice-Hall, 1984.