Next: Overview
Up: PDA [1ex
Previous: Replacing calls to E02CBF
Minimisation
The routines for minimisation (or optimisation) in this library are:
- PDA_DNLS1
Minimises the sum of squares of M non-linear functions.
- PDA_DNLS1E
Minimises the sum of squares of M non-linear functions (easy version).
- PDA_DQED
Solve bounded nonlinear least squares and nonlinear equations.
- PDA_LMDIF1 (MINPACK/NETLIB)
Minimise the sum of the squares of m nonlinear functions in n
variables, simple interface to PDA_LMDIF.
- PDA_LMDIF (MINPACK/NETLIB)
Minimise the sum of the squares of m nonlinear functions in n
variables with a modified Levenberg-Marquardt algorithm. Needs
function only, the Jacobian is calculated by a forward-difference
approximation.
- PDA_UNCMND (NMS/TIBER)
Minimises a smooth non-linear function of n variables. Needs
function values only.
- PDA_SA (module SIMANN from OPT/NETLIB)
Continuous simulated annealing global optimisation algorithm.
Simple constraints can be specified.
- PDA_SUMSL (module SUMSL from TOMS)
Minimises a general unconstrained objective function using
analytic gradients and a hessian approximation from secant update.
- PDA_SUBPLX (module SUBPLEX from OPT/NETLIB)
Subplex method to solve unconstrained optimisation problems. The
method is well suited for optimising objective functions that are
noisy or are discontinuous at the solution.
Subsections
Next: Overview
Up: PDA [1ex
Previous: Replacing calls to E02CBF
PDA [1ex
Starlink User Note 194
H. Meyerdierks, D. Berry, P. W. Draper, G. Privett, M. Currie
12th October 2005
E-mail:ussc@star.rl.ac.uk
Copyright © 2009 Science and Technology Facilities Council