| |
|
|
NLSSOL
Solves dense linear and nonlinear least squares problems,
with linear and nonlinear
constraints.
Main features
-
NLSSOL is built on top of the nonlinear solver NPSOL.
-
More information available about NPSOL at the
Stanford
Systems Optimization Laboratory (SOL) pages.
-
See the
User's Guide for NPSOL 5.0: A Fortran Package for
Nonlinear Programming (1986, Revised July 30, 1998).
-
NLSSOL has two extra parameters compared to NPSOL.
See the help for
nlssolTL.m.
The parameters are
JTJ Initial Hessian and
Reset Frequency.
If JTJ Initial Hessian is set true, then the Jacobian
matrix is used to set the initial
Hessian matrix in the Quasi-Newton approximation.
Otherwise the initial Hessian is set to the identity matrix.
The Reset Frequency
determines how often the algorithm should reset the Hessian to
the value determined by
the
JTJ Initial Hessian setting.
The
default value of the
Reset Frequency is two, which means that the Hessian
is reset rather often.
|
|
