Functional Summary

The NLP Procedure

Functional Summary

The following table outlines the options in the NLP statement classified by function.

Description Option

Input Data Set Specifications

data set DATA=

initial values and constraints INEST=

quadratic objective function INQUAD=

program statements MODEL=

skip missing value observations NOMISS

Output Data Set Specifications

variables and derivatives OUT=

result parameter values OUTEST=

program statements OUTMODEL=

combining various OUT... statements OUTALL

CRP Jacobian to OUTEST= data set OUTCRPJAC

derivatives in the OUT= data set OUTDER

grid in the OUTEST= data set OUTGRID

Hessian to OUTEST= data set OUTHESSIAN

iterative output to the OUTEST= data set OUTITER

Jacobian in the OUTEST= data set OUTJAC

NLC Jacobian in the OUTEST= data set OUTNLCJAC

time in the OUTEST= data set OUTTIME

Optimization Specifications

minimization method TECHNIQUE=

update technique UPDATE=

version of optimization technique VERSION=

line-search method LINESEARCH=

line-search precision LSPRECISION=

type of Hessian scaling HESCAL=

start for approximated Hessian INHESSIAN=

iteration number for update restart RESTART=

Initial Value Specifications

best grid point number BEST=

infeasible points in grid search INFEASIBLE=

pseudorandom initial values RANDOM=

constant initial values INITIAL=

Derivatives Specifications

finite-difference derivatives FD[=]

finite-difference derivatives FDHESSIAN[=]

compute finite-difference interval FDINT=

use only diagonal of Hessian DIAHES

test gradient specification GRADCHECK[=]

Constraint Specifications

range for active constraints LCEPSILON=

LM tolerance for deactivating LCDEACT=

tolerance for dependent constraints LCSINGULAR=

Termination Criteria Specifications

maximum number of function calls MAXFUNC=

maximum number of iterations MAXITER=

minimum number of iterations MINITER=

upper limit seconds of CPU time MAXTIME=

absolute function convergence criterion ABSCONV=

absolute function convergence criterion ABSFCONV =

absolute gradient convergence criterion ABSGCONV=

absolute parameter convergence criterion ABSXCONV=

relative function convergence criterion FCONV =

relative function convergence criterion FCONV 2=

relative gradient convergence criterion GCONV=

relative gradient convergence criterion GCONV2=

relative parameter convergence criterion XCONV=

used in FCONV , GCONV criterion FSIZE=

used in XCONV criterion XSIZE=

Covariance Matrix Specifications

kind of covariance matrix COVARIANCE=

$\sigma^2$ factor of COV matrix SIGSQ=

determines factor of COV matrix VARDEF=

absolute singularity for inertia ASINGULAR=

relative M singularity for inertia MSINGULAR=

relative V singularity for inertia VSINGULAR=

threshold for Moore-Penrose inverse G4

tolerance for singular COV matrix COVSING=

profile confidence limits CLPARM=

Printed Output Specifications

print (almost) all printed output PALL

suppresses all printed output NOPRINT

reduces some default output PSHORT

reduces most default output PSUMMARY

initial values PINIT

optimization history PHISTORY

Jacobian matrix PJACOBI

crossproduct Jacobian matrix PCRPJAC

Hessian matrix PHESSIAN

Jacobian of nonlinear constraints PNLCJAC

values of grid points PGRID

values of functions in LSQ, MIN, MAX PFUNCTION

approximate standard errors PSTDERR

covariance matrix PCOV

eigenvalues for covariance matrix PEIGVAL

prints code evaluation problems PERROR

model program, variables LIST

compiled model program LISTCODE

Step Length Specifications

damped steps in line-search DAMPSTEP[=]

maximum trust-region radius MAXSTEP=

initial trust-region radius INSTEP=

Miscellaneous Options

number accurate digits in objective function FDIGITS=

number accurate digits in nonlinear constraints CDIGITS=

general singularity criterion SINGULAR=

do not compute inertia of matrices NOEIGNUM

check optimality in neighborhood OPTCHECK[=]

Chapter Contents
Previous
Next
Top

Description	Option
Input Data Set Specifications
data set	DATA=
initial values and constraints	INEST=
quadratic objective function	INQUAD=
program statements	MODEL=
skip missing value observations	NOMISS

Output Data Set Specifications
variables and derivatives	OUT=
result parameter values	OUTEST=
program statements	OUTMODEL=
combining various OUT... statements	OUTALL
CRP Jacobian to OUTEST= data set	OUTCRPJAC
derivatives in the OUT= data set	OUTDER
grid in the OUTEST= data set	OUTGRID
Hessian to OUTEST= data set	OUTHESSIAN
iterative output to the OUTEST= data set	OUTITER
Jacobian in the OUTEST= data set	OUTJAC
NLC Jacobian in the OUTEST= data set	OUTNLCJAC
time in the OUTEST= data set	OUTTIME

Optimization Specifications
minimization method	TECHNIQUE=
update technique	UPDATE=
version of optimization technique	VERSION=
line-search method	LINESEARCH=
line-search precision	LSPRECISION=
type of Hessian scaling	HESCAL=
start for approximated Hessian	INHESSIAN=
iteration number for update restart	RESTART=

Initial Value Specifications
best grid point number	BEST=
infeasible points in grid search	INFEASIBLE=
pseudorandom initial values	RANDOM=
constant initial values	INITIAL=

Derivatives Specifications
finite-difference derivatives	FD[=]
finite-difference derivatives	FDHESSIAN[=]
compute finite-difference interval	FDINT=
use only diagonal of Hessian	DIAHES
test gradient specification	GRADCHECK[=]

Constraint Specifications
range for active constraints	LCEPSILON=
LM tolerance for deactivating	LCDEACT=
tolerance for dependent constraints	LCSINGULAR=

Termination Criteria Specifications
maximum number of function calls	MAXFUNC=
maximum number of iterations	MAXITER=
minimum number of iterations	MINITER=
upper limit seconds of CPU time	MAXTIME=
absolute function convergence criterion	ABSCONV=
absolute function convergence criterion	ABSFCONV =
absolute gradient convergence criterion	ABSGCONV=
absolute parameter convergence criterion	ABSXCONV=
relative function convergence criterion	FCONV =
relative function convergence criterion	FCONV 2=
relative gradient convergence criterion	GCONV=
relative gradient convergence criterion	GCONV2=
relative parameter convergence criterion	XCONV=
used in FCONV , GCONV criterion	FSIZE=
used in XCONV criterion	XSIZE=

Covariance Matrix Specifications
kind of covariance matrix	COVARIANCE=
$\sigma^2$ factor of COV matrix	SIGSQ=
determines factor of COV matrix	VARDEF=
absolute singularity for inertia	ASINGULAR=
relative M singularity for inertia	MSINGULAR=
relative V singularity for inertia	VSINGULAR=
threshold for Moore-Penrose inverse	G4
tolerance for singular COV matrix	COVSING=
profile confidence limits	CLPARM=

Printed Output Specifications
print (almost) all printed output	PALL
suppresses all printed output	NOPRINT
reduces some default output	PSHORT
reduces most default output	PSUMMARY
initial values	PINIT
optimization history	PHISTORY
Jacobian matrix	PJACOBI
crossproduct Jacobian matrix	PCRPJAC
Hessian matrix	PHESSIAN
Jacobian of nonlinear constraints	PNLCJAC
values of grid points	PGRID
values of functions in LSQ, MIN, MAX	PFUNCTION
approximate standard errors	PSTDERR
covariance matrix	PCOV
eigenvalues for covariance matrix	PEIGVAL
prints code evaluation problems	PERROR
model program, variables	LIST
compiled model program	LISTCODE

Step Length Specifications
damped steps in line-search	DAMPSTEP[=]
maximum trust-region radius	MAXSTEP=
initial trust-region radius	INSTEP=

Miscellaneous Options
number accurate digits in objective function	FDIGITS=
number accurate digits in nonlinear constraints	CDIGITS=
general singularity criterion	SINGULAR=
do not compute inertia of matrices	NOEIGNUM
check optimality in neighborhood	OPTCHECK[=]