The NLP Procedure

Functional Summary

The following table outlines the options in PROC NLP classified by function. An alphabetical list of options is provided in the Dictionary of Options.

Table 4.1: Functional Summary

Description Statement Option
Input Data Set Options: 
input data setPROC NLPDATA=
initial values and constraintsPROC NLPINEST=
quadratic objective functionPROC NLPINQUAD=
program statementsPROC NLPMODEL=
skip missing value observationsPROC NLPNOMISS
   
Output Data Set Options: 
variables and derivativesPROC NLPOUT=
result parameter valuesPROC NLPOUTEST=
program statementsPROC NLPOUTMODEL=
combine various OUT... statementsPROC NLPOUTALL
CRP Jacobian in the OUTEST= data setPROC NLPOUTCRPJAC
derivatives in the OUT= data setPROC NLPOUTDER=
grid in the OUTEST= data setPROC NLPOUTGRID
Hessian in the OUTEST= data setPROC NLPOUTHESSIAN
iterative output in the OUTEST= data setPROC NLPOUTITER
Jacobian in the OUTEST= data setPROC NLPOUTJAC
NLC Jacobian in the OUTEST= data setPROC NLPOUTNLCJAC
time in the OUTEST= data setPROC NLPOUTTIME
   
Optimization Options: 
minimization methodPROC NLPTECH=
update techniquePROC NLPUPDATE=
version of optimization techniquePROC NLPVERSION=
line-search methodPROC NLPLINESEARCH=
line-search precisionPROC NLPLSPRECISION=
type of Hessian scalingPROC NLPHESCAL=
start for approximated HessianPROC NLPINHESSIAN=
iteration number for update restartPROC NLPRESTART=
   
Initial Value Options: 
produce best grid pointsPROC NLPBEST=
infeasible points in grid searchPROC NLPINFEASIBLE
pseudorandom initial valuesPROC NLPRANDOM=
constant initial valuesPROC NLPINITIAL=
   
Derivative Options: 
finite-difference derivativesPROC NLPFD=
finite-difference derivativesPROC NLPFDHESSIAN=
compute finite-difference intervalPROC NLPFDINT=
use only diagonal of HessianPROC NLPDIAHES
test gradient specificationPROC NLPGRADCHECK=
   
Constraint Options: 
range for active constraintsPROC NLPLCEPSILON=
LM tolerance for deactivatingPROC NLPLCDEACT=
tolerance for dependent constraintsPROC NLPLCSINGULAR=
sum all observations for continuous functionsNLINCON/ SUMOBS
evaluate each observation for continuous functionsNLINCON/ EVERYOBS
   
Termination Criteria Options: 
maximum number of function callsPROC NLPMAXFUNC=
maximum number of iterationsPROC NLPMAXITER=
minimum number of iterationsPROC NLPMINITER=
upper limit on real timePROC NLPMAXTIME=
absolute function convergence criterionPROC NLPABSCONV=
absolute function convergence criterionPROC NLPABSFCONV=
absolute gradient convergence criterionPROC NLPABSGCONV=
absolute parameter convergence criterionPROC NLPABSXCONV=
relative function convergence criterionPROC NLPFCONV=
relative function convergence criterionPROC NLPFCONV2=
relative gradient convergence criterionPROC NLPGCONV=
relative gradient convergence criterionPROC NLPGCONV2=
relative parameter convergence criterionPROC NLPXCONV=
used in FCONV, GCONV criterionPROC NLPFSIZE=
used in XCONV criterionPROC NLPXSIZE=
   
Covariance Matrix Options: 
type of covariance matrixPROC NLPCOV=
\sigma^2 factor of COV matrixPROC NLPSIGSQ=
determine factor of COV matrixPROC NLPVARDEF=
absolute singularity for inertiaPROC NLPASINGULAR=
relative M singularity for inertiaPROC NLPMSINGULAR=
relative V singularity for inertiaPROC NLPVSINGULAR=
threshold for Moore-Penrose inversePROC NLPG4=
tolerance for singular COV matrixPROC NLPCOVSING=
profile confidence limitsPROC NLPCLPARM=
   
Printed Output Options: 
display (almost) all printed outputPROC NLPPALL
suppress all printed outputPROC NLPNOPRINT
reduce some default outputPROC NLPPSHORT
reduce most default outputPROC NLPPSUMMARY
display initial values and gradientsPROC NLPPINIT
display optimization historyPROC NLPPHISTORY
display Jacobian matrixPROC NLPPJACOBI
display crossproduct Jacobian matrixPROC NLPPCRPJAC
display Hessian matrixPROC NLPPHESSIAN
display Jacobian of nonlinear constraintsPROC NLPPNLCJAC
display values of grid pointsPROC NLPPGRID
display values of functions in LSQ, MIN, MAXPROC NLPPFUNCTION
display approximate standard errorsPROC NLPPSTDERR
display covariance matrixPROC NLPPCOV
display eigenvalues for covariance matrixPROC NLPPEIGVAL
print code evaluation problemsPROC NLPPERROR
print measures of real timePROC NLPPTIME
display model program, variablesPROC NLPLIST
display compiled model programPROC NLPLISTCODE
   
Step Length Options: 
damped steps in line searchPROC NLPDAMPSTEP=
maximum trust region radiusPROC NLPMAXSTEP=
initial trust region radiusPROC NLPINSTEP=
   
Profile Point and Confidence Interval Options: 
factor relating discrepancy function to \chi^2 quantilePROFILEFFACTOR=
scale for y values written to OUTEST= data setPROFILEFORCHI=
upper bound for confidence limit searchPROFILEFEASRATIO=
write all confidence limit parameter estimates to OUTEST= data setPROFILEOUTTABLE
   
Miscellaneous Options: 
number of accurate digits in objective functionPROC NLPFDIGITS=
number of accurate digits in nonlinear constraintsPROC NLPCDIGITS=
general singularity criterionPROC NLPSINGULAR=
do not compute inertia of matricesPROC NLPNOEIGNUM
check optimality in neighborhoodPROC NLPOPTCHECK=
   

Previous Page | Next Page | Top of Page