The arboretum procedure

The DMREG Procedure

NLOPTIONS Statement

Specifies options for nonlinear optimizations. These options only apply to logistic regression

models.

NLOPTIONS nonlinear-option(s);

Nonlinear-Options

ABSCONV= number

Specifies an absolute function convergence criterion. ABSCONV= is a function of the

log-likelihood for the intercept-only model. The optimization is to maximize the log-likelihood.

Default:__Range:___number'>Default:

The default value is 1e-3 times the log-likelihood of the null model

(intercept-only model).

Range:

number > 0

ABSFCONV= number

Specifies an absolute function convergence criterion.

Default:

 times the log-likelihood of the intercept-only model

Range:

number > 0

ABSGCONV= number

Specifies the absolute gradient convergence criterion.

Default:

1E-5

Range:

number > 0

ABSXCONV= number

Specifies the absolute parameter convergence criterion.

Default:

1E-8

Range:

number > 0

DAMPSTEP= number

Specifies that the initial step size value for each line search used by the QUANEW, CONGRA, or

NEWRAP techniques cannot be larger than the product of number and the step size value used in

the previous iteration.

---

Default:

2

Range:

number > 0

DIAHES

Forces the optimization algorithm (TRUREG, NEWRAP, or NRRIDG) to take advantage of the

diagonality.

FCONV= number

Specifies a function convergence criterion.

Default:

, where FDIGITS is the value of the FDIGITS= option.

Range:

number > 0

FCONV2= number

Specifies another function convergence criterion.

Default:

Range:

number > 0

FDIGITS= number

Specifies the number of accurate digits in evaluations of the objective function.

Default:

, where   is the machine precision.

Range:

number > 0

FSIZE= number

Specifies the parameter of the relative function and relative gradient termination criteria.

Default:

0

Range:

number   0

GCONV= number

Specifies the relative gradient convergence criterion.

Default:

Range:

number > 0

GCONV2= number

Specifies another relative gradient convergence criterion.

Default:

0

---

Range:

number > 0

HESCAL= 0 | 1 | 2 |3

Specifies the scaling version of the Hessian or cross-product Jacobian matrix used in NRRIDG,

TRUREG, LEVMAR, NEWRAP, or DBLDOG optimization.

Default:

1 - for LEVMAR minimization technique

0 - for all others

INHESSIAN= number

Specifies how to define the initial estimate of the approximate Hessian for the quasi-Newton

techniques QUANEW and DBLDOG.

Range:

number   0

Default:

The default is to use a Hessian based on the initial estimates as the initial

estimate of the approximate Hessian. When r=0, the initial estimate of the

approximate Hessian is computed from the magnitude of the initial gradient.

INSTEP= number

Specifies a larger or smaller radius of the trust region used in the TRUREG, DBLDOG, and

LEVMAR algorithms.

Default:

1

Range:

number > 0

LINESEARCH= number

Specifies the line-search method for the CONGRA, QUANEW, and NEWRAP optimization

techniques.

Default:

2

Range:

1   number   8

LSPRECISION= number

Specifies the degree of accuracy that should be obtained by the second and third line-search

algorithms.

---

Default:

Table of Line-Search Precision Values

TECHNIQUE= UPDATE= LSPRECISION

VALUE

QUANEW

DBFGS,

BFGS

0.4

QUANEW

DDFP,

DFP

0.06

CONGRA

all

0.1

NEWRAP

no update

0.9

Range:

number > 0

MAXFUNC= number

Specifies the maximum number of function calls in the optimization process. The objective

function that is minimized is the negative log-likelihood.

Default:

125 for TRUREG, NRRIDG, and NEWRAP.

500 for QUANEW and DBLDOG.

1000 for CONGRA.

Range:

number > 0

MAXITER= number

Specifies the maximum number of iterations in the optimization process.

Default:

50 for TRUREG, NRRIDG and NEWRAP

200 for QUANEW and DBLDOG

400 for CONGRA

Range:

number > 0

MAXSTEP= number

Specifies the upper bound for the step length of the line-search algorithms.

Default:

The largest double precision value

Range:

number > 0

---