algorithm used to maximize the likelihood function.  For {cmd:technique(nr)},
{cmd:vce(native)} is a synonym for {cmd:vce(oim)}; and for
{cmd:technique(bhhh)}, {cmd:vce(native)} is a synonym for {cmd:vce(opg)}.
{cmd:vce(native)} is not allowed when switching optimization techniques.


{title:Options for use with ml model in noninteractive mode}

{pstd}
The following additional options are for use with {opt ml model} in
noninteractive mode.  Noninteractive mode is for programmers who use {opt ml}
as a subroutine and want to issue a single command that will carry forth the
estimation from start to finish.

{phang}
{opt maximize} is required.  It specifies noninteractive mode.

{phang}
{opt init(ml_init_args)} sets the initial parameter values.
{it:ml_init_args} are whatever you would type after the {opt ml init} command.

{phang}
{cmd:search(}{opt on}|{opt norescale}|{opt quietly}|{opt off}{cmd:)} specifies whether
{cmd:ml search} is to be used to improve the initial values.  {cmd:search(on)}
is the default and is equivalent to running separately
{cmd:ml search, repeat(0)}.  {cmd:search(norescale)} is equivalent to running
separately {cmd:ml search, repeat(0) norescale}.  {cmd:search(quietly)} is the
same as {cmd:search(on)}, except that it suppresses {cmd:ml search}'s output.
{cmd:search(off)} prevents the calling of {cmd:ml search} altogether.

{phang}
{opt repeat(#)} is {opt ml search}'s {opt repeat()} option.  {cmd:repeat(0)}
is the default.

{phang}
{opt bounds(ml_search_bounds)} specifies the search bounds.  The command
{opt ml model} issues is
"{opt ml search} {it:ml_search_bounds}{opt , repeat(#)}".  Specifying search
bounds is optional.

{phang}
{opt nowarning}, {opt novce}, and {opt score()} are {opt ml maximize}'s
equivalent options.

{phang}{marker noninteractive_maxopts}
{it:maximize_options}:
{opt dif:ficult},
{opt tech:nique(algorithm_spec)},
{opt iter:ate(#)},
[{cmdab:no:}]{opt lo:g},
{opt tr:ace},
{opt grad:ient},
{opt showstep},
{opt hess:ian},
{opt shownr:tolerance},
{opt tol:erance(#)},
{opt ltol:erance(#)},
{opt gtol:erance(#)},
{opt nrtol:erance(#)},
{opt nonrtol:erance},
{opt from(init_specs)}; see {opt maximize}.  These options are seldom used.


{title:Options for use when specifying equations}

{phang}
{opt noconstant} specifies that the equation not include an intercept.

{phang}
{opth "offset(varname:varname_o)"} specifies that the equation be
xb + {it:varname_o}{hline 2}that it include {it:varname_o} with
coefficient constrained to be 1.

{phang}
{opth "exposure(varname:varname_e)"} is an alternative to
{opt offset(varname_o)}; it specifies that the equation be xb +
ln({it:varname_e}).  The equation is to include ln({it:varname_e}) with
coefficient constrained to be 1.


{title:Options for use with ml search}

{phang}
{opt repeat(#)} specifies the number of random attempts that
are to be made to find a better initial-value vector.  The default is
{cmd:repeat(10)}.

{pmore}
{cmd:repeat(0)} specifies that no random attempts be made.
More correctly, {cmd:repeat(0)} specifies that no random attempts be made if
the initial initial-value vector is a feasible starting point.  If it is not,
{opt ml search} will make random attempts, even if you specify
{cmd:repeat(0)}, because it has no alternative.  The {opt repeat()} option
refers to the number of random attempts to be made to improve the initial
values.  When the initial starting value vector is not feasible,
{opt ml search} will make up to 1,000 random attempts to find starting values.
It stops the instant it finds one set of values that works and then moves into
its improve-initial-values logic.

{pmore}
{opt repeat(k)}, {it:k} > 0, specifies the number of random attempts to be made to
improve the initial values.

{phang}
{opt restart} specifies that random actions be taken to obtain
starting values and that the resulting starting values not be a
deterministic function of the current values.  Generally,
you should not specify this option because, with {cmd:restart}, {cmd:ml search}
intentionally does not produce as good a set of starting values as it could.
{opt restart} is included for use by the optimizer when it gets into serious
trouble.  The random actions ensure that the actions of the optimizer and
{cmd:ml search}, working together, do not result in a long, endless loop.
See {bf:[R] ml} for details.

{phang}
{opt norescale} specifies that {opt ml search} not engage in its
rescaling actions to improve the parameter vector.  We do not recommend
specifying this option because rescaling tends to work so well.

{phang}{marker search_maxopts}
{it:maximize_options}:
[{cmdab:no:}]{cmdab:lo:g},
{opt tr:ace};
see {help maximize}.  These options are seldom used.


{title:Option for use with ml plot}

{phang}
{helpb saving_option:saving({it:filename}[, replace])} specifies that the
graph be saved in {it:filename}{cmd:.gph}.


{title:Options for use with ml init}

{phang}
{opt copy} specifies that the list of numbers or the initialization
vector be copied into the initial-value vector by position rather than
by name.

{phang}
{opt skip} specifies that any parameters found in the specified
initialization vector that are not also found in the model be ignored.
The default action is to issue an error message.


{title:Options for use with ml maximize}

{phang}
{opt nowarning} is allowed only with {cmd:iterate(0)}.  {opt nowarning}
suppresses the "convergence not achieved" message.
Programmers might specify {cmd:iterate(0) nowarning} when they have a vector
already containing the final estimates and want {opt ml} to calculate the
variance matrix and post final estimation results.  In that case, specify
{bind:{cmd:init(b) search(off) iterate(0) nowarning nolog}}.

{phang}
{opt novce} is allowed only with {cmd:iterate(0)}.  {opt novce}
substitutes the zero matrix for the variance matrix, which in effect posts
estimation results as fixed constants.

{phang}
{cmd:score(}{it:{help newvar:newvars}} | {it:stub*}{cmd:)}
creates new variables containing the contributions to the score for each
equation and ancillary parameter in the model; see
{bf:[U] 20.15 Obtaining scores}.

{pmore}
If {opt score(newvars)} is specified, the {it:newvars} must contain k
new variables, one for each equation in the model.  If {opt score(stub*)} is
specified, variables named {it:stub}{cmd:1}, {it:stub}{cmd:2}, {it:...}
{it:stub}{cmd:k} are created.

{pmore}
The first variable contains {bind:d(ln L_j)/d(x_j b)}; the second variable
contains {bind:d(ln L_j)/d(ln(alpha))}; and so on.

{phang}
{opt nooutput} suppresses display of the final results.  This is
different from prefixing {opt ml maximize} with {opt quietly} in that the
iteration log is still displayed (assuming that {opt nolog} is not specified).

{phang}
{opt noclear} specifies that after the model has converged, the ml
problem definition not be cleared.  Perhaps you are having convergence
problems and intend to run the model to convergence.  If so, use {opt ml search}
to see if those values can be improved, and then start the estimation again.

{phang}{marker ml_maxopts}
{it:maximize_options}:
{opt dif:ficult},
{opt iter:ate(#)},
[{cmdab:no:}]{opt lo:g},
{opt tr:ace},
{opt grad:ient},
{opt showstep},
{opt hess:ian},
{opt shownr:tolerance},
{opt tol:erance(#)},
{opt ltol:erance(#)},
{opt gtol:erance(#)},
{opt nrtol:erance(#)},
{opt nonrtol:erance}; see {help maximize}.  These options are seldom used.

{phang}
{it:display_options}; see
{help ml##mldisplay:Options for use with ml display}.

{phang}
{it:eform_option}; see
{help ml##mldisplay:Options for use with ml display}.


{title:Option for use with ml graph}

{phang}
{helpb saving_option:saving({it:filename}[, replace])} specifies that
the graph is to be saved in {it:filename}{cmd:.gph}.


{marker mldisplay}{...}
{title:Options for use with ml display}

{phang}
{opt noheader} suppresses the display of the header above the coefficient
table that displays the final log-likelihood value, the number of
observations, and the model significance test.

{phang}
{opt nofootnote} suppresses the display of the footnote below the coefficient
table, which displays a warning if the model fitted did not converge within
the specified number of iterations.  Use {cmd:ml} {cmd:footnote} to display
the warning (1) if you add to the coefficient table using the {cmd:plus}
option or (2) you have your own footnotes and want the warning to be last.

{phang}
{opt level(#)} is the standard confidence-level option.  It
specifies the confidence level, as a percentage, for confidence intervals of the
coefficients.  The default is {cmd:level(95)} or as set by {opt set level};
see {help level}.

{phang}
{opt first} displays a coefficient table reporting results for the
first equation only, and the report makes it appear that the first equation is
the only equation.  This is used by programmers who estimate ancillary
parameters in the second and subsequent equations and who wish to report the values
of such parameters themselves.

{phang}
{opt neq(#)} is an alternative to {opt first}.
{opt neq(#)} displays a coefficient table reporting results for
the first {it:#} equations.  This is used by programmers who estimate ancillary
parameters in the {it:#}+1 and subsequent equations and who wish to report the
values of such parameters themselves.

{phang}
{opt plus} displays the coefficient table, but then rather than ending the
table in a line of dashes, ends it in dashes{c -}plus-sign{c -}dashes.  This
is so that programmers can write additional display code to add more results
to the table and make it appear as if the combined result is one table.
Programmers typically specify {cmd:plus} with options {opt first} or
{opt neq()}.  This option implies {cmd:nofootnote}.

{phang}
{it:eform_option}: {opth "eform(data types:string)"}, {opt eform}, {opt hr},
{opt irr}, {opt or}, and
{opt rrr} display the coefficient table in exponentiated form:
for each coefficient, exp({it:b}) rather than {it:b} is displayed, and standard
errors and confidence intervals are transformed.  Display of the intercept, if
any, is suppressed.  {it:string} is the table header that will be displayed
above the transformed coefficients and must be 11 characters or fewer in
length, for example, {cmd:eform("Odds ratio")}.
The options {opt eform}, {opt hr}, {opt irr}, {opt or}, and {opt rrr}
provide a default {it:string} equivalent to {cmd:"exp(b)"}, {cmd:"Haz.  Ratio"},
{cmd:"IRR"}, {cmd:"Odds Ratio"}, and {cmd:"RRR"}, respectively.  These options may
not be combined.


{title:Examples}

{pstd}
See {bf:[R] ml} for examples.  More examples are available in
{it:{browse "http://www.stata.com/bookstore/mle.html":Maximum Likelihood Estimation with Stata, 2nd edition}}
(Gould, Pitblado, and Sribney 2003){hline 2}available from StataCorp.


{title:Also see}

{psee}
Manual:
{bf:[R] ml}

{psee}
Online:  
{helpb _estimates},
{helpb matrix},
{help maximize},
{helpb ml score},
{helpb mleval},
{helpb nl}{p_end}