{smcl}
{* 08apr2005}{...}
{cmd:help ml}
{hline}

{title:Title}

{p2colset 4 14 16 2}{...}
{p2col:{hi:[R] ml} {hline 2}}Maximum likelihood estimation{p_end}
{p2colreset}{...}


{title:Syntax}

{pstd}
ml model in interactive mode

{p 8 20 2}
{cmd:ml} {opt mod:el}{space 3}
{it:method progname eq}
[{it:eq ...}]
{ifin}
{weight}
[{cmd:,}
{it:{help ml##model_options:model_options}}
{opt svy}
{it:{help ml##diparm_options:diparm_options}}]

{pstd}
ml model in noninteractive mode

{p 8 20 2}
{cmd:ml} {opt mod:el}{space 3}
{it:method progname eq}
[{it:eq ...}]
{ifin}
{weight}{cmd:,}
{opt max:imize}
[{it:{help ml##model_options:model_options}}
{opt svy}
{it:{help ml##diparm_options:diparm_options}}
{it:{help ml##noninteractive_options:noninteractive_options}}]

{pstd}
Noninteractive mode is invoked by specifying option {opt maximize}.  Use
{opt maximize} when you want to use {opt ml} as a subroutine of another
ado-file or program and you want to carry forth the problem, from definition
to posting of final results, in one command.

{p 8 20 2}
{cmd:ml clear}

{p 8 20 2}
{cmd:ml} {opt q:uery}

{p 8 20 2}
{cmd:ml check}

{p 8 20 2}
{cmd:ml} {opt sea:rch}{space 2}
	[[{cmd:/}]{it:eqname}[{cmd::}]
		{it:#lb} {it:#ub} ]
	[{it:...}]
	[{cmd:,}
	{it:{help ml##search_options:search_options}}]

{p 8 20 2}
{cmd:ml} {opt pl:ot}{space 4}
	[{it:eqname}{cmd::}]{it:name}
	[{it:#} [{it:#} [{it:#}]]]
	[{cmd:,}
	{cmdab:sav:ing:(}{it:filename}[{cmd:,} {opt replace}]{cmd:)}]

{p 8 20 2}
{cmd:ml init}{space 4}
	{c -(}
	[{it:eqname}{cmd::}]{it:name}{cmd:=}{it:#} |
	{cmd:/}{it:eqname}{cmd:=}{it:#}
	{c )-}
	[{it:...}]

{p 8 20 2}
{cmd:ml init}{space 5}{it:#} [{it:#} {it:...}]{cmd:,} {opt copy}

{p 8 20 2}
{cmd:ml init}{space 5}{it:matname} [{cmd:,} {opt copy} {opt skip} ]

{p 8 20 2}
{cmd:ml} {opt rep:ort}

{p 8 20 2}
{cmd:ml trace}{space 4}{c -(} {opt on} | {opt off} {c )-}

{p 8 20 2}
{cmd:ml count}{space 4}[ {opt clear} | {opt on} | {opt off} ]

{p 8 20 2}
{cmd:ml} {opt max:imize} [{cmd:,}
	{it:{help ml##ml_maximize_options:ml_maximize_options}}
	{it:{help ml##display_options:display_options}}
	{it:{help ml##eform_option:eform_option}}]

{p 8 20 2}
{cmd:ml} {opt gr:aph}{space 4}[{it:#}]
	[{cmd:,}
	{cmdab:sav:ing:(}{it:filename}[{cmd:,} {opt replace}]{cmd:)}
	]

{p 8 20 2}
{cmd:ml} {opt di:splay}{space 2}[{cmd:,}
	{it:{help ml##display_options:display_options}}
	{it:{help ml##eform_option:eform_option}}]

{p 8 20 2}
{cmd:ml} {opt foot:note}
 
{phang}
where {it:method} is 
{c -(} {opt lf} | {opt d0} | {opt d1} | {opt d1debug} | {opt d2} | {opt d2debug} {c )-},

{phang}
{it:eq} is the equation to be estimated, enclosed in parentheses, and
optionally with a name to be given to the equation, preceded by a colon,

{p 12 16 2}
{cmd:(}
	[{it:eqname}{cmd::}]
	[{it:varnames} {cmd:=}]
	[{it:varnames}]
	[{cmd:,} {it:eq_options}]
{cmd:)}

{pmore}
or {it:eq} is the name of a parameter, such as sigma with a slash in front

{p 12 12 2}
{cmd:/}{it:eqname}{space 6}which
is equivalent to{space 4}{cmd:(}{it:eqname}{cmd::)}

{pmore}
{marker diparm_options}{...}
and {it:diparm_options} is one or more {opt diparm(diparm_args)} options where
{it:diparm_args} is either {opt __sep__} or anything accepted by the
"undocumented" {opt _diparm} command; see {helpb _diparm}.

{marker eq_options}{...}
{synoptset 28}{...}
{synopthdr:eq_options}
{synoptline}
{synopt:{opt nocons:tant}}do not include an intercept in the equation
{p_end}
{synopt:{opth off:set(varname:varname_o)}}include {it:varname_o} in model with
	coefficient constrained to 1
	{p_end}
{synopt:{opth exp:osure(varname:varname_e)}}include ln({it:varname_e}) in model
	with coefficient constrained to 1
	{p_end}
{synoptline}
{p2colreset}{...}

{marker model_options}{...}
{synoptset 28}{...}
{synopthdr:model_options}
{synoptline}
{synopt:{opt rob:ust}}compute standard errors using the robust/sandwich
estimator{p_end}
{synopt:{opth cl:uster(varname)}}adjust standard errors for intragroups
correlation; implies {opt robust}{p_end}
{synopt:{opth const:raints(numlist)}}constraints by number to be applied{p_end}
{synopt:{opt const:raints(matname)}}matrix that contains the constraints to be
applied{p_end}
{synopt:{opt nocnsnote:s}}do not display notes when constraints are dropped{p_end}
{synopt:{opth ti:tle(data types:string)}}place a title on the estimation output{p_end}
{synopt:{opt nopre:serve}}do not preserve the estimation subsample in
memory{p_end}
{synopt:{opt col:linear}}keep collinear variables within equations{p_end}
{synopt:{opt miss:ing}}keep observations containing variables with missing
values{p_end}
{synopt:{opt lf0(#k #ll)}}number of parameters and
log-likelihood value of the "constant only" model{p_end}
{synopt:{opt cont:inue}}specifies that a model has been fitted and sets the
initial  values for the model to be fitted based on those
results{p_end}
{synopt:{opt wald:test(#)}}perform a Wald test; see
{help ml##mlmode:Options for use with ml model in interactive or noninteractive mode} below
{p_end}
{synopt:{opt obs(#)}}number of observations{p_end}
{synopt:{opt noscvars}}do not create and pass score variables to
likelihood-evaluator program; seldom used{p_end}
{synopt:{opth "crittype(data types:string)"}}describe the criterion optimized
by {opt ml}{p_end}
{synopt:{opth sub:pop(varname)}}compute estimates for the single
subpopulation{p_end}
{synopt:{opt srs:subpop}}compute deff and deft using an estimate of
simple-random-sampling variance{p_end}
{synopt:{opt no}{opt svy:adjust}}carry out Wald test as W/k distributed F(k,d){p_end}
{synopt:{cmdab:tech:nique(nr)}}Stata's modified Newton-Raphson (NR) algorithm{p_end}
{synopt:{cmdab:tech:nique(bhhh)}}Berndt-Hall-Hall-Hausman (BHHH) algorithm{p_end}
{synopt:{cmdab:tech:nique(dfp)}}Davidon-Fletcher-Powell (DFP) algorithm{p_end}
{synopt:{cmdab:tech:nique(bfgs)}}Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm{p_end}
{synopt:{cmd:vce(oim)}}estimate using the observed information matrix{p_end}
{synopt:{cmd:vce(}{cmdab:o:pg)}}estimate using the outer product of the coefficient gradients{p_end}
{synopt:{cmd:vce(native)}}see {help ml##mlmode:Options for use with ml model in interactive or noninteractive mode} below{p_end}
{synoptline}
{p2colreset}{...}

{marker noninteractive_options}{...}
{synoptset 28}{...}
{synopthdr:noninteractive_options}
{synoptline}
{synopt:{opt init:(ml_init_args)}}set the initial parameter values{p_end}
{synopt:{cmdab:sea:rch(on)}}equivalent to {cmd:ml search, repeat(0)}; the
default{p_end}
{synopt:{cmdab:sea:rch(norescale)}}equivalent to {cmd:ml search, repeat(0) norescale}{p_end}
{synopt:{cmdab:sea:rch(quietly)}}same as {cmd:search(on)}, except that output
is suppressed{p_end}
{synopt:{cmdab:sea:rch(off)}}prevents calling {opt ml search}{p_end}
{synopt:{opt r:epeat(#)}}{opt ml search}'s {opt repeat()} option; see below{p_end}
{synopt:{opt b:ounds(ml_search_bounds)}}specify bounds for {opt ml search}
{p_end}
{synopt:{opt nowarn:ing}}suppress "convergence not achieved" message of
{cmd:iterate(0)}{p_end}
{synopt:{opt novce}}substitute the zero matrix for the variance matrix{p_end}
{synopt:{opth sc:ore(newvar:newvars)}}new variables containing the
contribution to the score{p_end}
{synopt:{it:{help ml##noninteractive_maxopts:maximize_options}}}control the
maximization process; seldom used{p_end}
{synoptline}
{p2colreset}{...}

{marker search_options}{...}
{synoptset 28}{...}
{synopthdr:search_options}
{synoptline}
{synopt:{opt r:epeat(#)}}number of random attempts to find better
initial-value vector; default is {cmd:repeat(10)} in interactive mode and
{cmd:repeat(0)} in noninteractive mode{p_end}
{synopt:{opt rest:art}}use random actions to find starting values; not
recommended{p_end}
{synopt:{opt noresc:ale}}do not rescale to improve vector parameter; not
recommended{p_end}
{synopt:{it:{help ml##search_maxopts:maximize_options}}}control the
maximization process; seldom used{p_end}
{synoptline}
{p2colreset}{...}

{marker ml_maximize_options}{...}
{synoptset 28}{...}
{synopthdr:ml_maximize_options}
{synoptline}
{synopt:{opt nowarn:ing}}suppress "convergence not achieved" message of
{cmd:iterate(0)}{p_end}
{synopt:{opt novce}}substitute the zero matrix for the variance matrix{p_end}
{synopt:{cmdab:sc:ore(}{it:{help newvar:newvars}} | {it:stub*}{cmd:)}}new variables containing the contribution
to the score{p_end}
{synopt:{opt noout:put}}suppress the display of the final results{p_end}
{synopt:{opt noclear}}do not clear ml problem definition after model has
converged{p_end}
{synopt:{it:{help ml##ml_maxopts:maximize_options}}}control the
maximization process; seldom used{p_end}
{synoptline}
{p2colreset}{...}

{marker display_options}{...}
{synoptset 28}{...}
{synopthdr:display_options}
{synoptline}
{synopt:{opt noh:eader}}suppress display of the header above the coefficient
table{p_end}
{synopt:{opt nofoot:note}}suppress display of the footnote below the coefficient table{p_end}
{synopt:{opt l:evel(#)}}set confidence level; default is
{cmd:level(95)}{p_end}
{synopt:{opt f:irst}}display coefficient table reporting results for first
equation only{p_end}
{synopt:{opt neq(#)}}display coefficient table reporting first {it:#}
equations{p_end}
{synopt:{opt pl:us}}display coefficient table ending in
dashes{c -}plus-sign{c -}dashes{p_end}
{synoptline}
{p2colreset}{...}

{marker eform_option}{...}
{synoptset 28}{...}
{synopthdr:eform_option}
{synoptline}
{synopt:{opth ef:orm(data types:string)}}display exponentiated coefficients;
column title is "string"{p_end}
{synopt:{opt ef:orm}}display exponentiated coefficients;
column title is "{cmd:exp(b)}"{p_end}
{synopt:{opt hr}}report hazard ratios{p_end}
{synopt:{opt ir:r}}report incidence-rate ratios{p_end}
{synopt:{opt or}}report odds ratios{p_end}
{synopt:{opt rr:r}}report relative risk ratios{p_end}
{synoptline}
{p2colreset}{...}
{p 4 6 2}
{opt fweight}s, {opt aweight}s, {opt iweight}s, and {opt pweight}s are
allowed; see {help weight}.
With all but method lf, you must write your likelihood-evaluation program
carefully if {opt pweight}s are to be specified, and {opt pweight}s may not be
specified with method d0.
{p_end}