/usr/share/doc/octave-optim/html/Common-optimization-options.html

<html lang="en">
<head>
<title>Common optimization options - optim_doc</title>
<meta http-equiv="Content-Type" content="text/html">
<meta name="description" content="optim_doc">
<meta name="generator" content="makeinfo 4.13">
<link title="Top" rel="start" href="index.html#Top">
<link rel="prev" href="Common-frontend-options.html#Common-frontend-options" title="Common frontend options">
<link rel="next" href="Parameter-structures.html#Parameter-structures" title="Parameter structures">
<link href="http://www.gnu.org/software/texinfo/" rel="generator-home" title="Texinfo Homepage">
<!--
Additional documentation for the optim package for Octave.

Copyright (C) <Olaf Till <i7tiol@t-online.de>>

You can redistribute this documentation and/or modify it under the terms
of the GNU General Public License as published by the Free Software
Foundation; either version 3 of the License, or (at your option) any
later version.

This documentation is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
Public License for more details.

You should have received a copy of the GNU General Public License along
with this documentation; if not, see <http://www.gnu.org/licenses/>.-->
<meta http-equiv="Content-Style-Type" content="text/css">
<style type="text/css"><!--
  pre.display { font-family:inherit }
  pre.format  { font-family:inherit }
  pre.smalldisplay { font-family:inherit; font-size:smaller }
  pre.smallformat  { font-family:inherit; font-size:smaller }
  pre.smallexample { font-size:smaller }
  pre.smalllisp    { font-size:smaller }
  span.sc    { font-variant:small-caps }
  span.roman { font-family:serif; font-weight:normal; } 
  span.sansserif { font-family:sans-serif; font-weight:normal; } 
--></style>
</head>
<body>
<div class="node">
<a name="Common-optimization-options"></a>
<p>
Next:&nbsp;<a rel="next" accesskey="n" href="Parameter-structures.html#Parameter-structures">Parameter structures</a>,
Previous:&nbsp;<a rel="previous" accesskey="p" href="Common-frontend-options.html#Common-frontend-options">Common frontend options</a>,
Up:&nbsp;<a rel="up" accesskey="u" href="index.html#Top">Top</a>
<hr>
</div>

<h2 class="chapter">8 Options common to all optimization frontends</h2>

<p><a name="index-common-optimization-options-94"></a>
All frontends for optimization (<a href="nonlin_005fmin.html#nonlin_005fmin">nonlin_min</a>, <a href="nonlin_005fresidmin.html#nonlin_005fresidmin">nonlin_residmin</a>,
<a href="nonlin_005fcurvefit.html#nonlin_005fcurvefit">nonlin_curvefit</a>) accept the following options, settable with
<a href="../octave/XREFoptimset.html#XREFoptimset">optimset</a>.

<h4 class="subheading">Settings handled within the frontend</h4>

     <dl>
<dt><code>Algorithm</code><dd>String specifying the backend. 
<br><dt><code>complex_step_derivative_inequc,</code><br><dt><code>complex_step_derivative_equc</code><dd>Logical scalars, default: <code>false</code>. Estimate Jacobian of general
inequality constraints and equality constraints, respectively, with
complex step derivative approximation. Use only if you know that your
function of general inequality constraints or function of general
equality constraints, respectively, is suitable for this. No user
function for the respective Jacobian must be specified. 
</dl>

<h4 class="subheading">Settings passed to the backend</h4>

<p>Which of these options are actually honored is noted in the descriptions
of the individual backends.

     <dl>
<dt><code>lbound,</code><br><dt><code>ubound</code><dd>Column vectors of lower and upper bounds for parameters.  Default:
<code>-Inf</code> and <code>+Inf</code>, respectively.  The bounds are non-strict,
i.e. parameters are allowed to be exactly equal to a bound.  The default
function for gradients or Jacobians will respect bounds (but no further
inequality constraints) in finite differencing if the backend respects
bounds even during the course of optimization. 
<br><dt><code>inequc</code><dd>Further inequality constraints.  Cell-array containing up to four
entries, two entries for linear inequality constraints and/or one or two
entries for general inequality constraints.  Either linear or general
constraints may be the first entries, but the two entries for linear
constraints must be adjacent and, if two entries are given for general
constraints, they also must be adjacent.  The two entries for linear
constraints are a matrix (say <code>m</code>) and a vector (say <code>v</code>),
specifying linear inequality constraints of the form <code>m.' *
parameters + v &gt;= 0</code>.  The first entry for general constraints must be a
differentiable column-vector valued function (say <code>h</code>), specifying
general inequality constraints of the form <code>h (p[, idx]) &gt;= 0</code>;
<code>p</code> is the column vector of optimized paraters and the optional
argument <code>idx</code> is a logical index.  <code>h</code> has to return the
values of all constraints if <code>idx</code> is not given.  It may choose to
return only the indexed constraints if <code>idx</code> is given (so
computation of the other constraints can be spared); in this case, the
additional setting <code>f_inequc_idx</code> has to be set to <code>true</code>.  In
gradient determination, this function may be called with an
informational third argument, whose content depends on the function for
gradient determination.  If a second entry for general inequality
constraints is given, it must be a function computing the jacobian of
the constraints with respect to the parameters.  For this function, the
description of the setting <code>dfdp</code>, see
 <a href="XREFoptiondfdp.html#XREFoptiondfdp">dfdp</a>,
applies, with 2 exceptions: 1) it is called with 3 arguments since it
has an additional argument <code>idx</code>, a logical index, at second
position, indicating which rows of the jacobian must be returned (if the
function chooses to return only indexed rows, the additional setting
<code>df_inequc_idx</code> has to be set to <code>true</code>).  2) the default
jacobian function calls <code>h</code> with 3 arguments, since the argument
<code>idx</code> is also supplied.  Note that specifying linear constraints as
general constraints will generally waste performance, even if further,
non-linear, general constraints are also specified. 
<br><dt><code>f_inequc_idx,</code><br><dt><code>df_inequc_idx</code><dd>Indicate that functions for general inequality constraints or their
jacobian, respectively, return only the values or derivatives for the
indexed parameters.  See description of setting <code>inequc</code> above. 
<br><dt><code>equc</code><dd>Equality constraints.  Specified the same way as inequality constraints
(see <code>inequc</code> above). 
<br><dt><code>f_equc_idx,</code><br><dt><code>df_equc_idx</code><dd>As <code>f_inequc_idx</code> and <code>df_inequc_idx</code> above, but for equality
constraints. 
<br><dt><code>cpiv</code><dd>Function for complementary pivoting, usable in algorithms for
constraints.  Default: <code>cpiv_bard</code>.  Only the default function is
supplied with the package. 
<br><dt><code>TolFun</code><dd>Minimum fractional improvement in objective function (e.g. sum of
squares) in an iteration (termination criterium).  Default: .0001. 
<br><dt><code>TolX</code><dd>Minimum fractional change in a norm of the parameters in an iteration
(termination criterium).  Default: backend specific. 
<br><dt><code>MaxIter</code><dd>Maximum number of iterations (termination criterium).  Default:
backend-specific. 
<br><dt><code>fract_prec</code><dd>Column Vector, minimum fractional changes of corresponding parameters in
an iteration (termination criterium if violated in two consecutive
iterations).  Default: backend-specific. 
<br><dt><code>max_fract_change</code><dd>Column Vector, enforced maximum fractional changes in corresponding
parameters in an iteration. Default: backend-specific. 
<br><dt><code>Display</code><dd>String indicating the degree of verbosity. Default:
<code>"off"</code>. Possible values are currently <code>"off"</code> (no messages)
and <code>"iter"</code> (some messages after each iteration).  Support of
this setting and its exact interpretation are backend-specific. 
<br><dt><code>debug</code><dd>Logical scalar, default: <code>false</code>. Will be passed to the backend,
which might print debugging information if <code>true</code>. 
<br><dt><code>FunValCheck</code><dd>If <code>"on"</code>, the output of user functions will be sanity-checked. 
Default: <code>"off"</code>. 
<br><dt><code>user_interaction</code><dd><!-- This setting has deliberately not been named as its Matlab equivalent -->
<!-- `OutputFcn' since it differs from the latter by requiring the -->
<!-- functions to return _two_ outputs. The rationale for the difference -->
<!-- is that information about the reason for a user-stop should be -->
<!-- possible to pass in the output. The second output can't be made -->
<!-- optional without possibly altering the warning state under which the -->
<!-- user-function runs. -->
Handle to a user function or cell-array with a number of these. 
Functions must have this interface:
     <pre class="example">          [<var>stop</var>, <var>info</var>] = some_user_function (<var>p</var>, <var>vals</var>,
                                                         <var>state</var>);
</pre>
     <p>If <var>stop</var> is <code>true</code>, the algorithm stops.  In <var>info</var>
information about the reason for stopping can be returned in a free
format.  <var>info</var> can be set to be empty, but it must be set.  Note
that this is different from the otherwise similar Matlab setting
<code>OutputFcn</code>.  The functions will be called by the algorithms at the
start with <var>state</var> set to <code>init</code>, after each iteration with
<var>state</var> set to <code>iter</code>, and at the end with <var>state</var> set to
<code>done</code>.  <var>p</var> contains the current parameters, and <var>vals</var>
is a structure with other current values, the possible fields are
currently:
          <dl>
<dt><code>iteration</code><dd>number of the current iteration,
<br><dt><code>fval</code><dd>value of objective function (for scalar optimization),
<br><dt><code>residual</code><dd>residuals (for residual-based optimization),
<br><dt><code>model_y</code><dd>in <code>nonlin_curvefit</code>, the output of the model function,
<br><dt><code>observations</code><dd>in <code>nonlin_curvefit</code>, the constant observations,
<br><dt><code>model_x</code><dd>in <code>nonlin_curvefit</code>, the constant argument <var>x</var>. 
</dl>
     Information about the output of these functions when they were called
the last time (possibly causing a stop) will be contained in the output
<var>outp</var> of the frontend in field <code>user_interaction</code>.  Subfield
<code>stop</code> is a vector containing the <var>stop</var> outputs of each
function, subfield <code>info</code> is a cell-array containing the output
<var>info</var> of each function.  In the case of a stop, the output
<var>cvg</var> of the frontent will be <code>-1</code>. 
</dl>

<!--  -->
   </body></html>
octave-optim 1.4.1-2 / usr / share / doc / octave-optim / html / Common-optimization-options.html
