source: sasview/dev/api/park.html @ a462c6a

gh-pages
Last change on this file since a462c6a was a462c6a, checked in by ajj, 9 years ago

Rebuild to fix index and modules docs
  • Property mode set to 100644
File size: 160.6 KB
Line 
1<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
2  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
3
4
5<html xmlns="http://www.w3.org/1999/xhtml">
6  <head>
7    <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
8   
9    <title>park package &mdash; SasView 3.0.0 documentation</title>
10   
11    <link rel="stylesheet" href="../../_static/default.css" type="text/css" />
12    <link rel="stylesheet" href="../../_static/pygments.css" type="text/css" />
13   
14    <script type="text/javascript">
15      var DOCUMENTATION_OPTIONS = {
16        URL_ROOT:    '../../',
17        VERSION:     '3.0.0',
18        COLLAPSE_INDEX: false,
19        FILE_SUFFIX: '.html',
20        HAS_SOURCE:  true
21      };
22    </script>
23    <script type="text/javascript" src="../../_static/jquery.js"></script>
24    <script type="text/javascript" src="../../_static/underscore.js"></script>
25    <script type="text/javascript" src="../../_static/doctools.js"></script>
26    <script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
27    <link rel="top" title="SasView 3.0.0 documentation" href="../../index.html" />
28    <link rel="up" title="lib.macosx-10.5-x86_64-2.7" href="modules.html" />
29    <link rel="next" title="sas package" href="sas.html" />
30    <link rel="prev" title="lib.macosx-10.5-x86_64-2.7" href="modules.html" /> 
31  </head>
32  <body>
33    <div class="related">
34      <h3>Navigation</h3>
35      <ul>
36        <li class="right" style="margin-right: 10px">
37          <a href="../../genindex.html" title="General Index"
38             accesskey="I">index</a></li>
39        <li class="right" >
40          <a href="../../py-modindex.html" title="Python Module Index"
41             >modules</a> |</li>
42        <li class="right" >
43          <a href="sas.html" title="sas package"
44             accesskey="N">next</a> |</li>
45        <li class="right" >
46          <a href="modules.html" title="lib.macosx-10.5-x86_64-2.7"
47             accesskey="P">previous</a> |</li>
48        <li><a href="../../index.html">SasView 3.0.0 documentation</a> &raquo;</li>
49          <li><a href="../dev.html" >Developer Documentation</a> &raquo;</li>
50          <li><a href="modules.html" accesskey="U">lib.macosx-10.5-x86_64-2.7</a> &raquo;</li> 
51      </ul>
52    </div> 
53
54    <div class="document">
55      <div class="documentwrapper">
56        <div class="bodywrapper">
57          <div class="body">
58           
59  <div class="section" id="park-package">
60<h1>park package</h1>
61<div class="section" id="submodules">
62<h2>Submodules</h2>
63</div>
64<div class="section" id="module-park.assembly">
65<span id="park-assembly-module"></span><h2>park.assembly module</h2>
66<p>An assembly is a collection of fitting functions.  This provides
67the model representation that is the basis of the park fitting engine.</p>
68<p>Models can range from very simple one dimensional theory functions
69to complex assemblies of multidimensional datasets from different
70experimental techniques, each with their own theory function and
71a common underlying physical model.</p>
72<div class="section" id="usage">
73<h3>Usage</h3>
74<p>First define the models you want to work with.  In the example
75below we will use an example of a simple multilayer system measured by
76specular reflection of xrays and neutrons.  The gold depth is the only
77fitting parameter, ranging from 10-30 A.  The interface depths are
78tied together using expressions.  In this case the expression is
79a simple copy, but any standard math functions can be used.  Some
80model developers may provide additional functions for use with the
81expression.</p>
82<p>Example models:</p>
83<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">reflectometry.model1d</span> <span class="kn">as</span> <span class="nn">refl</span>
84<span class="n">xray</span> <span class="o">=</span> <span class="n">refl</span><span class="o">.</span><span class="n">model</span><span class="p">(</span><span class="s">&#39;xray&#39;</span><span class="p">)</span>
85<span class="n">xray</span><span class="o">.</span><span class="n">incident</span><span class="p">(</span><span class="s">&#39;Air&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
86<span class="n">xray</span><span class="o">.</span><span class="n">interface</span><span class="p">(</span><span class="s">&#39;iAu&#39;</span><span class="p">,</span><span class="n">sigma</span><span class="o">=</span><span class="mi">5</span><span class="p">)</span>
87<span class="n">xray</span><span class="o">.</span><span class="n">layer</span><span class="p">(</span><span class="s">&#39;Au&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mf">124.68</span><span class="p">,</span><span class="n">depth</span><span class="o">=</span><span class="p">[</span><span class="mi">10</span><span class="p">,</span><span class="mi">30</span><span class="p">])</span>
88<span class="n">xray</span><span class="o">.</span><span class="n">interface</span><span class="p">(</span><span class="s">&#39;iSi&#39;</span><span class="p">,</span><span class="n">sigma</span><span class="o">=</span><span class="mi">5</span><span class="p">)</span>
89<span class="n">xray</span><span class="o">.</span><span class="n">substrate</span><span class="p">(</span><span class="s">&#39;Si&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mf">20.07</span><span class="p">)</span>
90<span class="n">datax</span> <span class="o">=</span> <span class="n">refl</span><span class="o">.</span><span class="n">data</span><span class="p">(</span><span class="s">&#39;xray.dat&#39;</span><span class="p">)</span>
91
92<span class="n">neutron</span> <span class="o">=</span> <span class="n">refl</span><span class="o">.</span><span class="n">model</span><span class="p">(</span><span class="s">&#39;neutron&#39;</span><span class="p">)</span>
93<span class="n">neutron</span><span class="o">.</span><span class="n">incident</span><span class="p">(</span><span class="s">&#39;Air&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
94<span class="n">neutron</span><span class="o">.</span><span class="n">interface</span><span class="p">(</span><span class="s">&#39;iAu&#39;</span><span class="p">,</span><span class="n">sigma</span><span class="o">=</span><span class="s">&#39;xray.iAu&#39;</span><span class="p">)</span>
95<span class="n">neutron</span><span class="o">.</span><span class="n">layer</span><span class="p">(</span><span class="s">&#39;Au&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mf">4.66</span><span class="p">,</span><span class="n">depth</span><span class="o">=</span><span class="s">&#39;xray.Au.depth&#39;</span><span class="p">)</span>
96<span class="n">neutron</span><span class="o">.</span><span class="n">interface</span><span class="p">(</span><span class="s">&#39;iSi&#39;</span><span class="p">,</span><span class="n">sigma</span><span class="o">=</span><span class="s">&#39;xray.iSi&#39;</span><span class="p">)</span>
97<span class="n">neutron</span><span class="o">.</span><span class="n">substrate</span><span class="p">(</span><span class="s">&#39;Si&#39;</span><span class="p">,</span><span class="n">rho</span><span class="o">=</span><span class="mf">2.07</span><span class="p">)</span>
98<span class="n">datan</span> <span class="o">=</span> <span class="n">refl</span><span class="o">.</span><span class="n">data</span><span class="p">(</span><span class="s">&#39;neutron.dat&#39;</span><span class="p">)</span>
99</pre></div>
100</div>
101<p>As you can see from the above, parameters can be set to a value if
102the parameter is fixed, to a range if the parametemr is fitted, or
103to a string expression if the parameter is calculated from other
104parameters.  See park.Parameter.set for further details.</p>
105<p>Having constructed the models, we can now create an assembly:</p>
106<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">park</span>
107<span class="n">assembly</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">Assembly</span><span class="p">([(</span><span class="n">xray</span><span class="p">,</span><span class="n">datax</span><span class="p">),</span> <span class="p">(</span><span class="n">neutron</span><span class="p">,</span><span class="n">datan</span><span class="p">)])</span>
108</pre></div>
109</div>
110<p>Note: this would normally be done in the context of a fit
111using fit = park.Fit([(xray,datax), (neutron,datan)]), and later referenced
112using fit.assembly.</p>
113<p>Individual parts of the assembly are accessable using the
114model number 0, 1, 2... or by the model name.  In the above,
115assembly[0] and assembly[&#8216;xray&#8217;] refer to the same model.
116Assemblies have insert and append functions for adding new
117models, and &#8220;del model[idx]&#8221; for removing them.</p>
118<p>Once the assembly is created computing the values for the system
119is a matter of calling:</p>
120<div class="highlight-python"><div class="highlight"><pre><span class="n">assembly</span><span class="o">.</span><span class="n">eval</span><span class="p">()</span>
121<span class="k">print</span> <span class="s">&quot;Chi**2&quot;</span><span class="p">,</span><span class="n">assembly</span><span class="o">.</span><span class="n">chisq</span>
122<span class="k">print</span> <span class="s">&quot;Reduced chi**2&quot;</span><span class="p">,</span><span class="n">assembly</span><span class="o">.</span><span class="n">chisq</span><span class="o">/</span><span class="n">assembly</span><span class="o">.</span><span class="n">degrees_of_freedom</span>
123<span class="n">plot</span><span class="p">(</span><span class="n">arange</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">assembly</span><span class="o">.</span><span class="n">residuals</span><span class="p">)),</span> <span class="n">assembly</span><span class="o">.</span><span class="n">residuals</span><span class="p">)</span>
124</pre></div>
125</div>
126<p>This defines the attributes residuals, degrees_of_freedom and chisq,
127which is what the optimizer uses as the cost function to minimize.</p>
128<p>assembly.eval uses the current values for the parameters in the
129individual models.  These parameters can be changed directly
130in the model.  In the reflectometry example above, you could
131set the gold thickness using xray.layer.Au.depth=156, or
132something similar (the details are model specific).  Parameters
133can also be changed through the assembly parameter set.  In the same
134example, this would be assembly.parameterset[&#8216;xray&#8217;][&#8216;Au&#8217;][&#8216;depth&#8217;].
135See parameter set for details.</p>
136<p>In the process of modeling data, particularly with multiple
137datasets, you will sometimes want to temporarily ignore
138how well one of the datasets matches so that you
139can more quickly refine the model for the other datasets,
140or see how particular models are influencing the fit.  To
141temporarily ignore the xray data in the example above use:</p>
142<div class="highlight-python"><div class="highlight"><pre><span class="n">assembly</span><span class="o">.</span><span class="n">parts</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">isfitted</span> <span class="o">=</span> <span class="bp">False</span>
143</pre></div>
144</div>
145<p>The model itself isn&#8217;t ignored since its parameters may be
146needed to compute the parameters for other models.  To
147reenable checking against the xray data, you would assign
148a True value instead.  More subtle weighting of the models
149can be controlled using assembly.parts[idx].weight, but
150see below for a note on model weighting.</p>
151<div class="section" id="a-note-on-model-weighting">
152<h4>A note on model weighting</h4>
153<p>Changing the weight is equivalent to scaling the error bars
154on the given model by a factor of weight/n where n is the
155number of data points.  It is better to set the correct error
156bars on your data in the first place than to adjust the weights.
157If you have the correct error bars, then you should expect
158roughly 2/3 of the data points to lie within one error bar of
159the theory curve.  If consecutive data points have largely
160overlapping errorbars, then your uncertainty is overestimated.</p>
161<p>Another case where weights are adjusted (abused?) is to
162compensate for systematic errors in the data by forcing the
163errorbars to be large enough to cover the systematic bias.
164This is a poor approach to the problem.  A better strategy
165is to capture the systematic effects in the model, and treat
166the measurement of the independent variable as an additional
167data point in the fit.  This is still not statistically sound
168as there is likely to be a large correlation between the
169uncertainty of the measurement and the values of all the
170other variables.</p>
171<p>That said, adjusting the weight on a dataset is a quick way
172of reducing its influence on the entire fit.  Please use it
173with care.</p>
174<dl class="class">
175<dt id="park.assembly.Assembly">
176<em class="property">class </em><tt class="descclassname">park.assembly.</tt><tt class="descname">Assembly</tt><big>(</big><em>models=[]</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly"><span class="viewcode-link">[source]</span></a></dt>
177<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
178<p>Collection of fit models.</p>
179<p>Assembly implements the <cite>park.fit.Objective</cite> interface.</p>
180<p>See <cite>park.assembly</cite> for usage.</p>
181<p>Instance variables:</p>
182<dl class="docutils">
183<dt>residuals <span class="classifier-delimiter">:</span> <span class="classifier">array</span></dt>
184<dd>a vector of residuals spanning all models, with model
185weights applied as appropriate.</dd>
186<dt>degrees_of_freedom <span class="classifier-delimiter">:</span> <span class="classifier">integer</span></dt>
187<dd>length of the residuals - number of fitted parameters</dd>
188<dt>chisq <span class="classifier-delimiter">:</span> <span class="classifier">float</span></dt>
189<dd>sum squared residuals; this is not the reduced chisq, which
190you can get using chisq/degrees_of_freedom</dd>
191</dl>
192<p>These fields are defined for the individual models as well, with
193degrees of freedom adjusted to the length of the individual data
194set.  If the model is not fitted or the weight is zero, the residual
195will not be calculated.</p>
196<p>The residuals fields are available only after the model has been
197evaluated.</p>
198<dl class="method">
199<dt id="park.assembly.Assembly.abort">
200<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.abort"><span class="viewcode-link">[source]</span></a></dt>
201<dd><p>Interrupt the current function evaluation.</p>
202<p>Forward this to the currently executing model if possible.</p>
203</dd></dl>
204
205<dl class="method">
206<dt id="park.assembly.Assembly.all_results">
207<tt class="descname">all_results</tt><big>(</big><em>result</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.all_results"><span class="viewcode-link">[source]</span></a></dt>
208<dd><p>Extend result from the fit with the calculated parameters.</p>
209</dd></dl>
210
211<dl class="method">
212<dt id="park.assembly.Assembly.append">
213<tt class="descname">append</tt><big>(</big><em>fitness</em>, <em>weight=1.0</em>, <em>isfitted=True</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.append"><span class="viewcode-link">[source]</span></a></dt>
214<dd><p>Add a model to the end of set.</p>
215<table class="docutils field-list" frame="void" rules="none">
216<col class="field-name" />
217<col class="field-body" />
218<tbody valign="top">
219<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple">
220<li><strong>fitness</strong> &#8211; the fitting model
221The fitting model can be an instance of <cite>park.assembly.Fitness</cite>,
222or a tuple of (<cite>park.model.Model</cite>,`park.data.Data1D`)</li>
223<li><strong>weight</strong> &#8211; model weighting (usually 1.0)</li>
224<li><strong>isfitted</strong> &#8211; whether model should be fit (equivalent to weight 0.)</li>
225</ul>
226</td>
227</tr>
228</tbody>
229</table>
230</dd></dl>
231
232<dl class="method">
233<dt id="park.assembly.Assembly.cov">
234<tt class="descname">cov</tt><big>(</big><em>pvec</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.cov"><span class="viewcode-link">[source]</span></a></dt>
235<dd><p>Return the covariance matrix inv(J&#8217;J) at point p.</p>
236</dd></dl>
237
238<dl class="method">
239<dt id="park.assembly.Assembly.eval">
240<tt class="descname">eval</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.eval"><span class="viewcode-link">[source]</span></a></dt>
241<dd><p>Recalculate the theory functions, and from them, the
242residuals and chisq.</p>
243<table class="docutils field-list" frame="void" rules="none">
244<col class="field-name" />
245<col class="field-body" />
246<tbody valign="top">
247<tr class="field-odd field"><th class="field-name">Note:</th><td class="field-body">Call this after the parameters have been updated.</td>
248</tr>
249</tbody>
250</table>
251</dd></dl>
252
253<dl class="method">
254<dt id="park.assembly.Assembly.fit_parameters">
255<tt class="descname">fit_parameters</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.fit_parameters"><span class="viewcode-link">[source]</span></a></dt>
256<dd><p>Return an alphabetical list of the fitting parameters.</p>
257<p>This function is called once at the beginning of a fit,
258and serves as a convenient place to precalculate what
259can be precalculated such as the set of fitting parameters
260and the parameter expressions evaluator.</p>
261</dd></dl>
262
263<dl class="method">
264<dt id="park.assembly.Assembly.fresiduals">
265<tt class="descname">fresiduals</tt><big>(</big><em>pvec</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.fresiduals"><span class="viewcode-link">[source]</span></a></dt>
266<dd></dd></dl>
267
268<dl class="method">
269<dt id="park.assembly.Assembly.insert">
270<tt class="descname">insert</tt><big>(</big><em>idx</em>, <em>fitness</em>, <em>weight=1.0</em>, <em>isfitted=True</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.insert"><span class="viewcode-link">[source]</span></a></dt>
271<dd><p>Add a model to a particular position in the set.</p>
272</dd></dl>
273
274<dl class="method">
275<dt id="park.assembly.Assembly.isfeasible">
276<tt class="descname">isfeasible</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.isfeasible"><span class="viewcode-link">[source]</span></a></dt>
277<dd><p>Returns true if the parameter set is in a feasible region of the
278modeling space.</p>
279</dd></dl>
280
281<dl class="method">
282<dt id="park.assembly.Assembly.isfitted">
283<tt class="descname">isfitted</tt><big>(</big><em>idx</em>, <em>value=None</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.isfitted"><span class="viewcode-link">[source]</span></a></dt>
284<dd><p>Query if a particular model is fitted.</p>
285<p>Set isfitted to value if value is supplied.</p>
286<table class="docutils field-list" frame="void" rules="none">
287<col class="field-name" />
288<col class="field-body" />
289<tbody valign="top">
290<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple">
291<li><strong>idx</strong> (<em>integer</em>) &#8211; model number</li>
292<li><strong>value</strong> &#8211; </li>
293</ul>
294</td>
295</tr>
296</tbody>
297</table>
298</dd></dl>
299
300<dl class="method">
301<dt id="park.assembly.Assembly.jacobian">
302<tt class="descname">jacobian</tt><big>(</big><em>pvec</em>, <em>step=1e-08</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.jacobian"><span class="viewcode-link">[source]</span></a></dt>
303<dd><p>Returns the derivative wrt the fit parameters at point p.</p>
304<p>Numeric derivatives are calculated based on step, where step is
305the portion of the total range for parameter j, or the portion of
306point value p_j if the range on parameter j is infinite.</p>
307</dd></dl>
308
309<dl class="method">
310<dt id="park.assembly.Assembly.result">
311<tt class="descname">result</tt><big>(</big><em>status='step'</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.result"><span class="viewcode-link">[source]</span></a></dt>
312<dd><p>Details to send back to the fitting client on an improved fit.</p>
313<p>status is &#8216;start&#8217;, &#8216;step&#8217; or &#8216;end&#8217; depending if this is the
314first result to return, an improved result, or the final result.</p>
315<p>[Not implemented]</p>
316</dd></dl>
317
318<dl class="method">
319<dt id="park.assembly.Assembly.set_result">
320<tt class="descname">set_result</tt><big>(</big><em>result</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.set_result"><span class="viewcode-link">[source]</span></a></dt>
321<dd><p>Set the parameters resulting from the fit into the parameter set,
322and update the calculated expression.</p>
323<p>The parameter values may be retrieved by walking the assembly.parameterset
324tree, checking each parameter for isfitted, iscomputed, or isfixed.
325For example:</p>
326<div class="highlight-python"><div class="highlight"><pre><span class="n">assembly</span><span class="o">.</span><span class="n">set_result</span><span class="p">(</span><span class="n">result</span><span class="p">)</span>
327<span class="k">for</span> <span class="n">p</span> <span class="ow">in</span> <span class="n">assembly</span><span class="o">.</span><span class="n">parameterset</span><span class="o">.</span><span class="n">flatten</span><span class="p">():</span>
328    <span class="k">if</span> <span class="n">p</span><span class="o">.</span><span class="n">isfitted</span><span class="p">():</span>
329        <span class="k">print</span> <span class="s">&quot;</span><span class="si">%s</span><span class="s"> </span><span class="si">%g</span><span class="s"> in [</span><span class="si">%g</span><span class="s">,</span><span class="si">%g</span><span class="s">]&quot;</span><span class="o">%</span><span class="p">(</span><span class="n">p</span><span class="o">.</span><span class="n">path</span><span class="p">,</span><span class="n">p</span><span class="o">.</span><span class="n">value</span><span class="p">,</span><span class="n">p</span><span class="o">.</span><span class="n">range</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span><span class="n">p</span><span class="o">.</span><span class="n">range</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span>
330    <span class="k">elif</span> <span class="n">p</span><span class="o">.</span><span class="n">iscomputed</span><span class="p">():</span>
331        <span class="k">print</span> <span class="s">&quot;</span><span class="si">%s</span><span class="s"> computed as </span><span class="si">%g</span><span class="s">&quot;</span><span class="o">%</span><span class="p">(</span><span class="n">p</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">p</span><span class="o">.</span><span class="n">value</span><span class="p">)</span>
332</pre></div>
333</div>
334<p>This does not calculate the function or the residuals for these parameters.
335You can call assembly.eval() to do this.  The residuals will be set in
336assembly[i].residuals.  The theory and data are model specific, and can
337be found in assembly[i].fitness.data.</p>
338</dd></dl>
339
340<dl class="method">
341<dt id="park.assembly.Assembly.stderr">
342<tt class="descname">stderr</tt><big>(</big><em>pvec</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.stderr"><span class="viewcode-link">[source]</span></a></dt>
343<dd><p>Return parameter uncertainty.</p>
344<p>This is just the sqrt diagonal of covariance matrix inv(J&#8217;J) at point p.</p>
345</dd></dl>
346
347<dl class="method">
348<dt id="park.assembly.Assembly.weight">
349<tt class="descname">weight</tt><big>(</big><em>idx</em>, <em>value=None</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Assembly.weight"><span class="viewcode-link">[source]</span></a></dt>
350<dd><p>Query the weight on a particular model.</p>
351<p>Set weight to value if value is supplied.</p>
352<table class="docutils field-list" frame="void" rules="none">
353<col class="field-name" />
354<col class="field-body" />
355<tbody valign="top">
356<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><dl class="first docutils">
357<dt>idx <span class="classifier-delimiter">:</span> <span class="classifier">integer</span></dt>
358<dd><p class="first last">model number</p>
359</dd>
360<dt>value <span class="classifier-delimiter">:</span> <span class="classifier">float</span></dt>
361<dd><p class="first last">model weight</p>
362</dd>
363</dl>
364</td>
365</tr>
366<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first last">model weight</p>
367</td>
368</tr>
369</tbody>
370</table>
371</dd></dl>
372
373</dd></dl>
374
375<dl class="class">
376<dt id="park.assembly.Fitness">
377<em class="property">class </em><tt class="descclassname">park.assembly.</tt><tt class="descname">Fitness</tt><big>(</big><em>model=None</em>, <em>data=None</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Fitness"><span class="viewcode-link">[source]</span></a></dt>
378<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
379<p>Container for theory and data.</p>
380<p>The fit object compares theory with data.</p>
381<p>TODO: what to do with fittable metadata (e.g., footprint correction)?</p>
382<dl class="method">
383<dt id="park.assembly.Fitness.abort">
384<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Fitness.abort"><span class="viewcode-link">[source]</span></a></dt>
385<dd></dd></dl>
386
387<dl class="attribute">
388<dt id="park.assembly.Fitness.data">
389<tt class="descname">data</tt><em class="property"> = None</em></dt>
390<dd></dd></dl>
391
392<dl class="attribute">
393<dt id="park.assembly.Fitness.model">
394<tt class="descname">model</tt><em class="property"> = None</em></dt>
395<dd></dd></dl>
396
397<dl class="attribute">
398<dt id="park.assembly.Fitness.parameterset">
399<tt class="descname">parameterset</tt></dt>
400<dd></dd></dl>
401
402<dl class="method">
403<dt id="park.assembly.Fitness.residuals">
404<tt class="descname">residuals</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Fitness.residuals"><span class="viewcode-link">[source]</span></a></dt>
405<dd></dd></dl>
406
407<dl class="method">
408<dt id="park.assembly.Fitness.residuals_deriv">
409<tt class="descname">residuals_deriv</tt><big>(</big><em>pars=[]</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Fitness.residuals_deriv"><span class="viewcode-link">[source]</span></a></dt>
410<dd></dd></dl>
411
412<dl class="method">
413<dt id="park.assembly.Fitness.set">
414<tt class="descname">set</tt><big>(</big><em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/assembly.html#Fitness.set"><span class="viewcode-link">[source]</span></a></dt>
415<dd><p>Set parameters in the model.</p>
416<p>User convenience function.  This allows a user with an assembly
417of models in a script to for example set the fit range for
418parameter &#8216;a&#8217; of the model:</p>
419<div class="highlight-python"><div class="highlight"><pre><span class="n">assembly</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="n">a</span><span class="o">=</span><span class="p">[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">])</span>
420</pre></div>
421</div>
422<p>Raises KeyError if the parameter is not in parameterset.</p>
423</dd></dl>
424
425</dd></dl>
426
427</div>
428</div>
429</div>
430<div class="section" id="module-park.data">
431<span id="park-data-module"></span><h2>park.data module</h2>
432<p>Park 1-D data objects.</p>
433<p>The class Data1D represents simple 1-D data objects, along with
434an ascii file loader.  This format will work well for many
435uses, but it is likely that more specialized models will have
436their own data file formats.</p>
437<p>The minimal data format for park must supply the following
438methods:</p>
439<blockquote>
440<div><dl class="docutils">
441<dt>residuals(fn)</dt>
442<dd>returns the residuals vector for the model function.</dd>
443<dt>residuals_deriv(fn_deriv,par)</dt>
444<dd>returns the residuals vector for the model function, and
445for the derivatives with respect to the given parameters.</dd>
446</dl>
447</div></blockquote>
448<p>The function passed is going to be model.eval or in the case
449where derivatives are available, model.eval_deriv.  Normally
450this will take a vector of dependent variables and return the
451theory function for that vector but this is only convention.
452The fitting service only uses the parameter set and the residuals
453method from the model.</p>
454<p>The park GUI will make more demands on the interface, but the
455details are not yet resolved.</p>
456<dl class="class">
457<dt id="park.data.Data1D">
458<em class="property">class </em><tt class="descclassname">park.data.</tt><tt class="descname">Data1D</tt><big>(</big><em>filename=''</em>, <em>x=None</em>, <em>y=None</em>, <em>dx=0</em>, <em>dy=1</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D"><span class="viewcode-link">[source]</span></a></dt>
459<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
460<p>Data representation for 1-D fitting.</p>
461<p>Attributes</p>
462<dl class="docutils">
463<dt>filename</dt>
464<dd>The source of the data.  This may be the empty string if the
465data is simulation data.</dd>
466<dt>x,y,dy</dt>
467<dd>The data values.
468x is the measurement points of data to be fitted. x must be sorted.
469y is the measured value
470dy is the measurement uncertainty.</dd>
471<dt>dx</dt>
472<dd>Resolution at the the measured points.  The resolution may be 0,
473constant, or defined for each data point.  dx is the 1-sigma
474width of the Gaussian resolution function at point x.  Note that
475dx_FWHM = sqrt(8 ln 2) dx_sigma, so scale dx appropriately.</dd>
476<dt>fit_x,fit_dx,fit_y,fit_dy</dt>
477<dd>The points used in evaluating the residuals.</dd>
478<dt>calc_x</dt>
479<dd>The points at which to evaluate the theory function.  This may be
480different from the measured points for a number of reasons, such
481as a resolution function which suggests over or under sampling of
482the points (see below).  By default calc_x is x, but it can be set
483explicitly by the user.</dd>
484<dt>calc_y, fx</dt>
485<dd>The value of the function at the theory points, and the value of
486the function after resolution has been applied.  These values are
487computed by a call to residuals.</dd>
488</dl>
489<p>Notes on calc_x</p>
490<p>The contribution of Q to a resolution of width dQo at point Qo is:</p>
491<div class="highlight-python"><div class="highlight"><pre>p(Q) = 1/sqrt(2 pi dQo**2) exp ( (Q-Qo)**2/(2 dQo**2) )
492</pre></div>
493</div>
494<p>We are approximating the convolution at Qo using a numerical
495approximation to the integral over the measured points, with the
496integral is limited to p(Q_i)/p(0)&gt;=0.001.</p>
497<p>Sometimes the function we are convoluting is rapidly changing.
498That means the correct convolution should uniformly sample across
499the entire width of the Gaussian.  This is not possible at the
500end points unless you calculate the theory function beyond what is
501strictly needed for the data. For a given dQ and step size,
502you need enough steps that the following is true:</p>
503<div class="highlight-python"><div class="highlight"><pre>(n*step)**2 &gt; -2 dQ**2 * ln 0.001
504</pre></div>
505</div>
506<p>The choice of sampling density is particularly important near
507critical points where the shape of the function changes.  In
508reflectometry, the function goes from flat below the critical
509edge to O(Q**4) above.  In one particular model, calculating every
5100.005 rather than every 0.02 changed a value above the critical
511edge by 15%.  In a fitting program, this would lead to a somewhat
512larger estimate of the critical edge for this sample.</p>
513<p>Sometimes the theory function is oscillating more rapidly than
514the instrument can resolve.  This happens for example in reflectivity
515measurements involving thick layers.  In these systems, the theory
516function should be oversampled around the measured points Q.  With
517a single thick layer, oversampling can be limited to just one
518period 2 pi/d.  With multiple thick layers, oscillations will
519show interference patterns and it will be necessary to oversample
520uniformly through the entire width of the resolution.  If this is
521not accommodated, then aliasing effects make it difficult to
522compute the correct model.</p>
523<dl class="attribute">
524<dt id="park.data.Data1D.calc_x">
525<tt class="descname">calc_x</tt><em class="property"> = None</em></dt>
526<dd></dd></dl>
527
528<dl class="attribute">
529<dt id="park.data.Data1D.calc_y">
530<tt class="descname">calc_y</tt><em class="property"> = None</em></dt>
531<dd></dd></dl>
532
533<dl class="attribute">
534<dt id="park.data.Data1D.dx">
535<tt class="descname">dx</tt><em class="property"> = 0</em></dt>
536<dd></dd></dl>
537
538<dl class="attribute">
539<dt id="park.data.Data1D.dy">
540<tt class="descname">dy</tt><em class="property"> = 1</em></dt>
541<dd></dd></dl>
542
543<dl class="attribute">
544<dt id="park.data.Data1D.filename">
545<tt class="descname">filename</tt><em class="property"> = ''</em></dt>
546<dd></dd></dl>
547
548<dl class="attribute">
549<dt id="park.data.Data1D.fit_dx">
550<tt class="descname">fit_dx</tt><em class="property"> = 0</em></dt>
551<dd></dd></dl>
552
553<dl class="attribute">
554<dt id="park.data.Data1D.fit_dy">
555<tt class="descname">fit_dy</tt><em class="property"> = 1</em></dt>
556<dd></dd></dl>
557
558<dl class="attribute">
559<dt id="park.data.Data1D.fit_x">
560<tt class="descname">fit_x</tt><em class="property"> = None</em></dt>
561<dd></dd></dl>
562
563<dl class="attribute">
564<dt id="park.data.Data1D.fit_y">
565<tt class="descname">fit_y</tt><em class="property"> = None</em></dt>
566<dd></dd></dl>
567
568<dl class="attribute">
569<dt id="park.data.Data1D.fx">
570<tt class="descname">fx</tt><em class="property"> = None</em></dt>
571<dd></dd></dl>
572
573<dl class="method">
574<dt id="park.data.Data1D.load">
575<tt class="descname">load</tt><big>(</big><em>filename</em>, <em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D.load"><span class="viewcode-link">[source]</span></a></dt>
576<dd><p>Load a multicolumn datafile.</p>
577<p>Data should be in columns, with the following defaults:</p>
578<div class="highlight-python"><div class="highlight"><pre><span class="n">x</span><span class="p">,</span><span class="n">y</span> <span class="ow">or</span> <span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">dy</span> <span class="ow">or</span> <span class="n">x</span><span class="p">,</span><span class="n">dx</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">dy</span>
579</pre></div>
580</div>
581<p>Note that this resets the selected fitting points calc_x and the
582computed results calc_y and fx.</p>
583<p>Data is sorted after loading.</p>
584<p>Any extra keyword arguments are passed to the numpy loadtxt
585function.  This allows you to select the columns you want,
586skip rows, set the column separator, change the comment character,
587amongst other things.</p>
588</dd></dl>
589
590<dl class="method">
591<dt id="park.data.Data1D.resample">
592<tt class="descname">resample</tt><big>(</big><em>minstep=None</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D.resample"><span class="viewcode-link">[source]</span></a></dt>
593<dd><p>Over/under sampling support.</p>
594<p>Compute the calc_x points required to adequately sample
595the function y=f(x) so that the value reported for each
596measured point is supported by the resolution.  minstep
597is the minimum allowed sampling density that should be
598used.</p>
599</dd></dl>
600
601<dl class="method">
602<dt id="park.data.Data1D.residuals">
603<tt class="descname">residuals</tt><big>(</big><em>fn</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D.residuals"><span class="viewcode-link">[source]</span></a></dt>
604<dd><p>Compute the residuals of the data wrt to the given function.</p>
605<p>y = fn(x) should be a callable accepting a list of points at which
606to calculate the function, returning the values at those
607points.</p>
608<p>Any resolution function will be applied after the theory points
609are calculated.</p>
610</dd></dl>
611
612<dl class="method">
613<dt id="park.data.Data1D.residuals_deriv">
614<tt class="descname">residuals_deriv</tt><big>(</big><em>fn</em>, <em>pars=[]</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D.residuals_deriv"><span class="viewcode-link">[source]</span></a></dt>
615<dd><p>Compute residuals and derivatives wrt the given parameters.</p>
616<p>fdf = fn(x,pars=pars) should be a callable accepting a list
617of points at which to calculate the function and a keyword
618argument listing the parameters for which the derivative will
619be calculated.</p>
620<p>Returns a list of the residuals and the derivative wrt the
621parameter for each parameter.</p>
622<p>Any resolution function will be applied after the theory points
623and derivatives are calculated.</p>
624</dd></dl>
625
626<dl class="method">
627<dt id="park.data.Data1D.select">
628<tt class="descname">select</tt><big>(</big><em>idx</em><big>)</big><a class="reference internal" href="../../_modules/park/data.html#Data1D.select"><span class="viewcode-link">[source]</span></a></dt>
629<dd><p>A selection vector for points to use in the evaluation of the
630residuals, or None if all points are to be used.</p>
631</dd></dl>
632
633<dl class="attribute">
634<dt id="park.data.Data1D.x">
635<tt class="descname">x</tt><em class="property"> = None</em></dt>
636<dd></dd></dl>
637
638<dl class="attribute">
639<dt id="park.data.Data1D.y">
640<tt class="descname">y</tt><em class="property"> = None</em></dt>
641<dd></dd></dl>
642
643</dd></dl>
644
645</div>
646<div class="section" id="module-park.deps">
647<span id="park-deps-module"></span><h2>park.deps module</h2>
648<p>Functions for manipulating dependencies.</p>
649<p>Parameter values must be updated in the correct order.  If parameter A
650depends on parameter B, then parameter B must be evaluated first.</p>
651<dl class="function">
652<dt id="park.deps.order_dependencies">
653<tt class="descclassname">park.deps.</tt><tt class="descname">order_dependencies</tt><big>(</big><em>pairs</em><big>)</big><a class="reference internal" href="../../_modules/park/deps.html#order_dependencies"><span class="viewcode-link">[source]</span></a></dt>
654<dd><p>Order elements from pairs so that b comes before a in the
655ordered list for all pairs (a,b).</p>
656</dd></dl>
657
658<dl class="function">
659<dt id="park.deps.test">
660<tt class="descclassname">park.deps.</tt><tt class="descname">test</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/deps.html#test"><span class="viewcode-link">[source]</span></a></dt>
661<dd></dd></dl>
662
663</div>
664<div class="section" id="module-park.diffev">
665<span id="park-diffev-module"></span><h2>park.diffev module</h2>
666<dl class="class">
667<dt id="park.diffev.DE">
668<em class="property">class </em><tt class="descclassname">park.diffev.</tt><tt class="descname">DE</tt><big>(</big><em>F</em>, <em>crossover_rate</em>, <em>func</em>, <em>func_dim</em>, <em>pop</em>, <em>l_bound</em>, <em>u_bound</em>, <em>tol=1e-07</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE"><span class="viewcode-link">[source]</span></a></dt>
669<dd><p>Differential evolution test implementation
670Implements the Scheme_DE_rand_1 variant</p>
671<blockquote>
672<div><dl class="docutils">
673<dt>F: float</dt>
674<dd>weighting factor which determines the magnitude of perturbation
675occurring in each generation.</dd>
676<dt>crossover_rate:  float</dt>
677<dd>In general, 0 &lt; crossover_rate &lt; 1.  Usually
678considerably lower than 1.  Convergence slows as the value
679increases, but higher crossover rates may be necessary when</dd>
680<dt>func: w = f(p)</dt>
681<dd>The function to be minimized, of the form w = f(p), where p
682is a vector of length func_dim and w is a scalar</dd>
683<dt>func_dim: int</dt>
684<dd>The dimension of the objective function</dd>
685<dt>pop: array</dt>
686<dd>The initial population.  This should be an iterable composed of
687Rank-1 numpy arrays.  The population size must be at least 4,
688preferably much greater.</dd>
689<dt>l_bound, u_bound: vector</dt>
690<dd>arrays of size func_dim representing the upper and lower bounds
691for the parameters in the solution vectors</dd>
692<dt>tol: float</dt>
693<dd>if (max(pop_values) - min(pop_values) &lt;= conv), the population
694has converged and the evolution will stop</dd>
695</dl>
696</div></blockquote>
697<dl class="method">
698<dt id="park.diffev.DE.adjust_F">
699<tt class="descname">adjust_F</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.adjust_F"><span class="viewcode-link">[source]</span></a></dt>
700<dd><p>Adjust F to account for stagnation of the population</p>
701</dd></dl>
702
703<dl class="method">
704<dt id="park.diffev.DE.crossover">
705<tt class="descname">crossover</tt><big>(</big><em>candidate</em>, <em>i</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.crossover"><span class="viewcode-link">[source]</span></a></dt>
706<dd><p>Perform a crossover between the candidate and the ith member of
707the previous generation.</p>
708</dd></dl>
709
710<dl class="method">
711<dt id="park.diffev.DE.evolve">
712<tt class="descname">evolve</tt><big>(</big><em>numgens=1000</em>, <em>monitor=None</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.evolve"><span class="viewcode-link">[source]</span></a></dt>
713<dd><p>Evolve the population for numgens generations, or until it converges.
714Returns the best vector from the run</p>
715</dd></dl>
716
717<dl class="method">
718<dt id="park.diffev.DE.get_best_gene">
719<tt class="descname">get_best_gene</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.get_best_gene"><span class="viewcode-link">[source]</span></a></dt>
720<dd><p>find the most fit gene in the current population</p>
721</dd></dl>
722
723<dl class="method">
724<dt id="park.diffev.DE.get_random_gene">
725<tt class="descname">get_random_gene</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.get_random_gene"><span class="viewcode-link">[source]</span></a></dt>
726<dd><p>Generate a random gene within the bounds constraints.
727Used to replace out-of-bounds genes</p>
728</dd></dl>
729
730<dl class="method">
731<dt id="park.diffev.DE.is_converged">
732<tt class="descname">is_converged</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.is_converged"><span class="viewcode-link">[source]</span></a></dt>
733<dd><p>check for convergence</p>
734</dd></dl>
735
736<dl class="method">
737<dt id="park.diffev.DE.is_within_bounds">
738<tt class="descname">is_within_bounds</tt><big>(</big><em>gene</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.is_within_bounds"><span class="viewcode-link">[source]</span></a></dt>
739<dd><p>Determine whether the gene meets the bounds constraints</p>
740</dd></dl>
741
742<dl class="method">
743<dt id="park.diffev.DE.select_participants">
744<tt class="descname">select_participants</tt><big>(</big><em>i</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#DE.select_participants"><span class="viewcode-link">[source]</span></a></dt>
745<dd><p>generate r1, r2, and r3 randomly from the range [0, NP-1]
746such that they are distinct values not equal to i</p>
747</dd></dl>
748
749</dd></dl>
750
751<dl class="function">
752<dt id="park.diffev.diffev">
753<tt class="descclassname">park.diffev.</tt><tt class="descname">diffev</tt><big>(</big><em>fn</em>, <em>bounds</em>, <em>x0=None</em>, <em>pop_scale=4</em>, <em>crossover_rate=0.8</em>, <em>Fscale=1</em>, <em>tolerance=1e-05</em>, <em>maxiter=1000</em>, <em>monitor=None</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#diffev"><span class="viewcode-link">[source]</span></a></dt>
754<dd><p>Run differential evolution, returning x,fx,ncalls</p>
755</dd></dl>
756
757<dl class="function">
758<dt id="park.diffev.gen_pop">
759<tt class="descclassname">park.diffev.</tt><tt class="descname">gen_pop</tt><big>(</big><em>size</em>, <em>l_bound</em>, <em>u_bound</em>, <em>dimension</em><big>)</big><a class="reference internal" href="../../_modules/park/diffev.html#gen_pop"><span class="viewcode-link">[source]</span></a></dt>
760<dd><p>generate a random population of vectors within the given bounds.  dimension
761indicates the length of the vectors.  l_bound and u_bound should be vectors
762of length dimension (any iterable container should work)</p>
763</dd></dl>
764
765<dl class="function">
766<dt id="park.diffev.random">
767<tt class="descclassname">park.diffev.</tt><tt class="descname">random</tt><big>(</big><big>)</big> &rarr; x in the interval [0, 1).</dt>
768<dd></dd></dl>
769
770</div>
771<div class="section" id="module-park.expression">
772<span id="park-expression-module"></span><h2>park.expression module</h2>
773<p>Functions for manipulating expressions.</p>
774<dl class="function">
775<dt id="park.expression.build_eval">
776<tt class="descclassname">park.expression.</tt><tt class="descname">build_eval</tt><big>(</big><em>pars</em>, <em>context={}</em><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#build_eval"><span class="viewcode-link">[source]</span></a></dt>
777<dd><p>Build and return a function to evaluate all parameter expressions in
778the proper order.</p>
779<dl class="docutils">
780<dt>Inputs:</dt>
781<dd>pars is a list of parameters
782context is a dictionary of additional symbols for the expressions</dd>
783<dt>Output:</dt>
784<dd>updater function</dd>
785<dt>Raises:</dt>
786<dd>AssertionError - model, parameter or function is missing
787SyntaxError - improper expression syntax
788ValueError - expressions have circular dependencies</dd>
789</dl>
790<p>This function is not terribly sophisticated, and it would be easy to
791trick.  However it handles the common cases cleanly and generates
792reasonable messages for the common errors.</p>
793<p>This code has not been fully audited for security.  While we have
794removed the builtins and the ability to import modules, there may
795be other vectors for users to perform more than simple function
796evaluations.  Unauthenticated users should not be running this code.</p>
797<p>Parameter names are assumed to contain only _.a-zA-Z0-9#[]</p>
798<p>The list of parameters is probably something like:</p>
799<div class="highlight-python"><div class="highlight"><pre><span class="n">parset</span><span class="o">.</span><span class="n">setprefix</span><span class="p">()</span>
800<span class="n">pars</span> <span class="o">=</span> <span class="n">parset</span><span class="o">.</span><span class="n">flatten</span><span class="p">()</span>
801</pre></div>
802</div>
803<p>Note that math uses acos while numpy uses arccos.  To avoid confusion
804we allow both.</p>
805<p>Should try running the function to identify syntax errors before
806running it in a fit.</p>
807<p>Use help(fn) to see the code generated for the returned function fn.
808dis.dis(fn) will show the corresponding python vm instructions.</p>
809</dd></dl>
810
811<dl class="function">
812<dt id="park.expression.find_dependencies">
813<tt class="descclassname">park.expression.</tt><tt class="descname">find_dependencies</tt><big>(</big><em>pars</em><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#find_dependencies"><span class="viewcode-link">[source]</span></a></dt>
814<dd><p>Returns a list of pair-wise dependencies from the parameter expressions.</p>
815<p>For example, if p3 = p1+p2, then find_dependencies([p1,p2,p3]) will
816return [(p3,p1),(p3,p2)].  For base expressions without dependencies,
817such as p4 = 2*pi, this should return [(p4, None)]</p>
818</dd></dl>
819
820<dl class="function">
821<dt id="park.expression.no_constraints">
822<tt class="descclassname">park.expression.</tt><tt class="descname">no_constraints</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#no_constraints"><span class="viewcode-link">[source]</span></a></dt>
823<dd><p>This parameter set has no constraints between the parameters.</p>
824</dd></dl>
825
826<dl class="function">
827<dt id="park.expression.parameter_mapping">
828<tt class="descclassname">park.expression.</tt><tt class="descname">parameter_mapping</tt><big>(</big><em>pairs</em><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#parameter_mapping"><span class="viewcode-link">[source]</span></a></dt>
829<dd><p>Find the parameter substitution we need so that expressions can
830be evaluated without having to traverse a chain of
831model.layer.parameter.value</p>
832</dd></dl>
833
834<dl class="function">
835<dt id="park.expression.substitute">
836<tt class="descclassname">park.expression.</tt><tt class="descname">substitute</tt><big>(</big><em>expr</em>, <em>mapping</em><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#substitute"><span class="viewcode-link">[source]</span></a></dt>
837<dd><p>Replace all occurrences of symbol s with mapping[s] for s in mapping.</p>
838</dd></dl>
839
840<dl class="function">
841<dt id="park.expression.symbols">
842<tt class="descclassname">park.expression.</tt><tt class="descname">symbols</tt><big>(</big><em>expr</em>, <em>symtab</em><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#symbols"><span class="viewcode-link">[source]</span></a></dt>
843<dd><p>Given an expression string and a symbol table, return the set of symbols
844used in the expression.  Symbols are only returned once even if they
845occur multiple times.  The return value is a set with the elements in
846no particular order.</p>
847<p>This is the first step in computing a dependency graph.</p>
848</dd></dl>
849
850<dl class="function">
851<dt id="park.expression.test">
852<tt class="descclassname">park.expression.</tt><tt class="descname">test</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/expression.html#test"><span class="viewcode-link">[source]</span></a></dt>
853<dd></dd></dl>
854
855</div>
856<div class="section" id="module-park.fit">
857<span id="park-fit-module"></span><h2>park.fit module</h2>
858<p>Fitting service interface.</p>
859<p>A fit consists of a set of models and a fitting engine.  The models are
860collected in an assembly, which manages the parameter set and the
861constraints between them.  The models themselves are tightly coupled
862to the data that they are modeling and the data is invisible to the fit.</p>
863<p>The fitting engine can use a variety of methods depending on model.</p>
864<div class="section" id="id1">
865<h3>Usage</h3>
866<p>The fitter can be run directly on the local machine:</p>
867<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">park</span>
868<span class="n">M1</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">models</span><span class="o">.</span><span class="n">Peaks</span><span class="p">(</span><span class="n">datafile</span><span class="o">=</span><span class="n">park</span><span class="o">.</span><span class="n">sampledata</span><span class="p">(</span><span class="s">&#39;peak.dat&#39;</span><span class="p">))</span>
869<span class="n">M1</span><span class="o">.</span><span class="n">add_peak</span><span class="p">(</span><span class="s">&#39;P1&#39;</span><span class="p">,</span> <span class="s">&#39;gaussian&#39;</span><span class="p">,</span> <span class="n">A</span><span class="o">=</span><span class="p">[</span><span class="mi">4</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> <span class="n">mu</span><span class="o">=</span><span class="p">[</span><span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">],</span> <span class="n">sigma</span><span class="o">=</span><span class="mf">0.1</span><span class="p">)</span>
870<span class="n">result</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">models</span><span class="o">=</span><span class="p">[</span><span class="n">M1</span><span class="p">])</span>
871<span class="k">print</span> <span class="n">result</span>
872</pre></div>
873</div>
874<p>The default settings print results every time the fit improves, and
875print a global result when the fit is complete.  This is a suitable
876interface for a fitting script.</p>
877<p>For larger fit jobs you will want to run the fit on a remote server.
878The model setup is identical, but the fit call is different:</p>
879<div class="highlight-python"><div class="highlight"><pre><span class="n">service</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">FitService</span><span class="p">(</span><span class="s">&#39;server:port&#39;</span><span class="p">)</span>
880<span class="n">result</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">models</span><span class="o">=</span><span class="p">[</span><span class="n">M1</span><span class="p">],</span> <span class="n">service</span><span class="o">=</span><span class="n">service</span><span class="p">)</span>
881<span class="k">print</span> <span class="n">result</span>
882</pre></div>
883</div>
884<p>Again, the default settings print results every time the fit improves,
885and print a global result when the fit is complete.</p>
886<p>For long running fit jobs, you want to be able to disconnect from
887the server after submitting the job, and later reconnect to fetch
888the results.  An additional email field will send notification by
889email when the fit starts and ends, and daily updates on the status
890of all fits:</p>
891<div class="highlight-python"><div class="highlight"><pre><span class="n">service</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">FitService</span><span class="p">(</span><span class="s">&#39;server:port&#39;</span><span class="p">)</span>
892<span class="n">service</span><span class="o">.</span><span class="n">notify</span><span class="p">(</span><span class="n">email</span><span class="o">=</span><span class="s">&#39;me@my.email.address&#39;</span><span class="p">,</span><span class="n">update</span><span class="o">=</span><span class="s">&#39;daily&#39;</span><span class="p">)</span>
893<span class="n">fit</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">Fit</span><span class="p">(</span><span class="n">models</span><span class="o">=</span><span class="p">[</span><span class="n">M1</span><span class="p">])</span>
894<span class="nb">id</span> <span class="o">=</span> <span class="n">service</span><span class="o">.</span><span class="n">submit_job</span><span class="p">(</span><span class="n">fit</span><span class="p">,</span> <span class="n">jobname</span><span class="o">=</span><span class="s">&#39;peaks&#39;</span><span class="p">)</span>
895<span class="k">print</span> <span class="nb">id</span>
896</pre></div>
897</div>
898<p>The results can be retrieved either by id returned from the server,
899or by the given jobname:</p>
900<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">park</span>
901<span class="n">service</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">FitService</span><span class="p">(</span><span class="s">&#39;server:port&#39;</span><span class="p">,</span><span class="n">user</span><span class="o">=</span><span class="s">&#39;userid&#39;</span><span class="p">)</span>
902<span class="n">fitlist</span> <span class="o">=</span> <span class="n">service</span><span class="o">.</span><span class="n">retrieve</span><span class="p">(</span><span class="s">&#39;peaks&#39;</span><span class="p">)</span>
903<span class="k">for</span> <span class="n">fit</span> <span class="ow">in</span> <span class="n">fitlist</span><span class="p">:</span>
904    <span class="k">print</span> <span class="n">fit</span><span class="o">.</span><span class="n">summary</span><span class="p">()</span>
905</pre></div>
906</div>
907<p>The fit itself is a complicated object, including the model, the
908optimizer, and the type of uncertainty analysis to perform.</p>
909</div>
910<div class="section" id="gui-usage">
911<h3>GUI Usage</h3>
912<p>When used from a graphical user interface, a different programming
913interface is needed.  In this case, the user may want to watch
914the progress of the fit and perhaps stop it.  Also, as fits can
915take some time to complete, the user would like to be able to
916set up additional fits and run them at the same time, switching
917between them as necessary to monitor progress.</p>
918<dl class="class">
919<dt id="park.fit.FitJob">
920<em class="property">class </em><tt class="descclassname">park.fit.</tt><tt class="descname">FitJob</tt><big>(</big><em>objective=None</em>, <em>fitter=None</em>, <em>handler=None</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#FitJob"><span class="viewcode-link">[source]</span></a></dt>
921<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
922<p>Fit job.</p>
923<p>This implements <cite>park.job.Job</cite>.</p>
924<dl class="method">
925<dt id="park.fit.FitJob.run">
926<tt class="descname">run</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#FitJob.run"><span class="viewcode-link">[source]</span></a></dt>
927<dd></dd></dl>
928
929</dd></dl>
930
931<dl class="class">
932<dt id="park.fit.Fitter">
933<em class="property">class </em><tt class="descclassname">park.fit.</tt><tt class="descname">Fitter</tt><big>(</big><em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Fitter"><span class="viewcode-link">[source]</span></a></dt>
934<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
935<p>Abstract interface for a fitness optimizer.</p>
936<p>A fitter has a single method, fit, which takes an objective
937function (<cite>park.fit.Objective</cite>) and a handler.</p>
938<p>For a concrete instance see <cite>park.fitmc.FitMC</cite>.</p>
939<dl class="method">
940<dt id="park.fit.Fitter.fit">
941<tt class="descname">fit</tt><big>(</big><em>fitness</em>, <em>handler</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Fitter.fit"><span class="viewcode-link">[source]</span></a></dt>
942<dd><p>Global optimizer.</p>
943<p>This function should return immediately</p>
944</dd></dl>
945
946</dd></dl>
947
948<dl class="class">
949<dt id="park.fit.LocalQueue">
950<em class="property">class </em><tt class="descclassname">park.fit.</tt><tt class="descname">LocalQueue</tt><a class="reference internal" href="../../_modules/park/fit.html#LocalQueue"><span class="viewcode-link">[source]</span></a></dt>
951<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
952<p>Simple interface to the local job queue.  Currently supports start and
953wait.  Needs to support stop and status.  Also, needs to be a proper queue,
954and needs to allow jobs to run in separate processes according to priority,
955etc.  All the essentials of the remote queuing system without the remote
956calls.</p>
957<p>Unlike the remote queue, the local queue need not maintain persistence.</p>
958<dl class="attribute">
959<dt id="park.fit.LocalQueue.running">
960<tt class="descname">running</tt><em class="property"> = False</em></dt>
961<dd></dd></dl>
962
963<dl class="method">
964<dt id="park.fit.LocalQueue.start">
965<tt class="descname">start</tt><big>(</big><em>job</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#LocalQueue.start"><span class="viewcode-link">[source]</span></a></dt>
966<dd></dd></dl>
967
968<dl class="method">
969<dt id="park.fit.LocalQueue.wait">
970<tt class="descname">wait</tt><big>(</big><em>interval=0.1</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#LocalQueue.wait"><span class="viewcode-link">[source]</span></a></dt>
971<dd><p>Wait for the job to complete.  This is used in scripts to impose
972a synchronous interface to the fitting service.</p>
973</dd></dl>
974
975</dd></dl>
976
977<dl class="class">
978<dt id="park.fit.Objective">
979<em class="property">class </em><tt class="descclassname">park.fit.</tt><tt class="descname">Objective</tt><a class="reference internal" href="../../_modules/park/fit.html#Objective"><span class="viewcode-link">[source]</span></a></dt>
980<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
981<p>Abstract interface to the fitness function for the park minimizer
982classes.</p>
983<p>Park provides a specific implementation <cite>park.assembly.Assembly</cite>.</p>
984<p>TODO: add a results() method to return model specific info to the
985TODO: fit handler.</p>
986<dl class="method">
987<dt id="park.fit.Objective.abort">
988<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Objective.abort"><span class="viewcode-link">[source]</span></a></dt>
989<dd><p>Halts the current function evaluation, and has it return inf.
990This will be called from a separate thread.  If the function
991contains an expensive calculation, it should reset an abort
992flag before each evaluation and test it periodically.</p>
993<p>This method is optional.</p>
994</dd></dl>
995
996<dl class="method">
997<dt id="park.fit.Objective.fit_parameters">
998<tt class="descname">fit_parameters</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Objective.fit_parameters"><span class="viewcode-link">[source]</span></a></dt>
999<dd><p>Returns a list of fit parameters.  Each parameter has a name,
1000an initial value and a range.</p>
1001<p>See <cite>park.fitresult.FitParameter</cite> for an example.</p>
1002<p>On each function evaluation a new parameter set will be passed
1003to the fitter, with values in the same order as the list of
1004parameters.</p>
1005</dd></dl>
1006
1007<dl class="method">
1008<dt id="park.fit.Objective.residuals">
1009<tt class="descname">residuals</tt><big>(</big><em>p</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Objective.residuals"><span class="viewcode-link">[source]</span></a></dt>
1010<dd><p>Some fitters, notably Levenberg-Marquardt, operate directly on the
1011residuals vector.  If the individual residuals are not available,
1012then LM cannot be used.</p>
1013<p>This method is optional.</p>
1014</dd></dl>
1015
1016<dl class="method">
1017<dt id="park.fit.Objective.residuals_deriv">
1018<tt class="descname">residuals_deriv</tt><big>(</big><em>p</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#Objective.residuals_deriv"><span class="viewcode-link">[source]</span></a></dt>
1019<dd><p>Returns residuals and derivatives with respect to the given
1020parameters.</p>
1021<p>If these are unavailable in the model, then they can be approximated
1022by numerical derivatives, though it is generally better to use a
1023derivative free optimizer such as coliny or cobyla which can use the
1024function evaluations more efficiently.  In any case, your objective
1025function is responsible for calculating these.</p>
1026<p>This method is optional.</p>
1027</dd></dl>
1028
1029</dd></dl>
1030
1031<dl class="function">
1032<dt id="park.fit.assembly_example">
1033<tt class="descclassname">park.fit.</tt><tt class="descname">assembly_example</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#assembly_example"><span class="viewcode-link">[source]</span></a></dt>
1034<dd></dd></dl>
1035
1036<dl class="function">
1037<dt id="park.fit.demo">
1038<tt class="descclassname">park.fit.</tt><tt class="descname">demo</tt><big>(</big><em>fitter=None</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#demo"><span class="viewcode-link">[source]</span></a></dt>
1039<dd><p>Multiple minima example</p>
1040</dd></dl>
1041
1042<dl class="function">
1043<dt id="park.fit.demo2">
1044<tt class="descclassname">park.fit.</tt><tt class="descname">demo2</tt><big>(</big><em>fitter=None</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#demo2"><span class="viewcode-link">[source]</span></a></dt>
1045<dd></dd></dl>
1046
1047<dl class="function">
1048<dt id="park.fit.fit">
1049<tt class="descclassname">park.fit.</tt><tt class="descname">fit</tt><big>(</big><em>models=None</em>, <em>fitter=None</em>, <em>service=None</em>, <em>handler=None</em><big>)</big><a class="reference internal" href="../../_modules/park/fit.html#fit"><span class="viewcode-link">[source]</span></a></dt>
1050<dd><p>Start a fit with a set of models.  The model set must be
1051in a form accepted by <cite>park.assembly.Assembly</cite>.</p>
1052<p>This is a convenience function which sets up the default
1053optimizer and uses the local fitting engine to do the work.
1054Progress reports are printed as they are received.</p>
1055<p>The choice of fitter, service and handler can be specified
1056by the caller.</p>
1057<p>The default fitter is FitMC, which is a monte carlo Nelder-Mead
1058simplex local optimizer with 100 random start points.</p>
1059<p>The default handler does nothing.  Instead, ConsoleUpdate could
1060be used to report progress during the fit.</p>
1061<p>The default service is to run in a separate thread with FitThread.
1062Note that this will change soon to run in a separate process on
1063the local machine so that python&#8217;s global interpreter lock does
1064not interfere with parallelism.</p>
1065</dd></dl>
1066
1067</div>
1068</div>
1069<div class="section" id="park-fitmc-module">
1070<h2>park.fitmc module</h2>
1071<span class="target" id="module-park.fitmc"></span><dl class="function">
1072<dt id="park.fitmc.fitmc">
1073<tt class="descclassname">park.fitmc.</tt><tt class="descname">fitmc</tt><big>(</big><em>fitness</em>, <em>x0</em>, <em>bounds</em>, <em>localfit</em>, <em>n</em>, <em>handler</em><big>)</big><a class="reference internal" href="../../_modules/park/fitmc.html#fitmc"><span class="viewcode-link">[source]</span></a></dt>
1074<dd><p>Run a monte carlo fit.</p>
1075<p>This procedure maps a local optimizer across a set of n initial points.
1076The initial parameter value defined by the fitness parameters defines
1077one initial point.  The remainder are randomly generated within the
1078bounds of the problem.</p>
1079<p>localfit is the local optimizer to use.  It should be a bounded
1080optimizer following the <cite>park.fitmc.LocalFit</cite> interface.</p>
1081<p>handler accepts updates to the current best set of fit parameters.
1082See <cite>park.fitresult.FitHandler</cite> for details.</p>
1083</dd></dl>
1084
1085</div>
1086<div class="section" id="module-park.fitresult">
1087<span id="park-fitresult-module"></span><h2>park.fitresult module</h2>
1088<dl class="class">
1089<dt id="park.fitresult.ConsoleUpdate">
1090<em class="property">class </em><tt class="descclassname">park.fitresult.</tt><tt class="descname">ConsoleUpdate</tt><big>(</big><em>quiet=False</em>, <em>progress_delta=60</em>, <em>improvement_delta=5</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate"><span class="viewcode-link">[source]</span></a></dt>
1091<dd><p>Bases: <a class="reference internal" href="#park.fitresult.FitHandler" title="park.fitresult.FitHandler"><tt class="xref py py-class docutils literal"><span class="pre">park.fitresult.FitHandler</span></tt></a></p>
1092<p>Print progress to the console.</p>
1093<dl class="method">
1094<dt id="park.fitresult.ConsoleUpdate.abort">
1095<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate.abort"><span class="viewcode-link">[source]</span></a></dt>
1096<dd></dd></dl>
1097
1098<dl class="method">
1099<dt id="park.fitresult.ConsoleUpdate.error">
1100<tt class="descname">error</tt><big>(</big><em>msg</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate.error"><span class="viewcode-link">[source]</span></a></dt>
1101<dd><p>Model had an error; print traceback</p>
1102</dd></dl>
1103
1104<dl class="method">
1105<dt id="park.fitresult.ConsoleUpdate.finalize">
1106<tt class="descname">finalize</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate.finalize"><span class="viewcode-link">[source]</span></a></dt>
1107<dd></dd></dl>
1108
1109<dl class="method">
1110<dt id="park.fitresult.ConsoleUpdate.improvement">
1111<tt class="descname">improvement</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate.improvement"><span class="viewcode-link">[source]</span></a></dt>
1112<dd><p>Called when a result is observed which is better than previous
1113results from the fit.</p>
1114</dd></dl>
1115
1116<dl class="attribute">
1117<dt id="park.fitresult.ConsoleUpdate.improvement_delta">
1118<tt class="descname">improvement_delta</tt><em class="property"> = 5</em></dt>
1119<dd><p>Number of seconds between improvement updates</p>
1120</dd></dl>
1121
1122<dl class="attribute">
1123<dt id="park.fitresult.ConsoleUpdate.isbetter">
1124<tt class="descname">isbetter</tt><em class="property"> = False</em></dt>
1125<dd><p>Record whether results improved since last update</p>
1126</dd></dl>
1127
1128<dl class="method">
1129<dt id="park.fitresult.ConsoleUpdate.progress">
1130<tt class="descname">progress</tt><big>(</big><em>k</em>, <em>n</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#ConsoleUpdate.progress"><span class="viewcode-link">[source]</span></a></dt>
1131<dd><p>Report on progress.</p>
1132</dd></dl>
1133
1134<dl class="attribute">
1135<dt id="park.fitresult.ConsoleUpdate.progress_delta">
1136<tt class="descname">progress_delta</tt><em class="property"> = 60</em></dt>
1137<dd><p>Number of seconds between progress updates</p>
1138</dd></dl>
1139
1140</dd></dl>
1141
1142<dl class="class">
1143<dt id="park.fitresult.FitHandler">
1144<em class="property">class </em><tt class="descclassname">park.fitresult.</tt><tt class="descname">FitHandler</tt><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler"><span class="viewcode-link">[source]</span></a></dt>
1145<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1146<p>Abstract interface for fit thread handler.</p>
1147<p>The methods in this class are called by the optimizer as the fit
1148progresses.</p>
1149<p>Note that it is up to the optimizer to call the fit handler correctly,
1150reporting all status changes and maintaining the &#8216;done&#8217; flag.</p>
1151<dl class="method">
1152<dt id="park.fitresult.FitHandler.abort">
1153<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler.abort"><span class="viewcode-link">[source]</span></a></dt>
1154<dd><p>Fit was aborted.</p>
1155</dd></dl>
1156
1157<dl class="attribute">
1158<dt id="park.fitresult.FitHandler.done">
1159<tt class="descname">done</tt><em class="property"> = False</em></dt>
1160<dd><p>True when the fit job is complete</p>
1161</dd></dl>
1162
1163<dl class="method">
1164<dt id="park.fitresult.FitHandler.error">
1165<tt class="descname">error</tt><big>(</big><em>msg</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler.error"><span class="viewcode-link">[source]</span></a></dt>
1166<dd><p>Model had an error; print traceback</p>
1167</dd></dl>
1168
1169<dl class="method">
1170<dt id="park.fitresult.FitHandler.finalize">
1171<tt class="descname">finalize</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler.finalize"><span class="viewcode-link">[source]</span></a></dt>
1172<dd><p>Fit is complete; best results are reported</p>
1173</dd></dl>
1174
1175<dl class="method">
1176<dt id="park.fitresult.FitHandler.improvement">
1177<tt class="descname">improvement</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler.improvement"><span class="viewcode-link">[source]</span></a></dt>
1178<dd><p>Called when a result is observed which is better than previous
1179results from the fit.</p>
1180<p>result is a FitResult object, with parameters, #calls and fitness.</p>
1181</dd></dl>
1182
1183<dl class="method">
1184<dt id="park.fitresult.FitHandler.progress">
1185<tt class="descname">progress</tt><big>(</big><em>current</em>, <em>expected</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitHandler.progress"><span class="viewcode-link">[source]</span></a></dt>
1186<dd><p>Called each cycle of the fit, reporting the current and the
1187expected amount of work.   The meaning of these values is
1188optimizer dependent, but they can be converted into a percent
1189complete using (100*current)//expected.</p>
1190<p>Progress is updated each iteration of the fit, whatever that
1191means for the particular optimization algorithm.  It is called
1192after any calls to improvement for the iteration so that the
1193update handler can control I/O bandwidth by suppressing
1194intermediate improvements until the fit is complete.</p>
1195</dd></dl>
1196
1197<dl class="attribute">
1198<dt id="park.fitresult.FitHandler.result">
1199<tt class="descname">result</tt><em class="property"> = None</em></dt>
1200<dd><p>The current best result of the fit</p>
1201</dd></dl>
1202
1203</dd></dl>
1204
1205<dl class="class">
1206<dt id="park.fitresult.FitParameter">
1207<em class="property">class </em><tt class="descclassname">park.fitresult.</tt><tt class="descname">FitParameter</tt><big>(</big><em>name</em>, <em>range</em>, <em>value</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitParameter"><span class="viewcode-link">[source]</span></a></dt>
1208<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1209<p>Fit result for an individual parameter.</p>
1210<dl class="method">
1211<dt id="park.fitresult.FitParameter.summarize">
1212<tt class="descname">summarize</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitParameter.summarize"><span class="viewcode-link">[source]</span></a></dt>
1213<dd><p>Return parameter range string.</p>
1214<p>E.g.,  &#8221;       Gold .....|.... 5.2043 in [2,7]&#8221;</p>
1215</dd></dl>
1216
1217</dd></dl>
1218
1219<dl class="class">
1220<dt id="park.fitresult.FitResult">
1221<em class="property">class </em><tt class="descclassname">park.fitresult.</tt><tt class="descname">FitResult</tt><big>(</big><em>parameters</em>, <em>calls</em>, <em>fitness</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitResult"><span class="viewcode-link">[source]</span></a></dt>
1222<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1223<p>Container for reporting fit results.</p>
1224<dl class="method">
1225<dt id="park.fitresult.FitResult.calc_cov">
1226<tt class="descname">calc_cov</tt><big>(</big><em>fn</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitResult.calc_cov"><span class="viewcode-link">[source]</span></a></dt>
1227<dd><p>Return the covariance matrix inv(J&#8217;J) at point p.</p>
1228</dd></dl>
1229
1230<dl class="attribute">
1231<dt id="park.fitresult.FitResult.calls">
1232<tt class="descname">calls</tt><em class="property"> = None</em></dt>
1233<dd><p>Number of function calls</p>
1234</dd></dl>
1235
1236<dl class="attribute">
1237<dt id="park.fitresult.FitResult.cov">
1238<tt class="descname">cov</tt><em class="property"> = None</em></dt>
1239<dd><p>Covariance matrix</p>
1240</dd></dl>
1241
1242<dl class="attribute">
1243<dt id="park.fitresult.FitResult.fitness">
1244<tt class="descname">fitness</tt><em class="property"> = None</em></dt>
1245<dd><p>Value of the goodness of fit metric</p>
1246</dd></dl>
1247
1248<dl class="attribute">
1249<dt id="park.fitresult.FitResult.parameters">
1250<tt class="descname">parameters</tt><em class="property"> = None</em></dt>
1251<dd><p>Fit parameter list, each with name, range and value attributes.</p>
1252</dd></dl>
1253
1254<dl class="method">
1255<dt id="park.fitresult.FitResult.print_summary">
1256<tt class="descname">print_summary</tt><big>(</big><em>fid=&lt;open file '&lt;stdout&gt;'</em>, <em>mode 'w' at 0x10028e150&gt;</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitResult.print_summary"><span class="viewcode-link">[source]</span></a></dt>
1257<dd></dd></dl>
1258
1259<dl class="attribute">
1260<dt id="park.fitresult.FitResult.pvec">
1261<tt class="descname">pvec</tt><em class="property"> = None</em></dt>
1262<dd><p>Parameter vector</p>
1263</dd></dl>
1264
1265<dl class="method">
1266<dt id="park.fitresult.FitResult.set_cov">
1267<tt class="descname">set_cov</tt><big>(</big><em>cov</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitResult.set_cov"><span class="viewcode-link">[source]</span></a></dt>
1268<dd><p>Return the covariance matrix inv(J&#8217;J) at point p.</p>
1269</dd></dl>
1270
1271<dl class="attribute">
1272<dt id="park.fitresult.FitResult.stderr">
1273<tt class="descname">stderr</tt><em class="property"> = None</em></dt>
1274<dd><p>Parameter uncertainties</p>
1275</dd></dl>
1276
1277<dl class="method">
1278<dt id="park.fitresult.FitResult.update">
1279<tt class="descname">update</tt><big>(</big><em>pvec</em>, <em>fitness</em>, <em>calls</em><big>)</big><a class="reference internal" href="../../_modules/park/fitresult.html#FitResult.update"><span class="viewcode-link">[source]</span></a></dt>
1280<dd></dd></dl>
1281
1282</dd></dl>
1283
1284</div>
1285<div class="section" id="park-fitservice-module">
1286<h2>park.fitservice module</h2>
1287</div>
1288<div class="section" id="module-park.formatnum">
1289<span id="park-formatnum-module"></span><h2>park.formatnum module</h2>
1290<p>Format numbers nicely for printing.</p>
1291<p>Usage:</p>
1292<div class="highlight-python"><div class="highlight"><pre>&gt;&gt; from danse.common.util.formatnum import *
1293&gt;&gt; v,dv = 757.2356,0.01032
1294&gt;&gt; print format_uncertainty_pm(v,dv)
1295757.235 +/- 0.010
1296&gt;&gt; format_uncertainty_compact(v,dv)
1297757.235(10)
1298&gt;&gt; format_uncertainty(v,dv)
1299757.235(10)
1300</pre></div>
1301</div>
1302<p>Set format_uncertainty.compact to False to use the +/-
1303format by default, otherwise leave it at True for compact
1304value(##) format.</p>
1305<dl class="function">
1306<dt id="park.formatnum.format_uncertainty_pm">
1307<tt class="descclassname">park.formatnum.</tt><tt class="descname">format_uncertainty_pm</tt><big>(</big><em>value</em>, <em>uncertainty=None</em><big>)</big><a class="reference internal" href="../../_modules/park/formatnum.html#format_uncertainty_pm"><span class="viewcode-link">[source]</span></a></dt>
1308<dd><p>Given <em>value</em> v and <em>uncertainty</em> dv, return a string v +/- dv.</p>
1309<p>The returned string uses only the number of digits warranted by
1310the uncertainty in the measurement.</p>
1311<p>If the uncertainty is 0 or not otherwise provided, the simple
1312%g floating point format option is used.</p>
1313<p>Infinite and indefinite numbers are represented as inf and NaN.</p>
1314</dd></dl>
1315
1316<dl class="function">
1317<dt id="park.formatnum.format_uncertainty_compact">
1318<tt class="descclassname">park.formatnum.</tt><tt class="descname">format_uncertainty_compact</tt><big>(</big><em>value</em>, <em>uncertainty=None</em><big>)</big><a class="reference internal" href="../../_modules/park/formatnum.html#format_uncertainty_compact"><span class="viewcode-link">[source]</span></a></dt>
1319<dd><p>Given <em>value</em> v and <em>uncertainty</em> dv, return the compact
1320representation v(##), where ## are the first two digits of
1321the uncertainty.</p>
1322<p>The returned string uses only the number of digits warranted by
1323the uncertainty in the measurement.</p>
1324<p>If the uncertainty is 0 or not otherwise provided, the simple
1325%g floating point format option is used.</p>
1326<p>Infinite and indefinite numbers are represented as inf and NaN.</p>
1327</dd></dl>
1328
1329</div>
1330<div class="section" id="module-park.message">
1331<span id="park-message-module"></span><h2>park.message module</h2>
1332<p>Asynchronous message streams.</p>
1333<p>When you have multiple listeners to a process, some of which can connect and
1334disconnect at different times, you need to dispatch the incoming messages to
1335all the listeners.  When a listener joins a running stream, they need to get
1336an immediate status update for the computation.  This is stored as the
1337stream header. Then, without dropping any messages, the listeners should
1338receive all subsequent messages in the stream in order.</p>
1339<p>The message stream is multi-channelled, and indexed by a key.  Within
1340the service framework the key is likely to be the jobid associated with
1341the message stream.</p>
1342<p>The contents of the message stream are expected to be monitor messages
1343(see <cite>park.monitor</cite> for details), with the stream header being a
1344<cite>park.monitor.Join</cite> message.  When a new listener is registered,
1345the header is immediately put on the queue (if there is one), then all
1346subsequent message are sent until the listener calls hangup().</p>
1347<p>To attach to a message stream you need an object which accepts put().
1348An asynchronous queue object is a good choice since it allows you to
1349buffer the messages in one thread and pull them off as needed in another.
1350You can also use a <cite>park.monitor.Monitor</cite> object to process the messages
1351directly.</p>
1352</div>
1353<div class="section" id="module-park.model">
1354<span id="park-model-module"></span><h2>park.model module</h2>
1355<p>Define a park fitting model.</p>
1356<div class="section" id="id2">
1357<h3>Usage</h3>
1358<p>The simplest sort of fitting model is something like the following:</p>
1359<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">numpy</span>
1360<span class="kn">import</span> <span class="nn">park</span>
1361<span class="k">def</span> <span class="nf">G</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">mu</span><span class="p">,</span><span class="n">sigma</span><span class="p">):</span>
1362    <span class="k">return</span> <span class="n">numpy</span><span class="o">.</span><span class="n">exp</span><span class="p">(</span><span class="o">-</span><span class="mf">0.5</span><span class="o">*</span><span class="p">(</span><span class="n">x</span><span class="o">-</span><span class="n">mu</span><span class="p">)</span><span class="o">**</span><span class="mi">2</span><span class="o">/</span><span class="n">sigma</span><span class="o">**</span><span class="mi">2</span><span class="p">)</span>
1363
1364<span class="k">class</span> <span class="nc">Gauss</span><span class="p">(</span><span class="n">park</span><span class="o">.</span><span class="n">Model</span><span class="p">):</span>
1365    <span class="n">parameters</span> <span class="o">=</span> <span class="p">[</span><span class="s">&#39;center&#39;</span><span class="p">,</span><span class="s">&#39;width&#39;</span><span class="p">,</span><span class="s">&#39;scale&#39;</span><span class="p">]</span>
1366    <span class="k">def</span> <span class="nf">eval</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="n">x</span><span class="p">):</span>
1367        <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">scale</span> <span class="o">*</span> <span class="n">G</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="bp">self</span><span class="o">.</span><span class="n">center</span><span class="p">,</span><span class="bp">self</span><span class="o">.</span><span class="n">width</span><span class="p">)</span>
1368</pre></div>
1369</div>
1370<p>It has a function which is evaluated at a series of x values and
1371a set of adjustable parameters controlling the shape of f(x).</p>
1372<p>You can check your module with something like the following:</p>
1373<div class="highlight-python"><div class="highlight"><pre>$ ipython -pylab
1374
1375from gauss import Gauss
1376
1377g = Gauss(center=5,width=1,scale=10)
1378x = asarray([1,2,3,4,5])
1379y = g(x)
1380plot(x,y)
1381</pre></div>
1382</div>
1383<p>This should produce a plot of the Gaussian peak.</p>
1384<p>You will then want to try your model with some data.  Create a file
1385with some dummy data, such as gauss.dat:</p>
1386<div class="highlight-python"><div class="highlight"><pre># x y
13874 2
13885 6
13896 7
13907 3
1391</pre></div>
1392</div>
1393<p>In order to match the model to data, you need to define a fitness
1394object.  This shows the difference between the model and the data,
1395which you can then plot or sum to create a weighted chisq value:</p>
1396<div class="highlight-python"><div class="highlight"><pre><span class="n">f</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">Fitness</span><span class="p">(</span><span class="n">g</span><span class="p">,</span> <span class="s">&#39;gauss.dat&#39;</span><span class="p">)</span>
1397<span class="n">plot</span><span class="p">(</span><span class="n">f</span><span class="o">.</span><span class="n">data</span><span class="o">.</span><span class="n">fit_x</span><span class="p">,</span> <span class="n">f</span><span class="o">.</span><span class="n">residuals</span><span class="p">())</span>
1398</pre></div>
1399</div>
1400<p>The data file can have up to four columns, either x,y or x,y,dy
1401or x,dx,y,dy where x,y are the measurement points and values,
1402dx is the instrument resolution in x and dy is the uncertainty
1403in the measurement value y.  You can pass any keyword arguments
1404to data.load that are accepted by numpy.loadtxt.  For example,
1405you can reorder columns or skip rows.  Additionally, you can modify
1406data attributes directly x,y,dx,dy and calc_x.  See help on park.Data1D
1407for details.</p>
1408<p>Once you have your model debugged you can use it in a fit:</p>
1409<div class="highlight-python"><div class="highlight"><pre><span class="n">g</span> <span class="o">=</span> <span class="n">Gauss</span><span class="p">(</span><span class="n">center</span><span class="o">=</span><span class="p">[</span><span class="mi">3</span><span class="p">,</span><span class="mi">5</span><span class="p">],</span><span class="n">scale</span><span class="o">=</span><span class="mf">7.2</span><span class="p">,</span><span class="n">width</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mi">3</span><span class="p">])</span>
1410<span class="n">result</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">fit</span><span class="p">((</span><span class="n">g</span><span class="p">,</span> <span class="s">&#39;gauss.dat&#39;</span><span class="p">))</span>
1411<span class="n">result</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
1412</pre></div>
1413</div>
1414<p>In this example, center and width are allowed to vary but scale is fixed.</p>
1415<p>Existing models can be readily adapted to Park:</p>
1416<div class="highlight-python"><div class="highlight"><pre><span class="k">class</span> <span class="nc">Gauss</span><span class="p">(</span><span class="nb">object</span><span class="p">):</span>
1417    <span class="s">&quot;Existing model&quot;</span>
1418    <span class="n">center</span><span class="p">,</span><span class="n">width</span><span class="p">,</span><span class="n">scale</span> <span class="o">=</span> <span class="mi">0</span><span class="p">,</span><span class="mi">1</span><span class="p">,</span><span class="mi">1</span>
1419    <span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="o">**</span><span class="n">kw</span><span class="p">):</span>
1420        <span class="k">for</span> <span class="n">k</span><span class="p">,</span><span class="n">v</span> <span class="ow">in</span> <span class="n">kw</span><span class="o">.</span><span class="n">items</span><span class="p">():</span> <span class="nb">setattr</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="n">k</span><span class="p">,</span><span class="n">v</span><span class="p">)</span>
1421    <span class="k">def</span> <span class="nf">eval</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="n">x</span><span class="p">):</span>
1422        <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">scale</span> <span class="o">*</span><span class="n">G</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="bp">self</span><span class="o">.</span><span class="n">center</span><span class="p">,</span><span class="bp">self</span><span class="o">.</span><span class="n">width</span><span class="p">)</span>
1423
1424<span class="k">class</span> <span class="nc">GaussAdaptor</span><span class="p">(</span><span class="n">Gauss</span><span class="p">,</span><span class="n">Model</span><span class="p">):</span>
1425    <span class="s">&quot;PARK adaptor&quot;</span>
1426    <span class="n">parameters</span> <span class="o">=</span> <span class="p">[</span><span class="s">&#39;center&#39;</span><span class="p">,</span><span class="s">&#39;width&#39;</span><span class="p">,</span><span class="s">&#39;scale&#39;</span><span class="p">]</span>
1427    <span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="o">*</span><span class="n">args</span><span class="p">,</span><span class="o">**</span><span class="n">kw</span><span class="p">):</span>
1428        <span class="n">Model</span><span class="o">.</span><span class="n">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="o">*</span><span class="n">args</span><span class="p">)</span>
1429        <span class="n">Gauss</span><span class="o">.</span><span class="n">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span><span class="o">**</span><span class="n">kw</span><span class="p">)</span>
1430
1431<span class="n">g</span> <span class="o">=</span> <span class="n">GaussAdaptor</span><span class="p">(</span><span class="n">center</span><span class="o">=</span><span class="p">[</span><span class="mi">3</span><span class="p">,</span><span class="mi">5</span><span class="p">],</span><span class="n">scale</span><span class="o">=</span><span class="mf">7.2</span><span class="p">,</span><span class="n">width</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mi">3</span><span class="p">])</span>
1432<span class="n">result</span> <span class="o">=</span> <span class="n">park</span><span class="o">.</span><span class="n">fit</span><span class="p">((</span><span class="n">g</span><span class="p">,</span> <span class="s">&#39;gauss.dat&#39;</span><span class="p">))</span>
1433<span class="n">result</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
1434</pre></div>
1435</div>
1436<p>Models can become much more complex than the ones described above,
1437including multilevel models where fitting parameters can be added
1438and removed dynamically.</p>
1439<p>In many cases the park optimizer will need an adaptor for pre-existing
1440models.  The adaptor above relies on python properties to translate
1441model.par access into model._par.get() and model._par.set() where _par
1442is the internal name for par.  This technique works for simple static
1443models, but will not work well for sophisticated models which have,
1444for example, a dynamic parameter set where the model parameters cannot
1445be set as properties.  A solution to this problem is to subclass the
1446park.Parameter and override the value attribute as a property.</p>
1447<p>Once models are defined they can be used in a variety of contexts, such
1448as simultaneous fitting with constraints between the parameters.  With
1449some care in defining the model, computationally intensive fits can
1450be distributed across multiple processors.  We provide a simple user
1451interface for interacting with the model parameters and managing fits.
1452This can be extended with model specialized model editors which format
1453the parameters in a sensible way for the model, or allow direct manipulation
1454of the model structure.  The underlying fitting engine can also be
1455used directly from your own user interface.</p>
1456<dl class="class">
1457<dt id="park.model.Model">
1458<em class="property">class </em><tt class="descclassname">park.model.</tt><tt class="descname">Model</tt><big>(</big><em>*args</em>, <em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/model.html#Model"><span class="viewcode-link">[source]</span></a></dt>
1459<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1460<p>Model definition.</p>
1461<p>The model manages attribute access to the fitting parameters and
1462also manages the dataset.</p>
1463<dl class="docutils">
1464<dt>derivatives [&#8216;p1&#8217;,&#8217;p2&#8217;,...]</dt>
1465<dd><p class="first">List of parameters for which the model can calculate
1466derivatives.  The derivs
1467The model function can compute the derivative with respect
1468to this parameter.  The function model.derivs(x,[p1,p2,...])
1469will return (f(x),df/dp1(x), ...).  The parameters and their
1470order are determined by the fitting engine.</p>
1471<p class="last">Note: This is a property of the model, not the fit.
1472Numerical derivatives will be used if the parameter is
1473used in an expression or if no analytic derivative is
1474available for the parameter.  Automatic differentiation
1475on parameter expressions is possible, but beyond the scope
1476of this project.</p>
1477</dd>
1478<dt>eval(x)</dt>
1479<dd>Evaluate the model at x.  This must be defined by the subclass.</dd>
1480<dt>eval_deriv(x,pars=[])</dt>
1481<dd>Evaluate the model and the derivatives at x.  This must be
1482defined by the subclass.</dd>
1483<dt>parameters</dt>
1484<dd>The names of the model parameters.  If this is not provided, then
1485the model will search the subclass for park.Parameter attributes
1486and construct the list of names from that.  Any parameters in the
1487list not already defined as park.Parameter attributes will be
1488defined as parameters with a default of 0.</dd>
1489<dt>parameterset</dt>
1490<dd>The set of parameters defined by the model.  These are the
1491parameters themselves, gathered into a park.ParameterSet.</dd>
1492</dl>
1493<p>The minimum fittng model if you choose not to subclass park.Model
1494requires parameterset and a residuals() method.</p>
1495<dl class="attribute">
1496<dt id="park.model.Model.derivatives">
1497<tt class="descname">derivatives</tt><em class="property"> = []</em></dt>
1498<dd></dd></dl>
1499
1500<dl class="method">
1501<dt id="park.model.Model.eval">
1502<tt class="descname">eval</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/model.html#Model.eval"><span class="viewcode-link">[source]</span></a></dt>
1503<dd><p>Evaluate the model at x.</p>
1504<p>This method needs to be specialized in the model to evaluate the
1505model function.  Alternatively, the model can implement is own
1506version of residuals which calculates the residuals directly
1507instead of calling eval.</p>
1508</dd></dl>
1509
1510<dl class="method">
1511<dt id="park.model.Model.eval_derivs">
1512<tt class="descname">eval_derivs</tt><big>(</big><em>x</em>, <em>pars=[]</em><big>)</big><a class="reference internal" href="../../_modules/park/model.html#Model.eval_derivs"><span class="viewcode-link">[source]</span></a></dt>
1513<dd><p>Evaluate the model and derivatives wrt pars at x.</p>
1514<p>pars is a list of the names of the parameters for which derivatives
1515are desired.</p>
1516<p>This method needs to be specialized in the model to evaluate the
1517model function.  Alternatively, the model can implement is own
1518version of residuals which calculates the residuals directly
1519instead of calling eval.</p>
1520</dd></dl>
1521
1522<dl class="attribute">
1523<dt id="park.model.Model.parameters">
1524<tt class="descname">parameters</tt><em class="property"> = []</em></dt>
1525<dd></dd></dl>
1526
1527<dl class="attribute">
1528<dt id="park.model.Model.parameterset">
1529<tt class="descname">parameterset</tt><em class="property"> = []</em></dt>
1530<dd></dd></dl>
1531
1532<dl class="method">
1533<dt id="park.model.Model.set">
1534<tt class="descname">set</tt><big>(</big><em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/model.html#Model.set"><span class="viewcode-link">[source]</span></a></dt>
1535<dd><p>Set the initial value for a set of parameters.</p>
1536<p>E.g., model.set(width=3,center=5)</p>
1537</dd></dl>
1538
1539</dd></dl>
1540
1541</div>
1542</div>
1543<div class="section" id="module-park.monitor">
1544<span id="park-monitor-module"></span><h2>park.monitor module</h2>
1545<p>Asychronous execution monitoring service.</p>
1546<p>Long running computations need to convey status information to the user.
1547This status can take multiple forms, such as output to the console or
1548activity on a GUI, or even mail to your inbox.</p>
1549<p>park.monitor defines several standard message types:</p>
1550<div class="highlight-python"><div class="highlight"><pre>`Start` for job start
1551`Join` first message when joining an already running job
1552`Progress` for job activity
1553`Improvement` for partial results
1554`Complete` for final result
1555`Abort` when job is killed
1556`Error` when job has an error
1557`Log` for various debugging messages
1558</pre></div>
1559</div>
1560<p>Individual services may have specialized message types.</p>
1561<p>park.monitor also defines <cite>Monitor</cite> to process the various kinds of messages,
1562and dispatch them to the various user defined handlers.</p>
1563<p>For each message type, the Monitor dispatcher will look for a function
1564named onMonitorQQQ where QQQ is the message type.  For example,
1565onMonitorStart(self, message) will be called in response to a Start message.
1566If onMonitorQQQ is not defined, then onMonitorMessage will be called.  The
1567default behaviour of onMonitorMessage is to print the message on the console.</p>
1568<p>Log messages are sent to the standard system logger.  See logging in the
1569python standard library for details.</p>
1570<p>The Monitor class has methods for onMonitorStart(message), etc.
1571In panel, be sure to have methods for onMonitorStart(message),
1572onMonitorProgress(message), etc., for the kinds of monitor messages
1573the application will send.  The catch-all method is onMonitorMessage.</p>
1574<p>See <cite>park.monitor</cite> for details on the message types.  Individual services
1575may have additional message types.</p>
1576<dl class="class">
1577<dt id="park.monitor.Monitor">
1578<em class="property">class </em><tt class="descclassname">park.monitor.</tt><tt class="descname">Monitor</tt><a class="reference internal" href="../../_modules/park/monitor.html#Monitor"><span class="viewcode-link">[source]</span></a></dt>
1579<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1580<p>Messages that are received during the processing of the job.</p>
1581<p>Standard message types:</p>
1582<div class="highlight-python"><div class="highlight"><pre><span class="sb">`Start`</span><span class="p">,</span> <span class="sb">`Progress`</span><span class="p">,</span> <span class="sb">`Improvement`</span><span class="p">,</span> <span class="sb">`Complete`</span><span class="p">,</span> <span class="sb">`Error`</span><span class="p">,</span> <span class="sb">`Abort`</span><span class="p">,</span> <span class="sb">`Log`</span>
1583</pre></div>
1584</div>
1585<p>Specific job types may have their own monitor messages.</p>
1586<p>The messages themselves should all produce nicely formatted results
1587in response to str(message).</p>
1588<p>The message dispatch calls on&lt;Class&gt;(message) if the on&lt;Class&gt;
1589method exists for the message type.  If not, then dispatch
1590calls otherwise(message).  By default onLog(message) submits the
1591log record to the logger.</p>
1592<p>Subclass Monitor to define your own behaviours.</p>
1593<dl class="method">
1594<dt id="park.monitor.Monitor.onMonitorLog">
1595<tt class="descname">onMonitorLog</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/monitor.html#Monitor.onMonitorLog"><span class="viewcode-link">[source]</span></a></dt>
1596<dd><p>Called when the job sends a logging record.</p>
1597<p>The logging record contains a normal python logging record.</p>
1598<p>The default behaviour is to tie into the application logging
1599system using:</p>
1600<div class="highlight-python"><div class="highlight"><pre><span class="n">logger</span> <span class="o">=</span> <span class="n">logging</span><span class="o">.</span><span class="n">getLogger</span><span class="p">(</span><span class="n">message</span><span class="o">.</span><span class="n">record</span><span class="o">.</span><span class="n">name</span><span class="p">)</span>
1601<span class="n">logger</span><span class="o">.</span><span class="n">handle</span><span class="p">(</span><span class="n">message</span><span class="o">.</span><span class="n">record</span><span class="p">)</span>
1602</pre></div>
1603</div>
1604<p>Logging levels are set in the job controller.</p>
1605</dd></dl>
1606
1607<dl class="method">
1608<dt id="park.monitor.Monitor.onMonitorMessage">
1609<tt class="descname">onMonitorMessage</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/monitor.html#Monitor.onMonitorMessage"><span class="viewcode-link">[source]</span></a></dt>
1610<dd><p>What to do if the message handler is not found.</p>
1611<p>Default is to ignore the message.</p>
1612</dd></dl>
1613
1614<dl class="method">
1615<dt id="park.monitor.Monitor.put">
1616<tt class="descname">put</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/monitor.html#Monitor.put"><span class="viewcode-link">[source]</span></a></dt>
1617<dd><p>Called from thread when new message has arrived.</p>
1618</dd></dl>
1619
1620</dd></dl>
1621
1622</div>
1623<div class="section" id="module-park.parameter">
1624<span id="park-parameter-module"></span><h2>park.parameter module</h2>
1625<p>Parameters and parameter sets.</p>
1626<p>Parameter defines an individual parameter, and ParameterSet groups them
1627into a hierarchy.</p>
1628<p>Individual models need to provide a parameter set with the correct
1629properties, either by using park.ParameterSet in their model definition,
1630or by providing a wrapper which can translate assignment to parameter.value
1631into the appropriate change in the wrapped model.  See wrapper.py for
1632an example.</p>
1633<dl class="class">
1634<dt id="park.parameter.Parameter">
1635<em class="property">class </em><tt class="descclassname">park.parameter.</tt><tt class="descname">Parameter</tt><big>(</big><em>name='unknown'</em>, <em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter"><span class="viewcode-link">[source]</span></a></dt>
1636<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
1637<p>A parameter is a box for communicating with the fitting service.
1638Parameters can have a number of properties,</p>
1639<p>Parameters have a number of properties:</p>
1640<dl class="docutils">
1641<dt>name &#8220;string&#8221;</dt>
1642<dd><p class="first">name of the parameter within the parameter set.</p>
1643<p class="last">The name is read only.  You can rename a parameter but only
1644in the context of the parameter set which contains it, using
1645parameterset.rename(par,name).  This will change all expressions
1646containing the named parameter.</p>
1647</dd>
1648<dt>path</dt>
1649<dd>dotted name of the parameter within the set of models.  The
1650dotted name is automatically generated by the parameter set
1651before expressions are parsed and evaluated.  There are
1652some operations on parameter sets (such as renaming the
1653layer containing a parameter) which will force an adjustment
1654of all the underlying parameter names, as well as any
1655expressions in which they are referenced.</dd>
1656<dt>limits (low, high)</dt>
1657<dd>hard limits on the range of allowed parameter values, dictated
1658by physics rather than by knowledge of the particular system.
1659For example, thickness parameters would have limits (0,inf)
1660because negative thicknesses are unphysical.  These limits
1661are enforced when setting range for the fit.</dd>
1662<dt>units &#8220;string&#8221;</dt>
1663<dd>units for the parameter.  This should be a string, but
1664be parsable by whatever units package your application
1665supports.</dd>
1666<dt>tip &#8220;string&#8221;</dt>
1667<dd>parameter description, suitable for use in a tooltip</dd>
1668<dt>value double</dt>
1669<dd>current value of the parameter, either fixed, fitted or computed</dd>
1670<dt>range (low, high)</dt>
1671<dd>range of expected values for the parameter in the model</dd>
1672<dt>expression &#8220;string&#8221;</dt>
1673<dd><p class="first">expression for the parameter in the model.  This is a string
1674containing a formula for updating the parameter value based
1675on the values of other parameters in the system.  The expression
1676is ignored if &#8216;calculated&#8217; is False.</p>
1677<p class="last">Note: the list of symbols available to the expression evaluator
1678defaults to the contents of the math module.  The caller will be
1679able to override this within the fitting fitting class.</p>
1680</dd>
1681<dt>status &#8216;fixed&#8217;|&#8217;computed&#8217;|&#8217;fitted&#8217;</dt>
1682<dd>the parameter type.  Choose &#8216;fixed&#8217; if the values is to
1683remain fixed throughout the fit, even if a range and an
1684expression have been specified.  Choose &#8216;computed&#8217; if the
1685value of the parameter is to be computed each time the
1686parameters are updated.  Choose &#8216;fitted&#8217; if an optimization
1687algorithm is supposed to search the parameter space.</dd>
1688<dt>likelihood</dt>
1689<dd><p class="first">function to return the negative log likelihood of seeing a
1690particular parameter value.  2*likelihood(value) will be added
1691to the total cost function for the particular parameter set
1692during the fit.  This will be on par with the probabilty
1693of seeing the particular theory function given the observed
1694datapoints when performing the fit (the residual term is
1695closely related to the log likelihood of the normal distribution).</p>
1696<p class="last">Note: we are minimizing chi^2 = sum [ ((y-f(x;p))/dy)^2 ] rather
1697than  -log P = sum [ ((y-f(x;p))/dy)^2/2 + log(2 pi dy^2) ],
1698where P is the probability of seeing f(x;p) given y,dy as the
1699mean and standard deviation of a normal distribution.  Because
1700chi^2_p = - 2 * log P_p + constant, the minima of p are the same
1701for chi^2 and negative log likelihood.  However, to weight the
1702likelihood properly when adding likelihood values to chisq, we
1703need the extra factor of 2 mentioned above.  The usual statistical
1704implications of normalized chi^2 will of course be suspect, both
1705because the assumption of independence between the points in
1706chi^2 (which definitely do not hold for the new &#8216;point&#8217; p_k), and
1707because of the additional 2 log(2 pi dp_k^2) constant, but given
1708the uncertainty in the estimate of the distribution parameters,
1709this is likely a minor point.</p>
1710</dd>
1711</dl>
1712<dl class="attribute">
1713<dt id="park.parameter.Parameter.deriv">
1714<tt class="descname">deriv</tt><em class="property"> = False</em></dt>
1715<dd></dd></dl>
1716
1717<dl class="attribute">
1718<dt id="park.parameter.Parameter.expression">
1719<tt class="descname">expression</tt><em class="property"> = ''</em></dt>
1720<dd></dd></dl>
1721
1722<dl class="method">
1723<dt id="park.parameter.Parameter.get">
1724<tt class="descname">get</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.get"><span class="viewcode-link">[source]</span></a></dt>
1725<dd><p>Return the current value for a parameter.</p>
1726</dd></dl>
1727
1728<dl class="method">
1729<dt id="park.parameter.Parameter.iscomputed">
1730<tt class="descname">iscomputed</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.iscomputed"><span class="viewcode-link">[source]</span></a></dt>
1731<dd></dd></dl>
1732
1733<dl class="method">
1734<dt id="park.parameter.Parameter.isfeasible">
1735<tt class="descname">isfeasible</tt><big>(</big><em>value</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.isfeasible"><span class="viewcode-link">[source]</span></a></dt>
1736<dd><p>Return true if the value is in the range</p>
1737</dd></dl>
1738
1739<dl class="method">
1740<dt id="park.parameter.Parameter.isfitted">
1741<tt class="descname">isfitted</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.isfitted"><span class="viewcode-link">[source]</span></a></dt>
1742<dd></dd></dl>
1743
1744<dl class="method">
1745<dt id="park.parameter.Parameter.isfixed">
1746<tt class="descname">isfixed</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.isfixed"><span class="viewcode-link">[source]</span></a></dt>
1747<dd></dd></dl>
1748
1749<dl class="method">
1750<dt id="park.parameter.Parameter.isrestrained">
1751<tt class="descname">isrestrained</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.isrestrained"><span class="viewcode-link">[source]</span></a></dt>
1752<dd></dd></dl>
1753
1754<dl class="attribute">
1755<dt id="park.parameter.Parameter.likelihood">
1756<tt class="descname">likelihood</tt><em class="property"> = None</em></dt>
1757<dd></dd></dl>
1758
1759<dl class="attribute">
1760<dt id="park.parameter.Parameter.limits">
1761<tt class="descname">limits</tt><em class="property"> = (-inf, inf)</em></dt>
1762<dd></dd></dl>
1763
1764<dl class="attribute">
1765<dt id="park.parameter.Parameter.name">
1766<tt class="descname">name</tt></dt>
1767<dd><p>parameter name</p>
1768</dd></dl>
1769
1770<dl class="attribute">
1771<dt id="park.parameter.Parameter.path">
1772<tt class="descname">path</tt><em class="property"> = ''</em></dt>
1773<dd></dd></dl>
1774
1775<dl class="attribute">
1776<dt id="park.parameter.Parameter.range">
1777<tt class="descname">range</tt></dt>
1778<dd><p>parameter range</p>
1779</dd></dl>
1780
1781<dl class="method">
1782<dt id="park.parameter.Parameter.set">
1783<tt class="descname">set</tt><big>(</big><em>value</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.set"><span class="viewcode-link">[source]</span></a></dt>
1784<dd><p>Set a parameter to a value, a range or an expression.  If it is a value,
1785the parameter will be fixed for the fit.  If it is a range, the value
1786will be varying for the fit.  If it is an expression, the parameter will
1787be calculated from the values of other parameters in the fit.</p>
1788<p>Raises ValueError if the value could not be interpreted.</p>
1789</dd></dl>
1790
1791<dl class="method">
1792<dt id="park.parameter.Parameter.setprefix">
1793<tt class="descname">setprefix</tt><big>(</big><em>prefix</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.setprefix"><span class="viewcode-link">[source]</span></a></dt>
1794<dd><p>Set the full path to the parameter as used in expressions involving
1795the parameter name.</p>
1796</dd></dl>
1797
1798<dl class="attribute">
1799<dt id="park.parameter.Parameter.status">
1800<tt class="descname">status</tt><em class="property"> = 'fixed'</em></dt>
1801<dd></dd></dl>
1802
1803<dl class="method">
1804<dt id="park.parameter.Parameter.summarize">
1805<tt class="descname">summarize</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#Parameter.summarize"><span class="viewcode-link">[source]</span></a></dt>
1806<dd><p>Return parameter range string.</p>
1807<p>E.g.,  &#8221;       Gold .....|.... 5.2043 in [2,7]&#8221;</p>
1808</dd></dl>
1809
1810<dl class="attribute">
1811<dt id="park.parameter.Parameter.tip">
1812<tt class="descname">tip</tt><em class="property"> = 'Fitting parameter'</em></dt>
1813<dd></dd></dl>
1814
1815<dl class="attribute">
1816<dt id="park.parameter.Parameter.units">
1817<tt class="descname">units</tt><em class="property"> = ''</em></dt>
1818<dd></dd></dl>
1819
1820<dl class="attribute">
1821<dt id="park.parameter.Parameter.value">
1822<tt class="descname">value</tt><em class="property"> = 0.0</em></dt>
1823<dd></dd></dl>
1824
1825</dd></dl>
1826
1827<dl class="class">
1828<dt id="park.parameter.ParameterSet">
1829<em class="property">class </em><tt class="descclassname">park.parameter.</tt><tt class="descname">ParameterSet</tt><big>(</big><em>name='unknown'</em>, <em>pars=[]</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet"><span class="viewcode-link">[source]</span></a></dt>
1830<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">list</span></tt></p>
1831<p>The set of parameters used to fit theory to data.</p>
1832<p>ParameterSet forms a hierarchy of parameters.  The parameters themselves
1833are referred to by the path through the hierarchy, usually:</p>
1834<div class="highlight-python"><div class="highlight"><pre><span class="n">fitname</span><span class="o">.</span><span class="n">component</span><span class="o">.</span><span class="n">parameter</span>
1835</pre></div>
1836</div>
1837<p>Though more or fewer levels are permitted.  Parameters are assumed to
1838have a unique label throughout the fit.  This is required so that
1839expressions tying the results of one fit to another can uniquely
1840reference a parameter.</p>
1841<p>Attributes:</p>
1842<dl class="docutils">
1843<dt>name</dt>
1844<dd>the name of the parameter set</dd>
1845<dt>path</dt>
1846<dd>the full dotted name of the parameter set</dd>
1847<dt>context</dt>
1848<dd>a dictionary providing additional context for evaluating parameters;
1849Note that this namespace is shared with other theory functions, so
1850populate it carefully.</dd>
1851</dl>
1852<dl class="method">
1853<dt id="park.parameter.ParameterSet.byname">
1854<tt class="descname">byname</tt><big>(</big><em>name</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet.byname"><span class="viewcode-link">[source]</span></a></dt>
1855<dd><p>Lookup parameter from dotted path</p>
1856</dd></dl>
1857
1858<dl class="attribute">
1859<dt id="park.parameter.ParameterSet.computed">
1860<tt class="descname">computed</tt></dt>
1861<dd><p>Return the subset of the parameters which are calculated</p>
1862</dd></dl>
1863
1864<dl class="attribute">
1865<dt id="park.parameter.ParameterSet.fitted">
1866<tt class="descname">fitted</tt></dt>
1867<dd><p>Return the subset of the paramters which are varying</p>
1868</dd></dl>
1869
1870<dl class="attribute">
1871<dt id="park.parameter.ParameterSet.fixed">
1872<tt class="descname">fixed</tt></dt>
1873<dd><p>Return the subset of the parameters which are fixed</p>
1874</dd></dl>
1875
1876<dl class="method">
1877<dt id="park.parameter.ParameterSet.flatten">
1878<tt class="descname">flatten</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet.flatten"><span class="viewcode-link">[source]</span></a></dt>
1879<dd><p>Iterate over the elements in depth first order.</p>
1880</dd></dl>
1881
1882<dl class="method">
1883<dt id="park.parameter.ParameterSet.gather_context">
1884<tt class="descname">gather_context</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet.gather_context"><span class="viewcode-link">[source]</span></a></dt>
1885<dd><p>Gather all additional symbols that can be used in expressions.</p>
1886<p>For example, if reflectometry provides a volume fraction
1887function volfrac(rho1,rho2,frac) to compute densities, then
1888this function can be added as a context dictionary to the
1889reflectometry parameter set.  Note that there is no guarantee
1890which function will be used if the same function exists in
1891two separate contexts.</p>
1892</dd></dl>
1893
1894<dl class="attribute">
1895<dt id="park.parameter.ParameterSet.name">
1896<tt class="descname">name</tt></dt>
1897<dd><p>parameter name</p>
1898</dd></dl>
1899
1900<dl class="attribute">
1901<dt id="park.parameter.ParameterSet.path">
1902<tt class="descname">path</tt><em class="property"> = ''</em></dt>
1903<dd></dd></dl>
1904
1905<dl class="method">
1906<dt id="park.parameter.ParameterSet.rename">
1907<tt class="descname">rename</tt><big>(</big><em>par</em>, <em>name</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet.rename"><span class="viewcode-link">[source]</span></a></dt>
1908<dd><p>Rename the parameter to something new.
1909Called from root of the parameter hierarchy, rename the particular
1910parameter object to something else.</p>
1911<p>This changes the internal name of the parameter, as well as all
1912expressions in which it occurs.  If the parameter is actually
1913a parameter set, then it renames all parameters in the set.</p>
1914</dd></dl>
1915
1916<dl class="attribute">
1917<dt id="park.parameter.ParameterSet.restrained">
1918<tt class="descname">restrained</tt></dt>
1919<dd><p>Return the subset of the parameters which have a likelihood
1920function associated with them.</p>
1921</dd></dl>
1922
1923<dl class="method">
1924<dt id="park.parameter.ParameterSet.setprefix">
1925<tt class="descname">setprefix</tt><big>(</big><em>prefix=None</em><big>)</big><a class="reference internal" href="../../_modules/park/parameter.html#ParameterSet.setprefix"><span class="viewcode-link">[source]</span></a></dt>
1926<dd><p>Fill in the full path name for all the parameters in the tree.</p>
1927<p>Note: this function must be called from the root parameter set
1928to build proper path names.</p>
1929<p>This is required before converting parameter expressions into
1930function calls.</p>
1931</dd></dl>
1932
1933</dd></dl>
1934
1935</div>
1936<div class="section" id="module-park.parkde">
1937<span id="park-parkde-module"></span><h2>park.parkde module</h2>
1938<dl class="class">
1939<dt id="park.parkde.DiffEv">
1940<em class="property">class </em><tt class="descclassname">park.parkde.</tt><tt class="descname">DiffEv</tt><big>(</big><em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/parkde.html#DiffEv"><span class="viewcode-link">[source]</span></a></dt>
1941<dd><p>Bases: <a class="reference internal" href="#park.fit.Fitter" title="park.fit.Fitter"><tt class="xref py py-class docutils literal"><span class="pre">park.fit.Fitter</span></tt></a></p>
1942<p>Differential evolution optimizer</p>
1943<p>This implements <cite>park.fit.Fitter</cite>.</p>
1944<dl class="attribute">
1945<dt id="park.parkde.DiffEv.Fscale">
1946<tt class="descname">Fscale</tt><em class="property"> = 0.5</em></dt>
1947<dd><p>Step size along difference vector</p>
1948</dd></dl>
1949
1950<dl class="attribute">
1951<dt id="park.parkde.DiffEv.crossover_rate">
1952<tt class="descname">crossover_rate</tt><em class="property"> = 0.9</em></dt>
1953<dd><p>Amount of mixing in population</p>
1954</dd></dl>
1955
1956<dl class="method">
1957<dt id="park.parkde.DiffEv.improvement">
1958<tt class="descname">improvement</tt><big>(</big><em>x</em>, <em>fx</em>, <em>ncalls=-1</em><big>)</big><a class="reference internal" href="../../_modules/park/parkde.html#DiffEv.improvement"><span class="viewcode-link">[source]</span></a></dt>
1959<dd></dd></dl>
1960
1961<dl class="attribute">
1962<dt id="park.parkde.DiffEv.maxiter">
1963<tt class="descname">maxiter</tt><em class="property"> = 1000</em></dt>
1964<dd><p>Maximum number of iterations</p>
1965</dd></dl>
1966
1967<dl class="attribute">
1968<dt id="park.parkde.DiffEv.pop_scale">
1969<tt class="descname">pop_scale</tt><em class="property"> = 4</em></dt>
1970<dd><p>Number of active points per dimension</p>
1971</dd></dl>
1972
1973<dl class="method">
1974<dt id="park.parkde.DiffEv.progress">
1975<tt class="descname">progress</tt><big>(</big><em>k</em>, <em>n</em><big>)</big><a class="reference internal" href="../../_modules/park/parkde.html#DiffEv.progress"><span class="viewcode-link">[source]</span></a></dt>
1976<dd></dd></dl>
1977
1978<dl class="attribute">
1979<dt id="park.parkde.DiffEv.tolerance">
1980<tt class="descname">tolerance</tt><em class="property"> = 1e-05</em></dt>
1981<dd><p>Fit tolerance</p>
1982</dd></dl>
1983
1984</dd></dl>
1985
1986</div>
1987<div class="section" id="park-parksnob-module">
1988<h2>park.parksnob module</h2>
1989</div>
1990<div class="section" id="module-park.peaks">
1991<span id="park-peaks-module"></span><h2>park.peaks module</h2>
1992<p>A park model implementing a multipeak fitter.</p>
1993<p><strong>* WARNING *</strong> this example was used to inform the design process,
1994and has not yet been updated to correspond to the current implementation.
1995Do not use this as a tutorial.</p>
1996<p>This is an example model showing how to put together a multi-part
1997fit objective function.</p>
1998<table class="docutils field-list" frame="void" rules="none">
1999<col class="field-name" />
2000<col class="field-body" />
2001<tbody valign="top">
2002<tr class="field-odd field"><th class="field-name">group main:</th><td class="field-body">Peaks</td>
2003</tr>
2004<tr class="field-even field"><th class="field-name">group peaks:</th><td class="field-body">Gaussian, Lorentzian, Voigt</td>
2005</tr>
2006<tr class="field-odd field"><th class="field-name" colspan="2">group backgrounds:</th></tr>
2007<tr class="field-odd field"><td>&nbsp;</td><td class="field-body">Constant, Quadratic, Linear</td>
2008</tr>
2009</tbody>
2010</table>
2011<dl class="class">
2012<dt id="park.peaks.Constant">
2013<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Constant</tt><big>(</big><em>name</em>, <em>C=0</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Constant"><span class="viewcode-link">[source]</span></a></dt>
2014<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2015<p>Constant background: C</p>
2016<dl class="method">
2017<dt id="park.peaks.Constant.dC">
2018<tt class="descname">dC</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Constant.dC"><span class="viewcode-link">[source]</span></a></dt>
2019<dd></dd></dl>
2020
2021<dl class="attribute">
2022<dt id="park.peaks.Constant.name">
2023<tt class="descname">name</tt><em class="property"> = 'Background: constant'</em></dt>
2024<dd></dd></dl>
2025
2026</dd></dl>
2027
2028<dl class="class">
2029<dt id="park.peaks.Gaussian">
2030<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Gaussian</tt><big>(</big><em>name</em>, <em>scale=1.0</em>, <em>center=0</em>, <em>sigma=1</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Gaussian"><span class="viewcode-link">[source]</span></a></dt>
2031<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2032<p>Gaussian peak: scale exp ( -0.5 (x-center)**2 / sigma**2 )</p>
2033<dl class="method">
2034<dt id="park.peaks.Gaussian.dcenter">
2035<tt class="descname">dcenter</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Gaussian.dcenter"><span class="viewcode-link">[source]</span></a></dt>
2036<dd></dd></dl>
2037
2038<dl class="method">
2039<dt id="park.peaks.Gaussian.dscale">
2040<tt class="descname">dscale</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Gaussian.dscale"><span class="viewcode-link">[source]</span></a></dt>
2041<dd></dd></dl>
2042
2043<dl class="method">
2044<dt id="park.peaks.Gaussian.dsigma">
2045<tt class="descname">dsigma</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Gaussian.dsigma"><span class="viewcode-link">[source]</span></a></dt>
2046<dd></dd></dl>
2047
2048</dd></dl>
2049
2050<dl class="class">
2051<dt id="park.peaks.Linear">
2052<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Linear</tt><big>(</big><em>name</em>, <em>B=0</em>, <em>C=0</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Linear"><span class="viewcode-link">[source]</span></a></dt>
2053<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2054<p>Linear background: B x**2 + C</p>
2055<dl class="method">
2056<dt id="park.peaks.Linear.dB">
2057<tt class="descname">dB</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Linear.dB"><span class="viewcode-link">[source]</span></a></dt>
2058<dd></dd></dl>
2059
2060<dl class="method">
2061<dt id="park.peaks.Linear.dC">
2062<tt class="descname">dC</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Linear.dC"><span class="viewcode-link">[source]</span></a></dt>
2063<dd></dd></dl>
2064
2065<dl class="attribute">
2066<dt id="park.peaks.Linear.name">
2067<tt class="descname">name</tt><em class="property"> = 'Background: linear'</em></dt>
2068<dd></dd></dl>
2069
2070</dd></dl>
2071
2072<dl class="class">
2073<dt id="park.peaks.Lorentzian">
2074<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Lorentzian</tt><big>(</big><em>name</em>, <em>scale=1.0</em>, <em>center=0</em>, <em>gamma=1</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Lorentzian"><span class="viewcode-link">[source]</span></a></dt>
2075<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2076<p>Lorentzian peak (HWHM): scale/pi  gamma/((x-center)**2 + gamma**2)</p>
2077<dl class="method">
2078<dt id="park.peaks.Lorentzian.dcenter">
2079<tt class="descname">dcenter</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Lorentzian.dcenter"><span class="viewcode-link">[source]</span></a></dt>
2080<dd></dd></dl>
2081
2082<dl class="method">
2083<dt id="park.peaks.Lorentzian.dgamma">
2084<tt class="descname">dgamma</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Lorentzian.dgamma"><span class="viewcode-link">[source]</span></a></dt>
2085<dd></dd></dl>
2086
2087<dl class="method">
2088<dt id="park.peaks.Lorentzian.dscale">
2089<tt class="descname">dscale</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Lorentzian.dscale"><span class="viewcode-link">[source]</span></a></dt>
2090<dd></dd></dl>
2091
2092</dd></dl>
2093
2094<dl class="class">
2095<dt id="park.peaks.Peaks">
2096<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Peaks</tt><big>(</big><em>datafile=None</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Peaks"><span class="viewcode-link">[source]</span></a></dt>
2097<dd><p>Bases: <a class="reference internal" href="#park.model.Model" title="park.model.Model"><tt class="xref py py-class docutils literal"><span class="pre">park.model.Model</span></tt></a></p>
2098<p>Peak fitter</p>
2099<dl class="method">
2100<dt id="park.peaks.Peaks.add_peak">
2101<tt class="descname">add_peak</tt><big>(</big><em>type</em>, <em>name=None</em>, <em>**kw</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Peaks.add_peak"><span class="viewcode-link">[source]</span></a></dt>
2102<dd><p>Add a peak to the model.</p>
2103<p>The name of the peak is used to distinguish it from other peaks
2104in the model, and appears as part of the parameter name in
2105constraint expressions.  For example, if name is &#8216;P3&#8217; and this
2106is part of model &#8216;M1&#8217; then the sigma parameter of a gaussian
2107peak would be &#8216;M1.P3.sigma&#8217;.</p>
2108<p>The peak can be specified either by peak type and initial arguments
2109or by the peak itself, which is a function of x with a parameters
2110attribute.</p>
2111<dl class="docutils">
2112<dt>Available peak types include:</dt>
2113<dd><dl class="first last docutils">
2114<dt>gaussian(scale=1, center=0, sigma=1) </dt>
2115<dd>= scale exp ( -0.5 (x-center)**2 / sigma**2 )</dd>
2116<dt>lorentzian(scale=1, center=0, gamma=1) </dt>
2117<dd>= scale/pi gamma / ((x-center)**2 + gamma**2))</dd>
2118<dt>voigt(scale=1, center=0, sigma=1, gamma=1)</dt>
2119<dd>= scale (lorentzian(mu,gamma) * gaussian(mu,sigma))(x-center)
2120where * represents convolution</dd>
2121</dl>
2122</dd>
2123<dt>Available background functions include:</dt>
2124<dd><dl class="first last docutils">
2125<dt>constant(C=0)</dt>
2126<dd>= C</dd>
2127<dt>slope(C=0, B=1)</dt>
2128<dd>= B x + C</dd>
2129<dt>quadratic(A=1, B=0, C=0)</dt>
2130<dd>= A x**2 + B x + C</dd>
2131</dl>
2132</dd>
2133</dl>
2134<p>Additional peak types can be registered.  The complete dictionary
2135of available types is in Peaks.peak_types.</p>
2136</dd></dl>
2137
2138<dl class="method">
2139<dt id="park.peaks.Peaks.eval">
2140<tt class="descname">eval</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Peaks.eval"><span class="viewcode-link">[source]</span></a></dt>
2141<dd><p>Returns the raw theory value at x, unconvoluted and unweighted.</p>
2142<p>The residuals will be calculated by park.Model.residuals by
2143calling</p>
2144</dd></dl>
2145
2146<dl class="attribute">
2147<dt id="park.peaks.Peaks.parameters">
2148<tt class="descname">parameters</tt><em class="property"> = []</em></dt>
2149<dd></dd></dl>
2150
2151<dl class="attribute">
2152<dt id="park.peaks.Peaks.parameterset">
2153<tt class="descname">parameterset</tt><em class="property"> = []</em></dt>
2154<dd></dd></dl>
2155
2156<dl class="attribute">
2157<dt id="park.peaks.Peaks.peak_types">
2158<tt class="descname">peak_types</tt><em class="property"> = {'Constant': &lt;class 'park.peaks.Constant'&gt;, 'Linear': &lt;class 'park.peaks.Linear'&gt;, 'Gaussian': &lt;class 'park.peaks.Gaussian'&gt;, 'Quadratic': &lt;class 'park.peaks.Quadratic'&gt;, 'Voigt': &lt;class 'park.peaks.Voigt'&gt;, 'Lorentzian': &lt;class 'park.peaks.Lorentzian'&gt;}</em></dt>
2159<dd></dd></dl>
2160
2161<dl class="classmethod">
2162<dt id="park.peaks.Peaks.register_peak_generator">
2163<em class="property">classmethod </em><tt class="descname">register_peak_generator</tt><big>(</big><em>type</em>, <em>factory</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Peaks.register_peak_generator"><span class="viewcode-link">[source]</span></a></dt>
2164<dd><p>Register a new peak generator.  This will return a callable
2165object with a parameter set that can be used in the peak
2166fitting class.</p>
2167<p>The individual peak functions a number of attributes:</p>
2168<dl class="docutils">
2169<dt>__name__</dt>
2170<dd>the name to be displayed in the list of peaks; the
2171default is __class__.__name__</dd>
2172<dt>__factory__</dt>
2173<dd>the name of a factory function in the python
2174namespace which returns a new peak.  These parameters
2175will be set from the stored parameter list when the
2176peak model is reloaded.  This can be omitted if it
2177is simply the module and name of the class.</dd>
2178<dt>__doc__</dt>
2179<dd>the tool tip to be displayed with the peak</dd>
2180<dt>__call__(x)</dt>
2181<dd>evaluate the peak function at x</dd>
2182</dl>
2183</dd></dl>
2184
2185<dl class="method">
2186<dt id="park.peaks.Peaks.remove_peak">
2187<tt class="descname">remove_peak</tt><big>(</big><em>name</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Peaks.remove_peak"><span class="viewcode-link">[source]</span></a></dt>
2188<dd><p>Remove the named peak from the model.</p>
2189</dd></dl>
2190
2191</dd></dl>
2192
2193<dl class="class">
2194<dt id="park.peaks.Quadratic">
2195<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Quadratic</tt><big>(</big><em>name</em>, <em>A=1</em>, <em>B=0</em>, <em>C=0</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Quadratic"><span class="viewcode-link">[source]</span></a></dt>
2196<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2197<p>Quadratic background: A x**2 + B x + C</p>
2198<dl class="method">
2199<dt id="park.peaks.Quadratic.dA">
2200<tt class="descname">dA</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Quadratic.dA"><span class="viewcode-link">[source]</span></a></dt>
2201<dd></dd></dl>
2202
2203<dl class="method">
2204<dt id="park.peaks.Quadratic.dB">
2205<tt class="descname">dB</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Quadratic.dB"><span class="viewcode-link">[source]</span></a></dt>
2206<dd></dd></dl>
2207
2208<dl class="method">
2209<dt id="park.peaks.Quadratic.dC">
2210<tt class="descname">dC</tt><big>(</big><em>x</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Quadratic.dC"><span class="viewcode-link">[source]</span></a></dt>
2211<dd></dd></dl>
2212
2213<dl class="attribute">
2214<dt id="park.peaks.Quadratic.name">
2215<tt class="descname">name</tt><em class="property"> = 'Background: quadratic'</em></dt>
2216<dd></dd></dl>
2217
2218</dd></dl>
2219
2220<dl class="class">
2221<dt id="park.peaks.Voigt">
2222<em class="property">class </em><tt class="descclassname">park.peaks.</tt><tt class="descname">Voigt</tt><big>(</big><em>name</em>, <em>scale=1</em>, <em>center=0</em>, <em>sigma=1</em>, <em>gamma=1</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#Voigt"><span class="viewcode-link">[source]</span></a></dt>
2223<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2224<p>Voigt peak (HWHM,sigma): A [G(sigma) * L(gamma)](x-center)</p>
2225</dd></dl>
2226
2227<dl class="function">
2228<dt id="park.peaks.init">
2229<tt class="descclassname">park.peaks.</tt><tt class="descname">init</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#init"><span class="viewcode-link">[source]</span></a></dt>
2230<dd><p>Register peak types with the Peaks model.</p>
2231</dd></dl>
2232
2233<dl class="function">
2234<dt id="park.peaks.test">
2235<tt class="descclassname">park.peaks.</tt><tt class="descname">test</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#test"><span class="viewcode-link">[source]</span></a></dt>
2236<dd></dd></dl>
2237
2238<dl class="function">
2239<dt id="park.peaks.voigt">
2240<tt class="descclassname">park.peaks.</tt><tt class="descname">voigt</tt><big>(</big><em>x</em>, <em>sigma</em>, <em>gamma</em><big>)</big><a class="reference internal" href="../../_modules/park/peaks.html#voigt"><span class="viewcode-link">[source]</span></a></dt>
2241<dd><p>Return the voigt function, which is the convolution of a Lorentz
2242function with a Gaussian.</p>
2243<table class="docutils field-list" frame="void" rules="none">
2244<col class="field-name" />
2245<col class="field-body" />
2246<tbody valign="top">
2247<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><dl class="first last docutils">
2248<dt>gamma <span class="classifier-delimiter">:</span> <span class="classifier">real</span></dt>
2249<dd><p class="first last">The half-width half-maximum of the Lorentzian</p>
2250</dd>
2251<dt>sigma <span class="classifier-delimiter">:</span> <span class="classifier">real</span></dt>
2252<dd><p class="first last">The 1-sigma width of the Gaussian, which is one standard deviation.</p>
2253</dd>
2254</dl>
2255</td>
2256</tr>
2257</tbody>
2258</table>
2259<p>Ref: W.I.F. David, J. Appl. Cryst. (1986). 19, 63-64</p>
2260<p>Note: adjusted to use stddev and HWHM rather than FWHM parameters</p>
2261</dd></dl>
2262
2263</div>
2264<div class="section" id="module-park.pmap">
2265<span id="park-pmap-module"></span><h2>park.pmap module</h2>
2266<p>Parallel map-reduce implementation using threads.</p>
2267<dl class="class">
2268<dt id="park.pmap.Collector">
2269<em class="property">class </em><tt class="descclassname">park.pmap.</tt><tt class="descname">Collector</tt><a class="reference internal" href="../../_modules/park/pmap.html#Collector"><span class="viewcode-link">[source]</span></a></dt>
2270<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2271<p>Abstract interface to map-reduce accumulator function.</p>
2272<dl class="method">
2273<dt id="park.pmap.Collector.error">
2274<tt class="descname">error</tt><big>(</big><em>part</em>, <em>msg</em><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#Collector.error"><span class="viewcode-link">[source]</span></a></dt>
2275<dd><p>Exception seen on executing map or reduce.  The collector
2276can adjust the accumulated result appropriately to reflect
2277the error.</p>
2278</dd></dl>
2279
2280<dl class="method">
2281<dt id="park.pmap.Collector.finalize">
2282<tt class="descname">finalize</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#Collector.finalize"><span class="viewcode-link">[source]</span></a></dt>
2283<dd><p>Called when all parts have been accumulated</p>
2284</dd></dl>
2285
2286</dd></dl>
2287
2288<dl class="class">
2289<dt id="park.pmap.Mapper">
2290<em class="property">class </em><tt class="descclassname">park.pmap.</tt><tt class="descname">Mapper</tt><a class="reference internal" href="../../_modules/park/pmap.html#Mapper"><span class="viewcode-link">[source]</span></a></dt>
2291<dd><p>Bases: <tt class="xref py py-class docutils literal"><span class="pre">object</span></tt></p>
2292<p>Abstract interface to map-reduce mapper function.</p>
2293<dl class="method">
2294<dt id="park.pmap.Mapper.abort">
2295<tt class="descname">abort</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#Mapper.abort"><span class="viewcode-link">[source]</span></a></dt>
2296<dd><p>Stop the mapper</p>
2297</dd></dl>
2298
2299</dd></dl>
2300
2301<dl class="function">
2302<dt id="park.pmap.main">
2303<tt class="descclassname">park.pmap.</tt><tt class="descname">main</tt><big>(</big><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#main"><span class="viewcode-link">[source]</span></a></dt>
2304<dd></dd></dl>
2305
2306<dl class="function">
2307<dt id="park.pmap.pmap">
2308<tt class="descclassname">park.pmap.</tt><tt class="descname">pmap</tt><big>(</big><em>mapper</em>, <em>inputs</em><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#pmap"><span class="viewcode-link">[source]</span></a></dt>
2309<dd><p>Apply function mapper to all inputs.</p>
2310<p>This is the serial version of a parallel iterator, yielding the next
2311sequence value as soon as it is available.  There is no guarantee
2312that the order of the inputs will be preserved in the parallel
2313version, so don&#8217;t depend on it!</p>
2314</dd></dl>
2315
2316<dl class="function">
2317<dt id="park.pmap.pmapreduce">
2318<tt class="descclassname">park.pmap.</tt><tt class="descname">pmapreduce</tt><big>(</big><em>mapper</em>, <em>collector</em>, <em>inputs</em><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#pmapreduce"><span class="viewcode-link">[source]</span></a></dt>
2319<dd><p>Apply function mapper to inputs, accumulating the results in collector.</p>
2320<p>Collector is a function which accepts the result of mapper(item) for
2321each item of inputs.  There is no guarantee that the outputs will be
2322received in order.</p>
2323<p>The map is executed in a separate thread so the function returns
2324to the caller immediately.</p>
2325</dd></dl>
2326
2327<dl class="function">
2328<dt id="park.pmap.preduce">
2329<tt class="descclassname">park.pmap.</tt><tt class="descname">preduce</tt><big>(</big><em>collector</em>, <em>outputs</em><big>)</big><a class="reference internal" href="../../_modules/park/pmap.html#preduce"><span class="viewcode-link">[source]</span></a></dt>
2330<dd><p>Collect all outputs, calling collector(item) for each item in the sequence.</p>
2331</dd></dl>
2332
2333<dl class="data">
2334<dt id="park.pmap.profile_mapper">
2335<tt class="descclassname">park.pmap.</tt><tt class="descname">profile_mapper</tt><em class="property"> = False</em></dt>
2336<dd><p>True if the mapper cost should be profiled.</p>
2337</dd></dl>
2338
2339</div>
2340<div class="section" id="park-rangemap-module">
2341<h2>park.rangemap module</h2>
2342</div>
2343<div class="section" id="park-serial-module">
2344<h2>park.serial module</h2>
2345</div>
2346<div class="section" id="module-park.setup">
2347<span id="park-setup-module"></span><h2>park.setup module</h2>
2348<dl class="function">
2349<dt id="park.setup.configuration">
2350<tt class="descclassname">park.setup.</tt><tt class="descname">configuration</tt><big>(</big><em>parent_package=''</em>, <em>top_path=None</em><big>)</big><a class="reference internal" href="../../_modules/park/setup.html#configuration"><span class="viewcode-link">[source]</span></a></dt>
2351<dd></dd></dl>
2352
2353</div>
2354<div class="section" id="module-park.simplex">
2355<span id="park-simplex-module"></span><h2>park.simplex module</h2>
2356<p>Downhill simplex optimizer.</p>
2357<dl class="function">
2358<dt id="park.simplex.simplex">
2359<tt class="descclassname">park.simplex.</tt><tt class="descname">simplex</tt><big>(</big><em>f</em>, <em>x0=None</em>, <em>bounds=None</em>, <em>radius=0.05</em>, <em>xtol=0.0001</em>, <em>ftol=0.0001</em>, <em>maxiter=None</em>, <em>update_handler=None</em>, <em>abort_test=&lt;function dont_abort at 0x105341500&gt;</em><big>)</big><a class="reference internal" href="../../_modules/park/simplex.html#simplex"><span class="viewcode-link">[source]</span></a></dt>
2360<dd><p>Minimize a function using Nelder-Mead downhill simplex algorithm.</p>
2361<p>This optimizer is also known as Amoeba (from Numerical Recipes) and
2362the Nealder-Mead simplex algorithm.  This is not the simplex algorithm
2363for solving constrained linear systems.</p>
2364<p>Downhill simplex is a robust derivative free algorithm for finding
2365minima.  It proceeds by choosing a set of points (the simplex) forming
2366an n-dimensional triangle, and transforming that triangle so that the
2367worst vertex is improved, either by stretching, shrinking or reflecting
2368it about the center of the triangle.  This algorithm is not known for
2369its speed, but for its simplicity and robustness, and is a good algorithm
2370to start your problem with.</p>
2371<p><em>Parameters</em>:</p>
2372<blockquote>
2373<div><dl class="docutils">
2374<dt>f <span class="classifier-delimiter">:</span> <span class="classifier">callable f(x,*args)</span></dt>
2375<dd>The objective function to be minimized.</dd>
2376<dt>x0 <span class="classifier-delimiter">:</span> <span class="classifier">ndarray</span></dt>
2377<dd>Initial guess.</dd>
2378<dt>bounds <span class="classifier-delimiter">:</span> <span class="classifier">(ndarray,ndarray) or None</span></dt>
2379<dd>Bounds on the parameter values for the function.</dd>
2380<dt>radius: float</dt>
2381<dd>Size of the initial simplex.  For bounded parameters (those
2382which have finite lower and upper bounds), radius is clipped
2383to a value in (0,0.5] representing the portion of the
2384range to use as the size of the initial simplex.</dd>
2385</dl>
2386</div></blockquote>
2387<p><em>Returns</em>: Result (<cite>park.simplex.Result</cite>)</p>
2388<blockquote>
2389<div><dl class="docutils">
2390<dt>x <span class="classifier-delimiter">:</span> <span class="classifier">ndarray</span></dt>
2391<dd>Parameter that minimizes function.</dd>
2392<dt>fx <span class="classifier-delimiter">:</span> <span class="classifier">float</span></dt>
2393<dd>Value of function at minimum: <tt class="docutils literal"><span class="pre">fopt</span> <span class="pre">=</span> <span class="pre">func(xopt)</span></tt>.</dd>
2394<dt>iters <span class="classifier-delimiter">:</span> <span class="classifier">int</span></dt>
2395<dd>Number of iterations performed.</dd>
2396<dt>calls <span class="classifier-delimiter">:</span> <span class="classifier">int</span></dt>
2397<dd>Number of function calls made.</dd>
2398<dt>success <span class="classifier-delimiter">:</span> <span class="classifier">boolean</span></dt>
2399<dd>True if fit completed successfully.</dd>
2400</dl>
2401</div></blockquote>
2402<p><em>Other Parameters</em>:</p>
2403<blockquote>
2404<div><dl class="docutils">
2405<dt>xtol <span class="classifier-delimiter">:</span> <span class="classifier">float</span></dt>
2406<dd>Relative error in xopt acceptable for convergence.</dd>
2407<dt>ftol <span class="classifier-delimiter">:</span> <span class="classifier">number</span></dt>
2408<dd>Relative error in func(xopt) acceptable for convergence.</dd>
2409<dt>maxiter <span class="classifier-delimiter">:</span> <span class="classifier">int=200*N</span></dt>
2410<dd>Maximum number of iterations to perform.  Defaults</dd>
2411<dt>update_handler <span class="classifier-delimiter">:</span> <span class="classifier">callable</span></dt>
2412<dd>Called after each iteration, as callback(xk,fxk), where xk
2413is the current parameter vector and fxk is the function value.
2414Returns True if the fit should continue.</dd>
2415</dl>
2416</div></blockquote>
2417<p><em>Notes</em></p>
2418<blockquote>
2419<div>Uses a Nelder-Mead simplex algorithm to find the minimum of
2420function of one or more variables.</div></blockquote>
2421</dd></dl>
2422
2423</div>
2424<div class="section" id="module-park.version">
2425<span id="park-version-module"></span><h2>park.version module</h2>
2426<p>version and package information about PARK.</p>
2427</div>
2428<div class="section" id="module-park.wxmonitor">
2429<span id="park-wxmonitor-module"></span><h2>park.wxmonitor module</h2>
2430<p>Asynchronous monitoring service for wx applications.</p>
2431<p>Define a monitor using park.wxmonitor.wxMonitor(panel) where panel is
2432the window which will receive the monitor updates.</p>
2433<p>In panel, be sure to have methods for onMonitorStart(message),
2434onMonitorProgress(message), etc., for the kinds of monitor messages
2435the application will send.  The catch-all method is onMonitorMessage,
2436which by default will print the messages on the console.  If you
2437don&#8217;t catch onMonitorLog messages then the log messages will be
2438sent to the standard python logger.</p>
2439<p>See <cite>park.monitor</cite> for details on the message types.</p>
2440<div class="section" id="example">
2441<h3>Example</h3>
2442<p>The following defines a panel which responds to monitor messages:</p>
2443<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">wx</span>
2444
2445<span class="k">class</span> <span class="nc">Panel</span><span class="p">(</span><span class="n">wx</span><span class="o">.</span><span class="n">Panel</span><span class="p">):</span>
2446    <span class="k">def</span> <span class="nf">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">*</span><span class="n">args</span><span class="p">,</span> <span class="o">**</span><span class="n">kw</span><span class="p">):</span>
2447        <span class="n">wx</span><span class="o">.</span><span class="n">Panel</span><span class="o">.</span><span class="n">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">*</span><span class="n">args</span><span class="p">,</span> <span class="o">**</span><span class="n">kw</span><span class="p">)</span>
2448        <span class="bp">self</span><span class="o">.</span><span class="n">text</span> <span class="o">=</span> <span class="n">wx</span><span class="o">.</span><span class="n">TextCtrl</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="p">(</span><span class="mi">200</span><span class="p">,</span><span class="mi">100</span><span class="p">),</span> <span class="n">style</span><span class="o">=</span><span class="n">wx</span><span class="o">.</span><span class="n">TE_MULTILINE</span><span class="p">)</span>
2449        <span class="bp">self</span><span class="o">.</span><span class="n">gauge</span> <span class="o">=</span> <span class="n">wx</span><span class="o">.</span><span class="n">Gauge</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="nb">range</span><span class="o">=</span><span class="mi">100</span><span class="p">)</span>
2450        <span class="n">sizer</span> <span class="o">=</span> <span class="n">wx</span><span class="o">.</span><span class="n">BoxSizer</span><span class="p">(</span><span class="n">wx</span><span class="o">.</span><span class="n">VERTICAL</span><span class="p">)</span>
2451        <span class="n">sizer</span><span class="o">.</span><span class="n">Add</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="n">wx</span><span class="o">.</span><span class="n">LEFT</span> <span class="o">|</span> <span class="n">wx</span><span class="o">.</span><span class="n">EXPAND</span><span class="p">)</span>
2452        <span class="n">sizer</span><span class="o">.</span><span class="n">Add</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">gauge</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="n">wx</span><span class="o">.</span><span class="n">LEFT</span> <span class="o">|</span> <span class="n">wx</span><span class="o">.</span><span class="n">EXPAND</span><span class="p">)</span>
2453        <span class="bp">self</span><span class="o">.</span><span class="n">SetSizer</span><span class="p">(</span><span class="n">sizer</span><span class="p">)</span>
2454        <span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="s">&#39;starting value&#39;</span><span class="p">)</span>
2455    <span class="k">def</span> <span class="nf">onMonitorMessage</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">message</span><span class="p">):</span>
2456        <span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="nb">str</span><span class="p">(</span><span class="n">message</span><span class="p">))</span>
2457    <span class="k">def</span> <span class="nf">onMonitorStart</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">message</span><span class="p">):</span>
2458        <span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="nb">str</span><span class="p">(</span><span class="n">message</span><span class="p">))</span>
2459        <span class="bp">self</span><span class="o">.</span><span class="n">gauge</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span>
2460    <span class="k">def</span> <span class="nf">onMonitorProgress</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">message</span><span class="p">):</span>
2461        <span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="nb">str</span><span class="p">(</span><span class="n">message</span><span class="p">))</span>
2462        <span class="bp">self</span><span class="o">.</span><span class="n">gauge</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="nb">int</span><span class="p">(</span><span class="mi">100</span><span class="o">*</span><span class="n">message</span><span class="o">.</span><span class="n">complete</span><span class="o">/</span><span class="n">message</span><span class="o">.</span><span class="n">total</span><span class="p">))</span>
2463    <span class="k">def</span> <span class="nf">onMonitorComplete</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">message</span><span class="p">):</span>
2464        <span class="bp">self</span><span class="o">.</span><span class="n">text</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="nb">str</span><span class="p">(</span><span class="n">message</span><span class="p">))</span>
2465        <span class="bp">self</span><span class="o">.</span><span class="n">gauge</span><span class="o">.</span><span class="n">SetValue</span><span class="p">(</span><span class="mi">100</span><span class="p">)</span>
2466</pre></div>
2467</div>
2468<p>We can put this panel in a simple app:</p>
2469<div class="highlight-python"><div class="highlight"><pre><span class="n">app</span> <span class="o">=</span> <span class="n">wx</span><span class="o">.</span><span class="n">PySimpleApp</span><span class="p">()</span>
2470<span class="n">frame</span> <span class="o">=</span> <span class="n">wx</span><span class="o">.</span><span class="n">Frame</span><span class="p">(</span><span class="bp">None</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="s">&#39;Test Monitor&#39;</span><span class="p">)</span>
2471<span class="n">panel</span> <span class="o">=</span> <span class="n">Panel</span><span class="p">(</span><span class="n">frame</span><span class="p">)</span>
2472<span class="n">frame</span><span class="o">.</span><span class="n">Show</span><span class="p">()</span>
2473</pre></div>
2474</div>
2475<p>Next we attach attach the monitor to this panel and feed some messages from
2476another thread:</p>
2477<div class="highlight-python"><div class="highlight"><pre>import time,thread
2478import park.wxmonitor, park.monitor
2479from park.monitor import Start, Progress, Improvement, Complete
2480monitor = park.wxmonitor.wxMonitor(panel)
2481msgs = [Start(), Progress(1,10), Progress(3,10),
2482        Improvement(&#39;Better!&#39;), Progerss(6,10), Complete(&#39;Best!&#39;)]:
2483def message_stream(monitor,msgs):
2484    time.sleep(1)
2485    for message in msgs:
2486      monitor.put(message)
2487      time.sleep(1)
2488thread.start_new_thread(message_stream, (monitor,msgs))
2489app.MainLoop()
2490</pre></div>
2491</div>
2492<p>You should see the progress bar jump from 10% to 30% to 60% then all the way
2493to the end.</p>
2494<dl class="function">
2495<dt id="park.wxmonitor.demo">
2496<tt class="descclassname">park.wxmonitor.</tt><tt class="descname">demo</tt><big>(</big><em>rate=0</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#demo"><span class="viewcode-link">[source]</span></a></dt>
2497<dd></dd></dl>
2498
2499<dl class="class">
2500<dt id="park.wxmonitor.wxMonitor">
2501<em class="property">class </em><tt class="descclassname">park.wxmonitor.</tt><tt class="descname">wxMonitor</tt><big>(</big><em>win</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#wxMonitor"><span class="viewcode-link">[source]</span></a></dt>
2502<dd><p>Bases: <a class="reference internal" href="#park.monitor.Monitor" title="park.monitor.Monitor"><tt class="xref py py-class docutils literal"><span class="pre">park.monitor.Monitor</span></tt></a></p>
2503<p>Attach a job monitor to a panel.</p>
2504<p>The monitor will perform callbacks to onMonitorStart(message),
2505onMonitorProgress(message), etc. if the associated method is
2506defined.  If the type specific method is not defined, then the
2507monitor will call onMonitorMessage(message).  Otherwise the
2508message is dropped.</p>
2509<p>See <cite>park.monitor</cite> for a description of the usual messages.</p>
2510<dl class="method">
2511<dt id="park.wxmonitor.wxMonitor.dispatch">
2512<tt class="descname">dispatch</tt><big>(</big><em>event</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#wxMonitor.dispatch"><span class="viewcode-link">[source]</span></a></dt>
2513<dd><p>Dispatch the event from the asynchronous monitor.  This is running
2514in the GUI thread.</p>
2515</dd></dl>
2516
2517<dl class="method">
2518<dt id="park.wxmonitor.wxMonitor.onMonitorLog">
2519<tt class="descname">onMonitorLog</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#wxMonitor.onMonitorLog"><span class="viewcode-link">[source]</span></a></dt>
2520<dd><p>Called when the job sends a logging record.</p>
2521<p>The logging record contains a normal python logging record.</p>
2522<p>The default behaviour is to tie into the application logging
2523system using:</p>
2524<div class="highlight-python"><div class="highlight"><pre><span class="n">logger</span> <span class="o">=</span> <span class="n">logging</span><span class="o">.</span><span class="n">getLogger</span><span class="p">(</span><span class="n">message</span><span class="o">.</span><span class="n">record</span><span class="o">.</span><span class="n">name</span><span class="p">)</span>
2525<span class="n">logger</span><span class="o">.</span><span class="n">handle</span><span class="p">(</span><span class="n">message</span><span class="o">.</span><span class="n">record</span><span class="p">)</span>
2526</pre></div>
2527</div>
2528<p>Logging levels are set in the job controller.</p>
2529</dd></dl>
2530
2531<dl class="method">
2532<dt id="park.wxmonitor.wxMonitor.onMonitorMessage">
2533<tt class="descname">onMonitorMessage</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#wxMonitor.onMonitorMessage"><span class="viewcode-link">[source]</span></a></dt>
2534<dd><p>Generic message handler: do nothing.</p>
2535</dd></dl>
2536
2537<dl class="method">
2538<dt id="park.wxmonitor.wxMonitor.put">
2539<tt class="descname">put</tt><big>(</big><em>message</em><big>)</big><a class="reference internal" href="../../_modules/park/wxmonitor.html#wxMonitor.put"><span class="viewcode-link">[source]</span></a></dt>
2540<dd><p>Intercept an event received from an asynchronous monitor.  This is
2541running in the asynchronous thread.</p>
2542</dd></dl>
2543
2544</dd></dl>
2545
2546</div>
2547</div>
2548<div class="section" id="module-park">
2549<span id="module-contents"></span><h2>Module contents</h2>
2550<p>PARK fitting service.</p>
2551<p>The PARK fitting service is a set of python packages to support
2552fitting in datasets.  Using common python infrastructure such
2553as scipy optimizers, numpy arrays, and matplotlib plotting, park
2554allows you to define models, associate them with datasets and
2555simultaneously fit them.  Park provides a simple job queue to
2556manage multiple fits using separate processes or running across
2557the network on separate machines.</p>
2558<div class="section" id="installation">
2559<h3>Installation</h3>
2560<p>The latest version of Park is available from
2561<a class="reference external" href="http://www.reflectometry.org/danse/park">http://www.reflectometry.org/danse/park</a>.</p>
2562<p>Currently this is supplied as source from a zip file.  You can
2563also retrieve the latest version from svn:</p>
2564<div class="highlight-python"><div class="highlight"><pre>svn co svn://danse.us/park/branches/park-1.2
2565</pre></div>
2566</div>
2567<p>If you are installing from source, you will need a python
2568environment with numpy and scipy.  For the GUI version, you
2569will also need matplotlib and wx.</p>
2570<p>You will need a C compiler to build the resolution convolution
2571function.  On Windows this may require installing MinGW and
2572adding distutils.cfg to your distutils directory:</p>
2573<div class="highlight-python"><div class="highlight"><pre><span class="p">[</span><span class="n">build</span><span class="p">]</span>
2574<span class="n">compiler</span> <span class="o">=</span> <span class="n">mingw</span>
2575</pre></div>
2576</div>
2577<p>Once you have the required supporting packages, use the following
2578to build and install:</p>
2579<div class="highlight-python"><div class="highlight"><pre>python setup.py install
2580</pre></div>
2581</div>
2582</div>
2583<div class="section" id="id3">
2584<h3>Usage</h3>
2585<p>To get started with park, you will need to first define the
2586models that you are using.  These can be very basic models,
2587listing all possible fitting parameters.  When setting up the
2588fit later, you will be able to combine models, using
2589parameter expressions to relate the values in one model with
2590those in another.  See <cite>park.model</cite> for details.</p>
2591<p>Once your models are constructed you can use them in a fit.
2592See <cite>park.fit</cite> for details.</p>
2593<p>Important classes and functions:
2594<cite>park.model.Model</cite>, <cite>park.parameter.ParameterSet</cite>, <cite>park.fit.Fit</cite></p>
2595<table class="docutils field-list" frame="void" rules="none">
2596<col class="field-name" />
2597<col class="field-body" />
2598<tbody valign="top">
2599<tr class="field-odd field"><th class="field-name">group models:</th><td class="field-body">model, assembly, parameter, data</td>
2600</tr>
2601<tr class="field-even field"><th class="field-name">group examples:</th><td class="field-body">peaks</td>
2602</tr>
2603<tr class="field-odd field"><th class="field-name" colspan="2">group optimizer:</th></tr>
2604<tr class="field-odd field"><td>&nbsp;</td><td class="field-body">fit, fitresult, fitmc, simplex</td>
2605</tr>
2606<tr class="field-even field"><th class="field-name">group support:</th><td class="field-body">expression, deps, version, setup, serial</td>
2607</tr>
2608<tr class="field-odd field"><th class="field-name">group server:</th><td class="field-body">fitservice</td>
2609</tr>
2610</tbody>
2611</table>
2612</div>
2613</div>
2614</div>
2615
2616
2617          </div>
2618        </div>
2619      </div>
2620      <div class="sphinxsidebar">
2621        <div class="sphinxsidebarwrapper">
2622  <h3><a href="../../index.html">Table Of Contents</a></h3>
2623  <ul>
2624<li><a class="reference internal" href="#">park package</a><ul>
2625<li><a class="reference internal" href="#submodules">Submodules</a></li>
2626<li><a class="reference internal" href="#module-park.assembly">park.assembly module</a><ul>
2627<li><a class="reference internal" href="#usage">Usage</a><ul>
2628<li><a class="reference internal" href="#a-note-on-model-weighting">A note on model weighting</a></li>
2629</ul>
2630</li>
2631</ul>
2632</li>
2633<li><a class="reference internal" href="#module-park.data">park.data module</a></li>
2634<li><a class="reference internal" href="#module-park.deps">park.deps module</a></li>
2635<li><a class="reference internal" href="#module-park.diffev">park.diffev module</a></li>
2636<li><a class="reference internal" href="#module-park.expression">park.expression module</a></li>
2637<li><a class="reference internal" href="#module-park.fit">park.fit module</a><ul>
2638<li><a class="reference internal" href="#id1">Usage</a></li>
2639<li><a class="reference internal" href="#gui-usage">GUI Usage</a></li>
2640</ul>
2641</li>
2642<li><a class="reference internal" href="#park-fitmc-module">park.fitmc module</a></li>
2643<li><a class="reference internal" href="#module-park.fitresult">park.fitresult module</a></li>
2644<li><a class="reference internal" href="#park-fitservice-module">park.fitservice module</a></li>
2645<li><a class="reference internal" href="#module-park.formatnum">park.formatnum module</a></li>
2646<li><a class="reference internal" href="#module-park.message">park.message module</a></li>
2647<li><a class="reference internal" href="#module-park.model">park.model module</a><ul>
2648<li><a class="reference internal" href="#id2">Usage</a></li>
2649</ul>
2650</li>
2651<li><a class="reference internal" href="#module-park.monitor">park.monitor module</a></li>
2652<li><a class="reference internal" href="#module-park.parameter">park.parameter module</a></li>
2653<li><a class="reference internal" href="#module-park.parkde">park.parkde module</a></li>
2654<li><a class="reference internal" href="#park-parksnob-module">park.parksnob module</a></li>
2655<li><a class="reference internal" href="#module-park.peaks">park.peaks module</a></li>
2656<li><a class="reference internal" href="#module-park.pmap">park.pmap module</a></li>
2657<li><a class="reference internal" href="#park-rangemap-module">park.rangemap module</a></li>
2658<li><a class="reference internal" href="#park-serial-module">park.serial module</a></li>
2659<li><a class="reference internal" href="#module-park.setup">park.setup module</a></li>
2660<li><a class="reference internal" href="#module-park.simplex">park.simplex module</a></li>
2661<li><a class="reference internal" href="#module-park.version">park.version module</a></li>
2662<li><a class="reference internal" href="#module-park.wxmonitor">park.wxmonitor module</a><ul>
2663<li><a class="reference internal" href="#example">Example</a></li>
2664</ul>
2665</li>
2666<li><a class="reference internal" href="#module-park">Module contents</a><ul>
2667<li><a class="reference internal" href="#installation">Installation</a></li>
2668<li><a class="reference internal" href="#id3">Usage</a></li>
2669</ul>
2670</li>
2671</ul>
2672</li>
2673</ul>
2674
2675  <h4>Previous topic</h4>
2676  <p class="topless"><a href="modules.html"
2677                        title="previous chapter">lib.macosx-10.5-x86_64-2.7</a></p>
2678  <h4>Next topic</h4>
2679  <p class="topless"><a href="sas.html"
2680                        title="next chapter">sas package</a></p>
2681  <h3>This Page</h3>
2682  <ul class="this-page-menu">
2683    <li><a href="../../_sources/dev/api/park.txt"
2684           rel="nofollow">Show Source</a></li>
2685  </ul>
2686<div id="searchbox" style="display: none">
2687  <h3>Quick search</h3>
2688    <form class="search" action="../../search.html" method="get">
2689      <input type="text" name="q" />
2690      <input type="submit" value="Go" />
2691      <input type="hidden" name="check_keywords" value="yes" />
2692      <input type="hidden" name="area" value="default" />
2693    </form>
2694    <p class="searchtip" style="font-size: 90%">
2695    Enter search terms or a module, class or function name.
2696    </p>
2697</div>
2698<script type="text/javascript">$('#searchbox').show(0);</script>
2699        </div>
2700      </div>
2701      <div class="clearer"></div>
2702    </div>
2703    <div class="related">
2704      <h3>Navigation</h3>
2705      <ul>
2706        <li class="right" style="margin-right: 10px">
2707          <a href="../../genindex.html" title="General Index"
2708             >index</a></li>
2709        <li class="right" >
2710          <a href="../../py-modindex.html" title="Python Module Index"
2711             >modules</a> |</li>
2712        <li class="right" >
2713          <a href="sas.html" title="sas package"
2714             >next</a> |</li>
2715        <li class="right" >
2716          <a href="modules.html" title="lib.macosx-10.5-x86_64-2.7"
2717             >previous</a> |</li>
2718        <li><a href="../../index.html">SasView 3.0.0 documentation</a> &raquo;</li>
2719          <li><a href="../dev.html" >Developer Documentation</a> &raquo;</li>
2720          <li><a href="modules.html" >lib.macosx-10.5-x86_64-2.7</a> &raquo;</li> 
2721      </ul>
2722    </div>
2723    <div class="footer">
2724        &copy; Copyright 2013, The SasView Project.
2725      Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.3.
2726    </div>
2727  </body>
2728</html>
Note: See TracBrowser for help on using the repository browser.