source: trunk/doc/expgui1.html @ 930

Last change on this file since 930 was 930, checked in by toby, 11 years ago

rcs:* properties removed

  • Property svn:keywords set to Author Date Revision Id
File size: 8.6 KB
Line 
1<html>
2<head>
3   <title>EXPGUI</title>
4   <meta name="keywords" content="crystallography, Rietveld, diffraction,
5   GSAS, EXPGUI">
6</HEAD>
7<style>
8A:link {text-decoration:none}
9A:vlink {text-decoration:none}
10</style>
11
12<BODY BGCOLOR="#FFFFFF"
13      topmargin="0" leftmargin="0" marginwidth="0" marginheight="0" 
14      text="#000000" link="#0033ff" vlink="#0033ff" alink="#0033ff">
15
16<?
17   include("/var/www/include/navigation.inc");
18   include("/var/www/include/utility.inc");
19?>
20<blockquote><font face="arial, helvetica, sans-serif">
21
22<TABLE BORDER BGCOLOR="#FFFF40" ALIGN=RIGHT>
23<TR><TH><A  Href="expgui.html">EXPGUI top</A> 
24</TH><TH><A Href="expgui2.html">Next page</A>
25</TH></TR></TABLE><BR CLEAR=ALL>
26
27<center><h1>
28<HR noshade width="75%" size="2" align="center">
29EXPGUI, part 1
30<HR noshade width="75%" size="2" align="center">
31</h1></center>
32<P>
33<h3>A.1 Least Squares (LS) Controls Panel</h3>
34<DL><DL>
35<p>The LS Controls panel shows information about the
36current experiment, typically found in the EXPEDT "Least
37Squares Controls" options.
38<P>
39Note that the order that histograms appear in this
40panel is determined by the
41<a href="expguic.html#sorthist">"Sort histograms by"</a> option in
42the Options Menu.
43</DL>
44<img SRC="1.gif" align=TEXTTOP alt="EXPGUI Screen snapshot">
45</DL>
46<DL><DL>
47The entries on the upper part of this panel are overall options for the
48entire experiment.
49<P>
50<DT><B>Last History</B><DD>
51This shows the last history record written into
52the experiment file, showing the last program that modified the file
53and when it was run.
54<DT><B>Title</B><DD>
55This is a title for the refinement. users can specify any information
56they want saved in the experiment file.
57<DT><B>Number of Cycles</B><DD>
58This is the number of refinement cycles to be performed in GENLES.
59If this number is zero when GENLES is run,
60powder diffraction intensities are computed and, when requested
61(<a href="#extract">see below</a>) reflection intensities are estimated
62but parameters are not refined. Note that when a
63<a href="#lebail">LeBail extraction</a> is performed
64with the cycles set at zero, reflection
65intensities are optimized even when though no cycles of refinement are
66performed.
67<DT><B>Print Options</B><DD>
68<img SRC="1a.gif" align=right alt="EXPGUI Screen snapshot">
69This allows you to control what types of output GENLES provides. The menu of
70options is shown to the right. I recommend including the summary of shifts and
71in most cases the correlation matrix in the output.
72<DT><B>Convergence Criterion</B><DD>
73GENLES stops refinements when the sum of the squares of each
74parameter shifts divided by its standard uncertainty is less than
75this "Convergence Criterion." Since this quantity is the <I>total</I>
76sum of squares, it is reasonable to raise this value for refinements
77where large numbers of parameters will be refined.
78<DT><B>Marquardt Damping</B><DD>
79Marquardt damping increases the weighting of the diagonal elements
80in the Hessian matrix, reducing the impact of parameter correlation
81on the refinement. It increases refinement stability at the cost
82of requiring additional cycles of refinement. A Marquardt term of 1.0
83corresponds to a standard least-squares refinement with no Marquardt
84damping. The value 1.2 has been recommended to me by Lachlan Cranswick as
85a good choice.
86<P>
87</DL>
88<br clear=all>
89<DL>
90The lower section, labeled "Reflection Intensity Extraction" has options
91for each histogram that determine if reflection intensities will be estimated,
92and if so, how.
93<P>
94<DT><B>Extract Fobs</B><DD>
95When the Extract Fobs option is on, reflection intensities are computed
96using the method developed by Hugo Rietveld. In this method
97the intensity for each reflection is determined by summing the
98appropriate data points, weighed by the ratio of the computed intensity
99from that reflection to the total computed intensity at that point. This means
100that in the case of severely overlapped reflections, "observed"
101intensities are apportioned according to the relative computed reflection
102intensities. This is clearly biased since it invokes the crystallographic
103model, but is about the best that can be done. Turning this option off
104saves a very small amount of computer time.
105<a name="extract">
106</a><DT><B>Intensity Extraction Methods</B><DD>
107There are two approaches to reflection intensity determination. In the
108conventional <B>Rietveld</B> approach, if the "Extract Fobs" flag is on,
109reflection intensities are determined
110as part of the Rietveld refinement, reflection R-factors are
111computed, and the reflection intensities
112are saved on disk file for use in Fourier or other computations.
113<P>
114In the extraction method developed by Armel <B>LeBail</B>,
115reflection intensities are "optimized" by treating the setting the F<sub>calc</sub> value
116for each reflection to the F<sub>obs</sub> value extracted
117during the previous cycle.
118By iterating, the F<sub>calc</sub> values slowly converge to a
119set of reflection
120intensities that yields a best fit to the pattern.
121The F<sub>obs</sub> values are determined every time GENLES is run,
122or a least squares
123refinement cycle is run. This it is possible to improve the LeBail fit, by
124running GENLES with the "Number of Cycles" set to zero.
125<P>
126Note that due to reflection
127overlap, there are usually many different ways to apportion intensities with 
128fits of comparable quality, depending on what starting values are used for
129F<sub>obs</sub>. Any time POWPREF is run, the reflection list
130is regenerated and the first time that GENLES is run, the
131starting F<sub>calc</sub> values are set one of two ways:
132<P><DL>
133<a name="lebail">
134<DT><B>F(calc) weighted</B><DD>
135In a "F(calc) weighted" LeBail extraction the initial F<sub>calc</sub> values are computed
136from the crystal structure model. If the model is fairly close to being
137correct, it will likely apportion intensity for overlapping reflections in
138a manner that is fairly close to correct. Thus, the F<sub>calc</sub> values obtained
139from a "F(calc) weighted" LeBail extraction are about as good as can be
140done for the case where the structure is pretty close to correct.
141<DT><B>Equally weighted</B><DD>
142On the other hand, if one has no good structural model, but would like to
143use LeBail extraction as a way to obtain F<sub>obs</sub> values for use in structure
144solution, for example, by direct methods, then it is best to assume that all
145reflections are equally likely to contain intensity. In the "Equally weighted"
146mode, all reflections are given an identical F<sub>obs</sub> starting value. Thus, if
147two reflections are completely overlapped, in this methods, they will
148be assigned equal F<sub>obs</sub> values through the LeBail fit.
149</DL>
150<P>
151It is possible to refine unit cell, background, profile and other
152non-structural parameters at the same time as a LeBail extraction is
153performed. I often do this, for two reasons. One is that the final LeBail
154R<sub>wp</sub> and Chi<sup>2</sup> provides a better measure of the
155best possible fit
156than the statistical values, particularly if the material has non-ideal
157peak shapes or other factors that cannot modeled. The second reason is the
158LeBail fit provides excellent starting values for the unit cell, background
159and profile parameters, so these terms need not be refined again
160until all structural terms have been fit well.
161<P>
162These LeBail refinements, alas, are prone to diverge
163if the the extracted intensities are changing rapidly and
164other parameters, such as unit cell parameters are compensating. It is
165a good practice to run GENLES several times with
166the "Number of Cycles" set to zero each time POWPREF is run -- to allow the
167reflection intensities to converge before refining parameters.
168<P>
169Note that extraction different methods can be used for different phases
170in a histogram.
171It can be convenient to use LeBail extraction for an impurity phase,
172in the case where the impurity has preferred orientation or has a known
173unit cell, but an unknown structure. I have also used LeBail fits to
174obtain precise lattice constants via profile fits of materials where the
175exact structure is not known, so a Rietveld refinement cannot be performed.
176<DT><B>LeBail Damping</B><DD>
177The shifts to the reflection intensities can damped. This is useful when
178refining lattice constants or other terms that might otherwise cause the
179reflection intensities to shift dramatically and in turn cause the refinement
180to diverge.
181</DL></DL>
182<hr>
183<TABLE BORDER BGCOLOR="#FFFF40" ALIGN=RIGHT>
184<TR><TH><A  Href="expgui.html">EXPGUI top</A> 
185</TH><TH><A Href="expgui2.html">Next page</A>
186</TH></TR></TABLE>
187
188<P><font size=-1><A HREF="MAILTO:crystal@NIST.gov?subject=WWW page <?=$PHP_SELF?>">Comments, corrections or questions: crystal@NIST.gov</A></font><BR>
189<font size=-1><? lastmod(); ?> </font>
190<br>$Revision: 930 $ $Date: 2009-12-04 23:14:35 +0000 (Fri, 04 Dec 2009) $
191</blockquote>
192</body>
193</html>
Note: See TracBrowser for help on using the repository browser.