Changeset 401 for trunk

Dec 4, 2009 5:05:33 PM (14 years ago)

# on 2001/06/29 18:20:12, toby did:
major expansion -- explain page contents

1 edited


  • trunk/doc/expgui1.html

    • Property rcs:date changed from 2000/10/18 00:11:17 to 2001/06/29 18:20:12
    • Property rcs:lines changed from +2 -2 to +137 -2
    • Property rcs:rev changed from 1.4 to 1.5
    r331 r401  
    4343<img SRC="1.gif" align=TEXTTOP alt="EXPGUI Screen snapshot">
     46The entries on the upper part of this pane are overall options for the
     47entire experiment.
     49<DT><B>Last History</B><DD>
     50This shows the last history record written into
     51the experiment file, showing the last program that modified the file
     52and when it was run.
     54This is a title for the refinement. users can specify any information
     55they want saved in the experiment file.
     56<DT><B>Number of Cycles</B><DD>
     57This is the number of refinement cycles to be performed in GENLES.
     58If this number is zero when GENLES is run,
     59powder diffraction intensities are computed and, when requested
     60(<a href="#extract">see below</a>) reflection intensities are estimated
     61but parameters are not refined. Note that when a
     62<a href="#lebail">LeBail extraction</a> is performed
     63with the cycles set at zero, reflection
     64intensities are optimized even when though no cycles of refinement are
     66<DT><B>Print Options</B><DD>
     67<img SRC="1a.gif" align=right alt="EXPGUI Screen snapshot">
     68This allows you to control what types of output GENLES provides. The menu of
     69options is shown to the right. I recommend including the summary of shifts and
     70in most cases the correlation matrix in the output.
     71<DT><B>Convergence Criterion</B><DD>
     72GENLES stops refinements when the sum of the squares of each
     73parameter shifts divided by its standard uncertainty is less than
     74this "Convergence Criterion." Since this quantity is the <I>total</I>
     75sum of squares, it is reasonable to raise this value for refinements
     76where large numbers of parameters will be refined.
     77<DT><B>Marquardt Damping</B><DD>
     78Marquardt damping increases the weighting of the diagonal elements
     79in the Hessian matrix, reducing the impact of parameter correlation
     80on the refinement. It increases refinement stability at the cost
     81of requiring additional cycles of refinement. A Marquardt term of 1.0
     82corresponds to a standard least-squares refinement with no Marquardt
     83damping. The value 1.2 has been recommended to me by Lachlan Cranswick as
     84a good choice.
     87<br clear=all>
     89The lower section, labeled "Reflection Intensity Extraction" has options
     90for each histogram that determine if reflection intensities will be estimated,
     91and if so, how.
     93<DT><B>Extract Fobs</B><DD>
     94When the Extract Fobs option is on, reflection intensities are computed
     95using the method developed by Hugo Rietveld. In this method
     96the intensity for each reflection is determined by summing the
     97appropriate data points, weighed by the ratio of the computed intensity
     98from that reflection to the total computed intensity at that point. This means
     99that in the case of severely overlapped reflections, "observed"
     100intensities are apportioned according to the relative computed reflection
     101intensities. This is clearly biased since it invokes the crystallographic
     102model, but is about the best that can be done. Turning this option off
     103saves a very small amount of computer time.
     104<a name="extract">
     105</a><DT><B>Intensity Extraction Methods</B><DD>
     106There are two approaches to reflection intensity determination. In the
     107conventional <B>Rietveld</B> approach, if the "Extract Fobs" flag is on,
     108reflection intensities are determined
     109as part of the Rietveld refinement, reflection R-factors are
     110computed, and the reflection intensities
     111are saved on disk file for use in Fourier or other computations.
     113In the extraction method developed by Armel <B>LeBail</B>,
     114reflection intensities are "optimized" by treating the setting the F<sub>calc</sub> value
     115for each reflection to the F<sub>obs</sub> value extracted
     116during the previous cycle.
     117By iterating, the F<sub>calc</sub> values slowly converge to a
     118set of reflection
     119intensities that yields a best fit to the pattern.
     120The F<sub>obs</sub> values are determined every time GENLES is run,
     121or a least squares
     122refinement cycle is run. This it is possible to improve the LeBail fit, by
     123running GENLES with the "Number of Cycles" set to zero.
     125Note that due to reflection
     126overlap, there are usually many different ways to apportion intensities with 
     127fits of comparable quality, depending on what starting values are used for
     128F<sub>obs</sub>. Any time POWPREF is run, the reflection list
     129is regenerated and the first time that GENLES is run, the
     130starting F<sub>calc</sub> values are set one of two ways:
     132<a name="lebail">
     133<DT><B>F(calc) weighted</B><DD>
     134In a "F(calc) weighted" LeBail extraction the initial F<sub>calc</sub> values are computed
     135from the crystal structure model. If the model is fairly close to being
     136correct, it will likely apportion intensity for overlapping reflections in
     137a manner that is fairly close to correct. Thus, the F<sub>calc</sub> values obtained
     138from a "F(calc) weighted" LeBail extraction are about as good as can be
     139done for the case where the structure is pretty close to correct.
     140<DT><B>Equally weighted</B><DD>
     141On the other hand, if one has no good structural model, but would like to
     142use LeBail extraction as a way to obtain F<sub>obs</sub> values for use in structure
     143solution, for example, by direct methods, then it is best to assume that all
     144reflections are equally likely to contain intensity. In the "Equally weighted"
     145mode, all reflections are given an identical F<sub>obs</sub> starting value. Thus, if
     146two reflections are completely overlapped, in this methods, they will
     147be assigned equal F<sub>obs</sub> values through the LeBail fit.
     150It is possible to refine unit cell, background, profile and other
     151non-structural parameters at the same time as a LeBail extraction is
     152performed. I often do this, for two reasons. One is that the final LeBail
     153R<sub>wp</sub> and Chi<sup>2</sup> provides a better measure of the
     154best possible fit
     155than the statistical values, particularly if the material has non-ideal
     156peak shapes or other factors that cannot modeled. The second reason is the
     157LeBail fit provides excellent starting values for the unit cell, background
     158and profile parameters, so these terms need not be refined again
     159until all structural terms have been fit well.
     161These LeBail refinements, alas, are prone to diverge
     162if the the extracted intensities are changing rapidly and
     163other parameters, such as unit cell parameters are compensating. It is
     164a good practice to run GENLES several times with
     165the "Number of Cycles" set to zero each time POWPREF is run -- to allow the
     166reflection intensities to converge before refining parameters.
     168Note that extraction different methods can be used for different phases
     169in a histogram.
     170It can be convenient to use LeBail extraction for an impurity phase,
     171in the case where the impurity has preferred orientation or has a known
     172unit cell, but an unknown structure. I have also used LeBail fits to
     173obtain precise lattice constants via profile fits of materials where the
     174exact structure is not known, so a Rietveld refinement cannot be performed.
     175<DT><B>LeBail Damping</B><DD>
     176The shifts to the reflection intensities can damped. This is useful when
     177refining lattice constants or other terms that might otherwise cause the
     178reflection intensities to shift dramatically and in turn cause the refinement
     179to diverge.
Note: See TracChangeset for help on using the changeset viewer.