# Changeset 4668

Ignore:
Timestamp:
Dec 2, 2020 11:39:06 AM (22 months ago)
Message:

updates to refinement options

File:
1 edited

Unmodified
Removed
• ## trunk/help/gsasII.html

 r4664

This window provides the main controls and a few global parameters for GSAS-II.

What can I do here?

On this page, there are three or four sets of controls. The first is for how refinements operate.

"Times New Roman"'>This window provides access to the controls that determine how GSAS-II performs minimizations as well as few global parameters for GSAS-II. Note that many other customization settings are set as configuration variables in the Preferences menu. (See the Programmer's documentation for a description of those.)

Refinement Controls: These controls determine how refinements are performed. The first determines the computational engine used to minimize the structure.

• analytic Hessian: This is the default option and is usually the most useful. It uses as custom-developed least-squares the most useful. It uses a custom-developed least-squares minimizer that uses singular-value decomposition (SVD) to reduce the errors caused by correlated variables and the Levenberg-Marquardt algorithm to down-weight the off-diagonal Levenberg-Marquardt algorithm to up-weight diagonal Hessian terms when refinements fail to lower χ2.
• analytic Jacobian: This uses a numpy-provided leastsq minimizer, which not applicable for larger problems as it requires minimizer, which not applicable for problem with a large number of histograms as it requires much more memory than the Hessian routines. This because it creates a Jacobian matrix is shaped N x M (N parameters x M observations) and uses that to create the N x N Hessian. The "Hessian" minimizers create the Hessian matrix directly. creates a Jacobian matrix that is shaped N x M (N parameters x M observations) while the Hessian methods create a Jacobian matrix only each histogram.
• numeric: This also uses the numpy leastsq minimizer, and is also not applicable for larger problems. Unlike, the "analytic Jacobian", numerical derivates are computed for derivatives rather than analytical derivatives problems. Unlike, the "analytic Jacobian", numerical derivatives are computed rather than use the analytical derivatives that are coded directly into GSAS-II. This will be slower than the analytical derivatives and will converge more slowly. It is the analytical derivatives and will is often less accurate which results in slower convergence. It is typically used for code development to check the accuracy of the analytical derivative formulations. Hessian but does not include the Levenberg-Marquardt algorithm. It can be faster, but is more prone to diverge when severe correlation is present. diverge when severe correlation is present. It is possible that it might be better for single-crystal refinements.

Min delta-M/M
This determines when convergence is recognized; the refinement will stop when the change in the minimization function is less than this value. Set Min delta-M/M = 1.0 to force just a single cycle to be performed. The default is 0.001. A value less than 10-4 causes the refinement to cycle with no meaningful improvement since shifts become a small fraction of the parameter's uncertainties. Set Min delta-M/M = 1.0 to force refinement to stop after a single refinement. The allowed range is 10-9 to 1.0.
A refinement will stop when the change in the minimization function (M=Σ[w(Io-Ic)2]) is less than this value.  The allowed range is 10-9 to 1.0, with a default of 0.001. A value of 1.0 stops the refinement after a single cycle. Values less than 10-4 cause refinements to continue even if there is no meaningful improvement.

Max cycles
This determines the maximum number of refinement cycles that will be performed. This is only available with the "Hessian" minimizers.
will be performed. This is only available with the "Hessian" minimizers.

Initial lambda
Note that here λ is the Marquardt coefficient, which when large down-weights the significance of the off-diagonal terms in the
Note that here λ is the Marquardt coefficient, where a weight of 1+λ is applied to the diagonal elements of the Hessian. When λ is large, this down-weights the significance of the off-diagonal terms in the Hessian. Thus, when λ is large, the refinement is effectively one of steepest-descents, where correlation between variables is slow and may not always find the local minimum. This is only available with the "analytical Hessian" minimizer.

SVD zero tolerance
This determines the level where SVD considers values to be the same. Default is 10-6. Make larger to where problems occur due to correlation. This is only available with the "Hessian" minimizers.

Initial shift factor
?
A “damping multiplier” applied during the first refinement cycle, for Jacobean/numeric refinements only. Should be in interval (0.1, 100). See the SciPy leastsq docs for more information.

A set of controls is provided for

Single Crystal: A set of controls is provided for control of single-crystal refinements. These only appear when single crystal (HKLF) histograms are present in the project. in the project.

When checked, refinements are against F2 rather than |F|.

Min obs/sig
?
Conventional cutoff for single crystal refinements as to what reflections should be considered observed, typical values are 2.0 (2σ) or 3.0 (3σ).

Min extinct.
?
(needs further work)

Max delt-F/sig
?
Removes reflections that are very poorly fit. Should be used only with extreme care, since poorly-fit reflections could be an indication that the structure is wrong.

Max d-spacing
Reflections with d-space values larger than this value are ignored.

Min d-spacing
Reflections with d-space values smaller than this value are ignored.

A set of controls is for sequential refinement. Settings here determine if "normal" or "sequential" refinement is performed. If no datasets are selected, then all "used" histograms are included in one combined refinement. However, if any number histogram are selected used here, then a sequential refinement is performed, where a fit is made to each histogram in turn. Only the first item below is shown in "normal" mode.

Sequential Settings: A set of controls is for sequential refinement. Settings here determine if a "normal" or "sequential" refinement is performed. If no datasets are selected here, then all histograms linked to phases in the project and that are flagged as "used" are included in one potentially large (combined) refinement. However, if any number of histograms are selected here, then a sequential refinement is performed, where a fit is made to the structural model(s) fitting each selected histogram in turn. Only the first item below is shown in "normal" mode.

set as "normal". The button is labeled "Select" when in normal refinement mode and "Reselect" in sequential refinement mode.

Reverse order?
Normally, in a sequential histograms are fit in the order they but when this option is selected, the sequential fit is performed with the last tree entry first.

Copy results to next histogram?
When this option is selected, the fitted parameters from each refinement is completed. For subsequent refinements, it is usually better to start with the results from the previous fit.

Clear previous seq. results
When this button is pressed, the "Sequential Results" entry with the results from the last sequential fit is deleted from the tree.

Global Settings: This is a location for parameters that apply to an entire project. At present there is only one.

CIF Author
The value provided here is used when creating a CIF of an entire project.

What can I do here?

This offers a place to change how GSAS-II performs refinements, but has no specific menu commands or graphics.

Last modified: Sun Nov 29 20:05:28 CST 2020 Last modified: Wed Dec  2 11:04:42 CST 2020

Note: See TracChangeset for help on using the changeset viewer.