Subsections

• GRAPHICAL INPUT

 INPUT PARAMETERS

INPUT PARAMETERS allows an initial fit model to be defined. The fit model can be defined GRAPHICALLY or from a FILE, which has previously been saved using OUTPUT PARAMETERS. Selecting INPUT PARAMETERS re-initialises the fit model so any previously defined fit model parameters are lost (prior to this operation values may be saved with OUTPUT PARAMETERS).

GRAPHICAL INPUT

This is the method to define an initial fit model. First any masked-off elements are defined (See Section 17.8, Page ).

The fit model can be made up of a general polynomial background plus one or more 2-D peaks. The order of the polynomial in the X and Y-directions is entered by the keyboard, followed by the a number of points used to define initial values for the polynomial. The initial polynomial may be defined from a lower order, which is more stable. The higher coefficients of the polynomial are then initialised to zero. To define the initial coefficients it is necessary to click on a given number of data points e.g. $(n+1)\times (m+1)$ where $n$ and $m$ are the orders of the initialisation polynomial. It is recommended to use only lower order polynomials, since higher order polynomials are not very stable. The user should click on the given number of data points spread throughout the data region, in an approximate grid. (If the points are all co-linear it will not be possible to set initial coefficient values.)

A choice of three peak types is available. A menu appears presenting the three types. the user should click on the required peak type, or on exit to finish this stage of the model definition. The three peak types are:

GAUSSIAN: 2-D Gaussian profile peak

POLAR GAUSSIAN 2-D Gaussian profile peak defined in a polar geometry i.e. Gaussian profile in a radial sense and in an angular arc

TWIN POLAR GAUSSIAN 2 polar Gaussians defined symmetrically about a centre of symmetry

For each peak a number of peak parameters need to be defined. These are all entered graphically by clicking on the centre of the peak (position and maximum intensity), and two half height positions. Care should be taken with the initialisation or greater danger of falling into local minima will be present.

Finally the user may specify a ``row-line'' as part of the model. A ``row-line'' is a line of polar Gaussians which all share the same radial and angular standard deviations. The peaks are placed at equal angles along a line at an angle from the vector from the symmetry centre to the ``row-line'' centre. The ``row-line'' is reflected in the vector from the symmetry centre to the row-line centre, but with a change in intensity. All the parameters which define a ``row-line'' are entered with the keyboard. (It is planned to define a tri-clinic unit cell diffraction pattern peak model to replace the ``row-line'' model which is an approximation.)