Analysis¶

Fitting adjusts the parameters you choose to bring the calculated result closer to the measured data. Begin with a structural model and experiment that plausibly describe the measurement; fitting is not a substitute for checking those choices.

The shared Analysis chapter explains the scientific model; the steps below describe this interface.

Load or prepare the project¶

Open an existing project or an included example on Project. To analyse your own measurement, follow create a project.

On Structure, check symmetry, cell and atom sites. On Experiment, check the data, instrument, peak profile, background, excluded regions and linked structures.

Compare the starting pattern¶

Look at peak positions, widths, intensities and the background before fitting. An incorrect wavelength or calibration can shift peaks; a missing phase can leave intensity the model cannot explain.

Choose free parameters¶

On Analysis, the parameter table lists every numerical input of the structures and experiments. Use the free column to choose which parameters may vary. When a fit cannot vary a value, its checkbox is disabled and hovering over it shows the reason. The name and variability filters help you find them.

A value cannot be fitted when it describes the setup, symmetry or a constraint sets it, the fit's data cannot determine it, or the calculation has no derivative for it yet. A value saved as free that cannot be fitted opens as fixed, and Messages says so. The minimum and maximum displayed in the table describe admissible model values; they are not a separate set of user-defined optimizer bounds.

Free parameters deliberately, checking the effect of each stage. Starting with every parameter free can hide an unsuitable model behind strongly correlated values.

Select a fitting mode¶

The available mode depends on the project. One experiment is fitted on its own; several experiments can be fitted jointly. A scan can run sequentially or independently across its datasets. See fitting modes.

Start and inspect the fit¶

Press Fit at the right of the status bar. Review the calculated curve and residual as well as the reported agreement factors. Parameter uncertainties and correlations help you understand what the data constrain.

A small agreement factor alone does not establish that the structural model is correct. Inspect unmodelled features and physically implausible values before accepting the result.

Inspect parameter correlations¶

Open Correlations in the Analysis main view after fitting. Inspect strongly related free parameters alongside their uncertainties and the residual. A correlation can reveal that the data do not determine two values independently; see covariance and correlations.

Read recorded fit statistics¶

Open a fitted project and select Analysis → Extra to inspect its recorded results. Fit result holds the overall outcome and statistics; Experiment fit result separates the measurements; Structure fit result reports the structure's reflection agreement for each experiment.

These groups are read-only. Assess the agreement factors alongside the pattern, uncertainties and parameter correlations.

Save the project and report¶

Save the updated project if you want to keep working from its fitted values. Use the Report page to prepare a separate document containing the selected model, results and figures.

The same project can be opened in Python or run with the Command line.