Table 1. Selected program parameters and their effects on the computation.
For the advanced settings and defaults that can be found in the mcsasparameters.json file, only selected values are listed.
| Location | Parameter name | Effect |
|---|---|---|
| GUI Algorithm panel | Convergence criterion | The least-squares value ( ) at which the fit is considered a success. For data with good uncertainty estimates, this can be set to 1. For a quick fit, it can be set to larger values. Values below 1 are not recommended. |
| Number of repetitions | The number of independent optimizations to be run. Larger values will result in improved uncertainty estimates on the result (and a slightly smoother result), but calculation time increases proportionally. | |
| Number of contributions | The number of individual contributions whose weighted sum comprises the total model intensity. Too few or too many will result in slow optimization times. Most patterns can be fitted using 300 contributions quickly, but times can be optimized using the timing information shown in the result. | |
| Find background level | If selected, a flat background is fitted during matching of model and data. This speeds up the fit with minimal effect on the result, as many scattering patterns contain a flat scattering component as well (due to density variations or incoherent scattering). | |
| GUI Post-fit Analysis panel | Parameter | The parameter to show the distribution of. |
| Lower upper | The distribution will be shown in this parameter range only. This can be used to cut off regions outside the range of interest. Population statistics also apply only to this range. | |
| Number of bins | The number of divisions to use in the distribution display. By increasing this number, more detail may be visible provided one stays within the Shannon channel limit (indicated in the ‘Data Files’ panel). An increase in the number of divisions will also negatively affect uncertainty estimates and observability limits. | |
| X-axis scaling | Scaling (linear or logarithmic) for the parameter axis of the distribution. Logarithmic recommended for wide parameter ranges. | |
| Y-axis weighting | The vertical axis can be shown in volume or number distributions. Volume-weighted distributions recommended; number-weighted distributions can be used for samples with a narrow dispersity. | |
| mcsasparameters.json (file, advanced settings and defaults) | maxIterations | If convergence has not been reached within this number of iterations, the optimization attempt is aborted. Larger values may allow complex calculations to finish successfully, but often nonconvergence can be traced back to poor initialization settings. Increasing this value increases the maximum possible calculation time. |
| compensationExponent | Adjusts internal weighting of scattering pattern contributions. Adjustment between 0.3 and 0.7 may lead to slight speed increases for some samples. | |
| eMin | Minimum uncertainty estimate in fraction of intensity. Default 0.01 sets the uncertainty value to be no less than 1% of the data intensity value. Can be increased or reduced based on best guess estimate for minimum inter-related data point uncertainty. A too low value may prevent reaching convergence. | |