← Algorithms · Parametric sphere mixture

Parametric sphere mixture

Purpose. When D(R) or I(q) suggests one or more spherical populations, a parametric mixture (mean radii, polydispersities, volume fractions) is a compact physical model — easier to interpret than a free D(R) alone, and useful for comparing samples.

Engine. ATSAS MIXTURE fits hard-sphere populations. Autosaxs decides which models to try and which fit wins.

What autosaxs adds (custom).

1. Runs a fixed grid over phase count (up to max_nph, default 3) and size-distribution families (Gauss / Schultz) for SPHERE phases.
2. Applies careful initialization (e.g. radius / polydispersity bounds; defaults for r_max often taken from fit_sizes when unset).
3. Selects the preferred model by minimizing BIC_log (Bayesian information criterion on the log-intensity fit), balancing goodness-of-fit against extra phases.
4. Writes comparison plots and a results table so you can inspect runners-up, not only the winner.

In liveview, enable MIXTURE in the polydisperse window, then use Start modeling and Confirm in the Distribution Modeling app (MIXTURE is not run by the liveview skill queue). After a run, click the Distribution plot in that app to open the advanced MIXTURE results viewer.

← D(R) · Geometric bodies →