An efficient method for smoothing indicator-dilution and other unimodal curves.

A mathematical function has been developed for approximating unimodal functions, particularly those which are non-Gaussian, skewed, and incomplete. It is useful as an alternative to cubic splines in smoothing noisy experimental data. Particular applications are to indicator-dilution curves and probability density functions of varied form. In its Fortran implementation, SMOEX, it is computationally inexpensive compared to standard cubic spline smoothing routines and requires less storage to preserve the smoothed function for retrieval or interpolation.