Repository navigation
Release date: 30-09-2026
New features
- Add the continuous plateau to
gt_phase_approx: with
trunc_frac=Noneand an imposed angleth_f, the phase matrix is cut
flat at its value atth_fand the truncation fraction follows from
that continuity (returned inf; the new output variabletrunc_frac
echoes the parameter, None in that case). Imposing bothtrunc_frac
andth_fsets the plateau by the normalization instead, in general
not continuous: 2.5 times the truncated phase matrix at 5 degrees for
Fournier-Forand (n = 1.10, mu = 3.5) andtrunc_frac=0.3, where the
continuous plateau takes f = 0.467. The docstring Notes describe the
three ways the angle, the fraction and the plateau set each other
Other changes
- Fix the truncation angle search of
gt_phase_approx, which could
return a truncated phase matrix with a negative plateau: for a peak as
sharp as Fournier-Forand's, the first moment matches best at angles
within which the peak holds less than the fractiontrunc_fracof the
scattering, where the plateau is negative (-445 at 1.1 degree for
n = 1.10, mu = 3.5,trunc_frac=0.3and the trapezoid rule). The
search now skips them, and raises a ValueError when no angle below
th_tolleaves a non-negative plateau. An imposed angle (th_f) is
unchanged. The searched references of the two-term Henyey-Greenstein
test case, whose plateaus were negative, are regenerated