Andrey Miroshnichenko reported that obtained values of Csca for a limited
angular range (e.g. from 0 to 33 degrees in theta) are not consistent (doesn't
satisfy the optical theorem). In particular, they may even be larger than Csca
for a complete range.
This is caused by the fact that all integrations in ADDA are actually
averaging. So smaller interval leads to proportional scaling of the integrand.
This is logical for other parts of ADDA (orientation averaging, or azimuthal
integration of light-scattering pattern), but is definitely wrong in this
particular case for Csca.
So we need to devise a logical default behavior (i.e. integration over theta,
while averaging over phi - the latter is convenient for symmetry
considerations). Also a lot of comments should be added both to manual and
parameter files.
Original issue reported on code.google.com by
yurkinon 19 May 2014 at 10:46