T-MATRIX COMPUTATIONS OF LIGHT-SCATTERING BY LARGE SPHEROIDAL PARTICLES

It is well known that T-matrix computations of light scattering by nonspherical particles may suffer from the ill-conditionality of the process of matrix inversion, which has precluded calculations for particle size parameters larger than about 25. It is demonstrated that calculating the T-matrix using extended-precision instead of double-precision floating-point variables is an effective approach for suppressing the numerical instability in computations for spheroids and allows one to increase the maximum particle size parameter for which T-matrix computations converge by as significant a factor as 2-2.7. Yet this approach requires only a negligibly small extra memory, an affordable increase in CPU time consumption, and practically no additional programming effort. As a result, the range of particle size parameters, for which rigorous T-matrix computations of spheroidal scattering can be performed, now covers a substantial fraction of the gap between the domains of applicability of the Rayleigh and geometrical optics approximations.