We performed a rigorous theoretical convergence analysis of the discretedipole approximation (DDA). We prove that errors in any measured quantity arebounded by a sum of a linear and quadratic term in the size of a dipole d, whenthe latter is in the range of DDA applicability. Moreover, the linear term issignificantly smaller for cubically than for non-cubically shaped scatterers.Therefore, for small d errors for cubically shaped particles are much smallerthan for non-cubically shaped. The relative importance of the linear termdecreases with increasing size, hence convergence of DDA for large enoughscatterers is quadratic in the common range of d. Extensive numericalsimulations were carried out for a wide range of d. Finally we discuss a numberof new developments in DDA and their consequences for convergence.

我们对离散偶极近似 (DDA) 进行了严格的理论收敛分析。我们证明，当偶极子 d 在 DDA 适用范围内时，任何测量量的误差都受偶极子 d 大小的线性项和二次项之和的限制。此外，立方体的线性项明显小于非立方体形状的散射体。因此，对于小的 d 误差，立方体形状的粒子比非立方体形状的要小得多。线性项的相对重要性随着尺寸的增加而降低，因此对于足够大的散射体，DDA 的收敛在 d 的公共范围内是二次的。对大范围的 d 进行了广泛的数值模拟。最后，我们讨论了 DDA 中的一些新发展及其对收敛的影响。