Direct calculation of probabilities of sums of independet lattice random variables (Q1897897)

The method described is applicable to sums of independent random variables taking values in \(\{0,1,2, \ldots\}^N\) for some \(N \geq 1\). Consider the case with \(N = 1\) and where all summands have the distribution \(P(X = k) = p_k\), \(k \geq 0\); write \(q_{j \nu} = P(S_j = \nu)\) where \(S_j\) denotes the sum of \(j\) summands. The method is based on an expression for the ``diagonal'' element \(q_{\nu \nu}\) in terms of \(\{p_0, p_1, \ldots, p_\nu\}\) \((\nu = 1,2, \ldots)\), and a system of differential equations relating the off-diagonal elements to \(\{q_{\nu \nu}\}\). The technique is rather complicated but is said to be suitable for implementation on a modern computer.