We solve the boson normal ordering problem for F (a † ) r a s , with r, s positive integers, a, a † = 1, i.e. we provide exact and explicit expressions for its normal form N F (a † ) r a s , where in N (F ) all a's are to the right. The solution involves integer sequences of numbers which are generalizations of the conventional Bell and Stirling numbers whose values they assume for r = s = 1. A comprehensive theory of such generalized combinatorial numbers is given including closed-form expressions (extended Dobinski -type formulas) and generating functions. These last are special expectation values in boson coherent states.
Consider a function F (x) having a Taylor expansion aroundIn this note we will collect the formulas concerning our solution of the normal ordering problem for F (a † ) r a s , where a, a † are the boson annihilation and creation operators, a, a † = 1, and r and s are positive integers. The normally ordered form of the operator, where in N (F ) all the a's are to the right. It satisfies the operator identity:Furthermore, an auxiliary symbol : O(a, a † ) : will be used, which means expand O in powers of a and a † and order normally assuming they commute [1], [2].The combinatorial numbers S(n, k), known as Stirling numbers of the second kind, and their sums B(n), the Bell numbers, arise naturally in the normal ordering procedure for r = s = 1, as follows [3]:which may be taken as definitions of S(n, k) and B(n). In the present work we shall treat the case of general (r, s), consequently generalizing these combinatorial numbers.For the moment we restrict ourselves to the case r ≥ s, the other alternative being treated later. We do not give the proofs of the formulas here; they will be given elsewhere [4]. The case r = s = 1 is known [1], [2] and some features of the r > 1, s = 1 case have been published [5]. When needed we shall refer to [1], [2] and [5] where particular cases of our more general formulas are discussed.We define the set of positive integers S r,s (n, k) entering the expansion:(a † ) r a s n = N (a † ) r a s n = (a † )n(r−s) ns k=s S r,s (n, k)(a † ) k a