Some errors have been detected in the above paper. In particular, in table 2 the headers of the first three columns should be K, L, M, while the headers of the next three columns should be n1, n2, n3, respectively. In the hypergeometric function shown in equation (20) the third argument must contain +4, rather than +3. In table 3 the used values of real parameters were a = 2.5, b = 2, c = 1.5 and d = 1. Furthermore, to evaluate the four-body (or three-electron) integrals of lowest order we have developed an alternative approach. In the lowest order three-electron integrals at least one of the n1, n2 and n3 parameters equals -1 (and all of them are odd). The approach is based on the Remiddi formula [1] and allows one to produce closed analytical expressions for all such integrals. The numerical values obtained with these formulae for the first and second integrals in table 2 differ slightly from the presented result values. For instance, the new value for the first integral is '...282 100 113 999 4...' (all 19 digits before '100' coincide with the result shown in table 2). Analogously, in the second integral in table 2 the first 32 digits coincide exactly with the new (i.e. Remiddi-based) value. Starting from the 33rd decimal digit the Remiddi-based value is '25 221 389 892 711 866'. The reason for such deviations is not quite clear and is currently under investigation. We are probably dealing with some unknown instability of the Perkins formula for lowest order four-body (or three-electron) integrals. It was also found that this instability disappears rapidly when the n1, n2, n3 parameters grow.
References
[1] Remiddi E 1991 Phys. Rev. A 44 5492