ABSTRACT

This has the probability interpretation that for j specified cells the ith. one has frequency exactly r* (i = 1 , 2 , . . . , j) and for the re­ maining b — j cells the ath cell has frequency at least rj+a (a — 1, 2 , . . . , b — j). One immediate advantage of this more general form

In some applications, we have to compute, or transform a formula with the j-value (i.e. that part of the second superscript which is after a comma) not equal to zero. In such cases we use the so called “reduction formula”, which enables us to reduce the superscript j to 0 and express (b, j) by (b — j, 0), for which we have fast programs.