Counting descent pairs with prescribed tops and bottoms (Q942164)

In this excellent paper the authors give nice combinatorial proofs of several Theorems and Formula's concerned with counting descent pairs within permutations in which the top of the pair lies in some fixed set \(X\) and the bottom lies in another fixed set \(Y\). It is shown that the connection (equality) of certain formula's is related to some known general transformations for hypergeometric series of Karlsson-Minton type. Also the authors reveal how their results are special cases of hit polynomials for Ferrer's boards (i.e. counting the number of specific kinds of (non-attacking) rook placements on a board. This gives yet an alternative way of proving the earlier mentioned results. Next the results are generalized to counting descent pairs within words of specific types, rather than in permutations. The paper concludes with some suggestions for future work. One of the combinatorial arguments used is a very nice one, it introduces a weight function assigning weights \(1\) or \(-1\) to certain structures and an involution that pairs the structures so that if the image of the structure is not equal to the structure itself the sign of the weight of the structure differs from the sign of the weight of its image. Furthermore the weight of a fixed point of the involution equals always 1. Summing over the weigts one gets the formula's to be proved, the fixed points of the defined involution turn out to be the objects to be counted. It turns out that this same argument can be easily generalized so that it works also for words in stead of permutations. At a certain point the authors use the Pfaff-Saalschütz \({}_3F^2\) summation formula, which I did not check. I spent quite some time on deriving the result in Remark 3.3 using what the authors call a similar computation and finally I succeeded. This made me fully appreciate the nice short reasoning given on page 713 leading to the same result. At first glance I did not see the connection between the result in Corollary 3.5 and the integral form of a transformation of Karlsson-Minton type hypergeometric series given on page 707. However, I saw that Corollary 3.5 is a special case of Corollary 3.6 and the result given in Corollary 3.6 is easily seen to be connected to the given transformation (after some computation). The reason why using rook placements on Ferrer's boards can be used in such a nice way to derive the earlier results lies in the fact that the nasty calculations are now hidden in a very powerful result of Foata and Schützenberger, stating that certain Ferrer's boards are rook equivalent under certain conditions. This appears to be a highly non-trivial but very useful result. In Remark 5.4 the authors mention the use of Foata's transformation but actually this transformation is defined for permutations rather than for words and so it has to be redefined for words first. I found out how that can be done but would have appreciated if the authors had given some hints. Just for the sake of completeness I want to point out a number of errors in the paper (one of which is a major one). I start with Corollary 2.7, Formula 2.6 in which the product has to be taken over \(X_n \). Next in the proof of Corollary 3.6 on page 707 \(b\) should be defined as \(b=a+1-M\) in stead of \(b=a+1-M-u_1\). Furthermore in the Formula given in Corollary 5.7 the binomial coefficient with upper part \(n+1\) should have as a lower part \(n-a-s-r\) in stead of \(s-r\). To cotinue, the number of even descents of the word \(w\) given ate the end of page 722 is three rather than two, the descents being 41, 42 and 21. Finally the results given in Theorems 6.1 and 6.2 are incorrect. The Formua of Theorem 6.1 should read: \[ (kr+(k-1)m)!\sum_{p=0}^s (-1)^{s-p}\binom{kr+(k-1)m+p}{p} \binom{k(r+m)+1}{s-p}(kr+(k-1)m+p)^m. \] The formula of Theorem 6.2 should read: \[ (kr+(k-1)m)!\sum_{p=0}^{m-s}(-1)^{m-s-p} \binom{kr+(k-1)m+p}{p} \binom{k(r+m)+1}{m-s-p}(p+1)^m. \]