Statistics on permutations

Toufik Mansour1, Yidong Sun2
1Department of Mathematics, University of Haifa, 31905 Haifa, Israel
2Department of Mathematics, Dalian Maritime University, 116026 Dalian, P.R. China

Abstract

Let \( \pi = \pi_1 \pi_2 \cdots \pi_n \) be any permutation of length \( n \), we say a descent \( \pi_i \pi_{i+1} \) is a {lower}, {middle}, {upper} if there exists \( j > i+1 \) such that \( \pi_j < \pi_{i+1}, \pi_{i+1} < \pi_j < \pi_i, \pi_i < \pi_j \), respectively. Similarly, we say a rise \( \pi_i \pi_{i+1} \) is a {lower}, {middle}, {upper} if there exists \( j > i+1 \) such that \( \pi_j < \pi_i, \pi_i < \pi_j < \pi_{i+1}, \pi_{i+1} < \pi_j \), respectively. In this paper, we give an explicit formula for the generating function for the number of permutations of length \( n \) according to the number of upper, middle, lower rises, and upper, middle, lower descents. This allows us to recover several known results in the combinatorics of permutation patterns as well as many new results. For example, we give an explicit formula for the generating function for the number of permutations of length \( n \) having exactly \( m \) middle descents.

Keywords: Statistics; Kernel method; Generating functions; Descents; Rises