[Paper Review] Bijective Proofs for "Enumerative Properties of Ferrers Graphs"

Jason Matthew Burns|ArXiv.org|Dec 15, 2003

Advanced Combinatorial Mathematics3 citations

TL;DR

This paper provides bijective proofs for two enumerative properties of Ferrers graphs: the number of Hamiltonian paths and the number of spanning trees. Using explicit, invertible constructions between combinatorial objects—pairs of rook placements and Hamiltonian paths, and configurations of R/C marks with an X and spanning trees—it establishes a direct, constructive correspondence that proves the known formulas bijectively, extending to weighted cases via generating functions.

ABSTRACT

Recently, Ehrenborg and Van Willenburg defined a class of bipartite graphs that correspond naturally to Ferrers diagrams, and proved several results about them. We give bijective proofs for the (already known) expressions for the number of spanning trees and (where applicable) Hamiltonian paths of these graphs. Their paper can be found at http://www.ms.uky.edu/~jrge/Papers/Ferrers_graphs.pdf .

Motivation & Objective

• To provide bijective proofs for the number of Hamiltonian paths in Ferrers graphs, which Ehrenborg and van Willigenburg had previously proven non-bijectively.
• To establish a bijective correspondence between spanning trees of Ferrers graphs and combinatorial configurations of R’s, C’s, and an X in the top-left corner.
• To extend these bijections to weighted spanning trees by assigning edge weights and showing the total weight matches a product formula.
• To explore the potential of extending these results to skew Ferrers diagrams and Tutte polynomial computations.
• To investigate whether the symmetries observed in the bijections (e.g., swapping roles of rows/columns or A/B labels) have deeper algebraic or structural significance.

Proposed method

• Construct a forward map from Hamiltonian paths to pairs of n-rook placements by marking A’s and B’s in specific columns based on path vertex order, ensuring column availability via a 'next available' rule.
• Define the inverse map from pairs of n-rook placements to Hamiltonian paths by successively removing A’s and B’s in row and column order, reconstructing the path step-by-step.
• For spanning trees, use a two-stage construction: first remove rows/columns with only one R or C, then reduce to irreducible configurations equivalent to rook placements or Hamiltonian paths.
• Represent the weighted case by assigning weights $ x_a y_b $ to edges and showing the total weight of all spanning trees equals the product of weights from all valid configurations of R’s, C’s, and an X.
• Use the fact that each symbol (R, C, X) contributes independently to the total weight, with R’s in row $ a $ contributing $ x_a (y_1 + \cdots + y_{\lambda_a}) $, and C’s in column $ b $ contributing $ y_b (x_1 + \cdots + x_{\lambda_b'}) $.
• Leverage the structure of Ferrers diagrams to ensure that in the irreducible case, the number of R’s and C’s are equal and form valid rook placements, enabling a bijection with Hamiltonian paths on the remaining subgraph.

Experimental results

Research questions

• RQ1 Can the number of Hamiltonian paths in a Ferrers graph be proven bijectively, matching the square of the number of n-rook placements on the diagram?
• RQ2 Is there a constructive, invertible mapping between Hamiltonian paths and pairs of n-rook placements that preserves the combinatorial structure?
• RQ3 Can the formula for the number of spanning trees in a Ferrers graph be given a bijective interpretation using marked configurations of R’s, C’s, and an X?
• RQ4 How can the weighted version of the spanning tree formula be derived and verified via a weight-preserving bijection?
• RQ5 What is the structural significance of the symmetries observed when swapping row/column roles or A/B labels in the bijection?

Key findings

• The number of Hamiltonian paths in a Ferrers graph with n rows and n columns is equal to the square of the number of n-rook placements on the corresponding Ferrers diagram, and this equality is established via a bijective, invertible construction.
• The bijection between Hamiltonian paths and pairs of n-rook placements is explicitly defined and reversible, with the 'next available column' condition guaranteed by the structure of the path and the placement rules.
• The number of spanning trees in a Ferrers graph is given by the product of the lengths of all but the first row and all but the first column, and this formula is proven bijectively via a two-stage reduction to irreducible configurations.
• The weighted spanning tree formula is derived by assigning weights $ x_a y_b $ to edges and showing the total weight of all spanning trees equals the product of individual symbol contributions: $ x_1y_1 \cdot \prod_{a=2}^m x_a (\sum_{j=1}^{\lambda_a} y_j) \cdot \prod_{b=2}^n y_b (\sum_{i=1}^{\lambda_b'} x_i) $.
• The bijection preserves weight at each stage: edge removal in the first stage preserves weight, and the second stage’s contribution matches the weight of the original configuration.
• The irreducible configurations in the spanning tree construction correspond exactly to pairs of n-rook placements and Hamiltonian paths on the subgraph, confirming the bijection through structural equivalence.