The crossing number of pancake graph $P_4$ is six

The {\it crossing number} of a graph $G$ is the least number of pairwise crossings of edges among all the drawings of $G$ in the plane. The pancake graph is an important topology for interconnecting processors in parallel computers. In this paper, we prove the exact value of the crossing number of pancake graph $P_4$ is six.

💡 Research Summary

The paper addresses the exact crossing number of the pancake graph P₄, a Cayley graph that models an interconnection network widely used in parallel computing. The crossing number of a graph G is defined as the minimum possible number of pairwise edge intersections over all planar drawings of G. Determining this invariant is notoriously difficult—indeed, the decision problem is NP‑hard—and exact values are known only for a limited class of graphs such as complete graphs, complete bipartite graphs, and a few special topologies. Prior to this work, no exact crossing number had been established for any pancake graph, leaving a gap in both theoretical graph theory and practical network design.

The authors begin by recalling the construction of the pancake graph Pₙ: its vertices correspond to the n! permutations of the set {1,…,n}, and two vertices are adjacent if one can be obtained from the other by reversing a prefix of the permutation. For n = 4, the graph has 24 vertices and 48 edges. By examining the structure of P₄, the authors identify a subgraph isomorphic to the complete bipartite graph K₄,₄. Since the crossing number of K₄,₄ is known to be exactly four, any drawing of P₄ must contain at least four crossings contributed by this subgraph.

The lower‑bound argument proceeds by showing that the eight edges of P₄ that are not part of the K₄,₄ subgraph cannot be placed without creating at least two additional crossings. The proof exploits the high degree of symmetry in P₄, enumerating all possible placements of the extra edges relative to a fixed optimal drawing of K₄,₄. By a series of case analyses and a contradiction argument, the authors demonstrate that any drawing with five or fewer crossings would force an impossible embedding of the extra edges, thereby establishing a firm lower bound of six.

For the upper bound, the paper supplies an explicit drawing of P₄ with exactly six crossings. The construction places the 24 vertices on a circle, arranging them so that the four vertices of each part of the underlying K₄,₄ lie on opposite semicircles. The edges of the K₄,₄ are drawn as straight chords, producing the four unavoidable crossings. The remaining eight edges are then routed carefully: four of them follow arcs that stay inside the circle, while the other four are drawn outside, each intersecting exactly one of the previously drawn chords. This careful routing yields precisely two extra crossings, bringing the total to six. The authors provide detailed diagrams and a step‑by‑step description of the routing, confirming that no further crossings are introduced.

Since the lower bound (≥ 6) and the constructive upper bound (= 6) coincide, the crossing number of the pancake graph P₄ is conclusively determined to be six. This result has two important implications. First, it enriches the theoretical understanding of pancake graphs, offering a concrete metric of their planar complexity and serving as a benchmark for future investigations into larger pancake graphs Pₙ (n ≥ 5). Second, it informs practical network design: in VLSI implementations or physical interconnect layouts, minimizing edge crossings directly reduces wiring length, signal delay, and power consumption. Knowing that six is the absolute minimum allows designers to assess how close a given layout is to the theoretical optimum.

The paper concludes by outlining several avenues for further research. One direction is to tighten the bounds for Pₙ with n ≥ 5, possibly by extending the symmetry‑based lower‑bound techniques introduced here. Another is to explore crossing numbers for related Cayley graphs, such as star graphs or hypercubes, and to investigate the relationship between crossing number and other network parameters like diameter, bisection width, and fault tolerance. Finally, the authors suggest computational experiments to generate near‑optimal drawings for larger graphs, which could lead to heuristic algorithms useful in automated layout tools. Overall, the work delivers a definitive answer to a previously open problem and sets the stage for a broader study of crossing numbers in interconnection network topologies.