[Paper Review] Sandwiching saturation number of fullerene graphs

Vesna Andova, Kardoš, František+1 more|arXiv (Cornell University)|May 9, 2014

Fullerene Chemistry and Applications20 references14 citations

TL;DR

This paper establishes that the saturation number of fullerene graphs—defined as the size of the smallest maximal matching—is asymptotically n/3, where n is the number of vertices. Using a discharging method on face and vertex charge distributions, the authors prove a tight lower bound of n/3 − 2 and show that the previous bound of 3n/10 is only tight for small fullerenes (n ≤ 60), resolving a key gap in understanding structural stability predictors in carbon molecules.

ABSTRACT

The saturation number of a graph $G$ is the cardinality of any smallest maximal matching of $G$, and it is denoted by $s(G)$. Fullerene graphs are cubic planar graphs with exactly twelve 5-faces; all the other faces are hexagons. They are used to capture the structure of carbon molecules. Here we show that the saturation number of fullerenes on $n$ vertices is essentially $n/3$.

Motivation & Objective

• To refine the known bounds on the saturation number of fullerene graphs, which is a key graph-theoretical invariant linked to molecular stability.
• To determine whether the previously established lower bound of 3n/10 is tight or can be improved for larger fullerenes.
• To investigate the structural conditions under which the saturation number reaches its theoretical minimum.
• To explore the computational complexity of computing the saturation number for fullerene graphs.
• To provide a tighter, asymptotically optimal bound on the saturation number using a discharging method on face and vertex charges.

Proposed method

• A discharging method is applied to fullerene graphs, assigning initial charges to faces and vertices based on their type (pentagon, hexagon, white/black vertices).
• Charges are redistributed via three rules: (R3) good faces send +1 to incident white vertices; (R4) bad hexagons send −1 to incident white vertices; (R5) transition hexagons send −1 to incoming and +1 to outgoing white vertices.
• The method ensures no face ends with negative charge, and negative charges at white vertices are offset by positive contributions from adjacent faces.
• The analysis shows that every white vertex receiving −1 charge from a bad or transition hexagon is compensated by a neighboring face sending +1, preserving non-negative total charge.
• The total charge sum is non-negative, which implies a lower bound on the saturation number through combinatorial and topological constraints of the graph.
• The proof leverages the 3-regularity and planarity of fullerene graphs, along with face and vertex degree constraints derived from Euler's formula.

Experimental results

Research questions

• RQ1 What is the asymptotic behavior of the saturation number in fullerene graphs as the number of vertices increases?
• RQ2 Can the previously known lower bound of 3n/10 for the saturation number be improved for larger fullerenes?
• RQ3 Under what structural conditions does the saturation number achieve its theoretical minimum of approximately n/3?
• RQ4 Are there infinitely many fullerene graphs for which the saturation number is exactly n/3 − 2?
• RQ5 Is the problem of computing the saturation number for fullerene graphs NP-complete?

Key findings

• The saturation number of any fullerene graph on n vertices is at least n/3 − 2, which is asymptotically tight.
• The previous lower bound of 3n/10 is only valid for fullerenes with at most 60 vertices, and only for specific small cases.
• There exist infinitely many fullerenes—such as (8,0)-nanotubes with 3k+1 hexagonal rings—where the saturation number is exactly n/3 − 2.
• The dodecahedron (20 vertices) and buckminsterfullerene (60 vertices) are the only known fullerenes achieving the 3n/10 bound, and only for n ≤ 60.
• The saturation number is strictly larger than 3n/10 for all fullerenes with more than 60 vertices.
• The authors conjecture that s(F) ≤ n/3 + C for some absolute constant C and all fullerene graphs F, suggesting a universal upper bound near n/3.