The study of graph colorings leads to rich mathematical objects called chromatic quasisymmetric functions, which encode information about how a graph can be properly colored (where adjacent vertices get different colors) while also tracking additional combinatorial data. For a special family of graphs called natural unit interval graphs, researchers had conjectured that these functions have a particularly nice property called e-log-concavity when expanded in terms of a standard algebraic basis. Log-concavity is a regularity condition on a sequence of numbers, roughly meaning each term squared is at least as large as the product of its two neighbors. Several research groups independently put forward versions of this conjecture, and it had remained open as a plausible structural feature of these functions.
The paper provides a concrete counterexample that disproves all those conjectures at once. The authors identify a specific connected graph on 13 vertices, described by a particular combinatorial recipe, and compute the relevant expansion of its chromatic quasisymmetric function. One specific coefficient sequence turns out to be 1, 6, 38, and since 6 squared equals 36, which is less than 1 times 38, the log-concavity condition fails right there. The sequence is still positive, symmetric, and unimodal (rising then falling), so it has many of the good properties researchers expected, just not quite all of them.
The authors are careful to make this result fully trustworthy. The computation was verified using a standalone program that relies entirely on exact integer and rational arithmetic, with no floating-point approximations that could introduce errors. This kind of certified calculation is important because the counterexample involves intricate combinatorial objects where numerical mistakes could easily mislead. The work closes off an entire line of conjectures and redirects future research toward understanding which weaker or modified regularity properties these functions do reliably satisfy.