General Science 3 JAN, 2023

A Fine-Grained Classification of the Complexity of Evaluating the Tutte Polynomial on Integer Points Parameterized by Treewidth and Cutwidth

A Fine-Grained Classification of the Complexity of Evaluating the Tutte Polynomial on Integer Points Parameterized by Treewidth and Cutwidth

We give a fine-grained classification of evaluating the Tutte polynomial $T(G;x,y)$ on all integer points on graphs with small treewidth and cutwidth. Specifically, we show for any point $(x,y) \in \mathbb{Z}^2$ that either - can be computed in polynomial time, - can be computed in $2^{O(tw)}n^{O(1)}$ time, but not in $2^{o(ctw)}n^{O(1)}$ time assuming the Exponential Time Hypothesis (ETH), - can be computed in $2^{O(tw \log tw)}n^{O(1)}$ time, but not in $2^{o(ctw \log ctw)}n^{O(1)}$ time assuming the ETH, where we assume tree decompositions of treewidth $tw$ and cutwidth decompositions of cutwidth $ctw$ are given as input along with the input graph on $n$ vertices and point $(x,y)$. To obtain these results, we refine the existing reductions that were instrumental for the seminal dichotomy by Jaeger, Welsh and Vertigan~[Math. Proc. Cambridge Philos. Soc'90]. One of our technical contributions is a new rank bound of a matrix that indicates whether the union of two forests is a forest itself, which we use to show that the number of forests of a graph can be counted in $2^{O(tw)}n^{O(1)}$ time.

💡 Research Summary

The paper delivers a fine‑grained, parameterized complexity classification for evaluating the Tutte polynomial T(G;x,y) at every integer point (x,y) on graphs whose structural parameters—treewidth (tw) and cutwidth (ctw)—are bounded. Building on the classic Jaeger‑Welsh‑Vertigan dichotomy (1990), which separates points into “polynomial‑time” and “#P‑hard”, the authors refine the reduction machinery so that, when a tree decomposition of width tw and a cutwidth decomposition of width ctw are supplied, each integer point falls into exactly one of three algorithmic regimes:

  1. Polynomial‑time points – a small, well‑known set of integer coordinates (e.g., (1,1), (−1,−1), (0,−1), (−1,0)) for which the Tutte value can be computed in O(n^c) time irrespective of the graph’s structure.

  2. Treewidth‑exponential but cutwidth‑subexponential points – for these points the Tutte value can be computed in time 2^{O(tw)}·n^{O(1)} by a dynamic‑programming (DP) algorithm that processes a given tree decomposition. The key technical device is a new rank bound on a matrix that encodes whether the union of two forests is itself a forest. This bound (rank = O(tw)) enables a compact representation of DP states, yielding the 2^{O(tw)} factor. Assuming the Exponential Time Hypothesis (ETH), the authors prove that no algorithm can achieve 2^{o(ctw)}·n^{O(1)} time, establishing cutwidth as a strictly stronger lower‑bound parameter for this class.

  3. Log‑factor treewidth‑exponential points – for a second set of integer points the best possible algorithm runs in 2^{O(tw log tw)}·n^{O(1)} time. Here the DP must keep track of edge‑subset configurations whose number grows like tw·log tw. Again, using the forest‑union rank bound the authors compress the state space to this size. Under ETH they show that any algorithm running in 2^{o(ctw log ctw)}·n^{O(1)} would contradict known hardness results, so the log‑factor bound is tight with respect to cutwidth.

A central contribution is the forest‑union rank lemma: given two forests on the same vertex set, the matrix whose rows correspond to edges of the first forest and columns to edges of the second forest has rank at most O(tw) when the underlying graph is embedded in a tree decomposition of width tw. This lemma not only underpins the 2^{O(tw)} algorithm for counting all forests (a result that generalises the classic spanning‑tree counting algorithm) but also provides the structural insight needed to prove ETH‑based lower bounds for cutwidth.

The paper’s structure proceeds as follows. After an introductory section reviewing the Tutte polynomial, previous dichotomies, and parameterized complexity basics, the authors formalize the problem and the input model (tree and cutwidth decompositions are part of the input). Section 3 develops the linear‑algebraic machinery, proving the rank bound and demonstrating how it yields an exact counting algorithm for forests in 2^{O(tw)}·n^{O(1)} time. Section 4 presents the DP over tree decompositions that leads to the 2^{O(tw)} algorithm for the second class of points, and it contains the ETH‑based reduction from the canonical cutwidth‑hard problem (e.g., Hamiltonian Path on bounded‑cutwidth graphs) to show the 2^{o(ctw)} lower bound. Section 5 extends the approach to the log‑factor regime, detailing how the DP state space expands to tw log tw and proving the matching ETH lower bound with the log factor. The final section discusses implications, possible extensions to other graph parameters (pathwidth, clique‑width), approximation schemes, and open questions such as whether similar rank‑based techniques can be applied to other graph polynomials.

In summary, the authors achieve a complete fine‑grained classification: every integer evaluation point of the Tutte polynomial on bounded‑tw/ctw graphs is either (i) polynomial‑time, (ii) solvable in 2^{O(tw)}·n^{O(1)} but not faster than 2^{o(ctw)}·n^{O(1)} (ETH), or (iii) solvable in 2^{O(tw log tw)}·n^{O(1)} but not faster than 2^{o(ctw log ctw)}·n^{O(1)} (ETH). The work advances both algorithmic graph theory—by providing concrete DP algorithms with provably optimal dependence on structural parameters—and complexity theory, by delivering ETH‑tight lower bounds that differentiate treewidth from cutwidth. This fine‑grained dichotomy sets a new benchmark for future studies of graph polynomials under structural parameterizations.