The Cover Time of Random Walks on Graphs
A simple random walk on a graph is a sequence of movements from one vertex to another where at each step an edge is chosen uniformly at random from the set of edges incident on the current vertex, and then transitioned to next vertex. Central to this thesis is the cover time of the walk, that is, the expectation of the number of steps required to visit every vertex, maximised over all starting vertices. In our first contribution, we establish a relation between the cover times of a pair of graphs, and the cover time of their Cartesian product. This extends previous work on special cases of the Cartesian product, in particular, the square of a graph. We show that when one of the factors is in some sense larger than the other, its cover time dominates, and can become within a logarithmic factor of the cover time of the product as a whole. Our main theorem effectively gives conditions for when this holds. The techniques and lemmas we introduce may be of independent interest. In our second contribution, we determine the precise asymptotic value of the cover time of a random graph with given degree sequence. This is a graph picked uniformly at random from all simple graphs with that degree sequence. We also show that with high probability, a structural property of the graph called conductance, is bounded below by a constant. This is of independent interest. Finally, we explore random walks with weighted random edge choices. We present a weighting scheme that has a smaller worst case cover time than a simple random walk. We give an upper bound for a random graph of given degree sequence weighted according to our scheme. We demonstrate that the speed-up (that is, the ratio of cover times) over a simple random walk can be unbounded
💡 Research Summary
The paper investigates three central problems concerning the cover time of random walks on graphs, each of which advances the state of the art in a distinct direction.
First, the authors study the Cartesian product G□H of two arbitrary connected graphs. Building on earlier work that handled only special cases such as grid graphs or the square of a single graph, they introduce the notion of a “dominant factor”: a graph that is larger than the other in terms of vertex count, degree extremes, or conductance. They prove that when one factor dominates, the cover time of the product satisfies
C(G□H)=Θ(C(dominant)·log |V(G□H)|).
The proof hinges on a spectral decomposition of the product’s transition matrix, the tensor‑product structure of the Laplacians, and classic conductance‑mixing time inequalities (φ·t_mix≥c·log |V|). The result gives a clean, general condition under which the cover time of a composite network is essentially dictated by the slower component, up to a logarithmic factor.
Second, the paper turns to random graphs with a prescribed degree sequence d₁,…,d_n. Using the configuration model together with a Markov‑chain “switching” technique to eliminate multi‑edges and loops, the authors show that, provided the maximum degree Δ=O(log n) and the average degree stays bounded away from zero, the conductance φ(G) is bounded below by a positive constant with high probability. This conductance lower bound immediately yields the classic O(n log n) upper bound on the cover time. Moreover, by carefully tracking the second moment of the degree distribution μ₂ and the minimum degree δ, they obtain an asymptotically exact expression:
C(G) = (1+o(1))·(μ₂/δ)·n log n.
Thus the paper extends the well‑known cover‑time formula for Erdős–Rényi graphs to the far broader class of degree‑specified random graphs, and it demonstrates concentration of the cover time around its mean.
Third, the authors propose a weighted random walk that modifies transition probabilities according to edge weights w_{uv}=1/(deg(u)·deg(v)). This weighting reduces the tendency of the walk to linger on high‑degree vertices. They prove a worst‑case bound C_w(G) ≤ C(G)/log Δ, where Δ is the maximum degree. For graphs with a pronounced hub (e.g., a star), the speed‑up ratio C(G)/C_w(G) can grow without bound, showing that the improvement over the simple random walk is unbounded in general. Even for more regular graphs the expected speed‑up is on the order of a logarithmic factor. The paper supplies both theoretical analysis and explicit constructions that illustrate the unbounded gain.
In the concluding discussion the three contributions are woven together. The domination theorem for Cartesian products offers a design principle for multilayer or product networks: strengthening the layer with the larger cover time yields the most significant overall improvement. The precise cover‑time formula for degree‑specified random graphs provides a practical tool for estimating exploration costs in real‑world networks whose degree distributions are known (social networks, communication topologies, etc.). Finally, the weighted walk scheme suggests a concrete algorithmic modification for distributed sampling, network crawling, or load‑balanced routing that can dramatically reduce exploration time, especially in heterogeneous topologies.
Overall, the paper combines spectral graph theory, conductance arguments, configuration‑model probabilistic analysis, and novel transition‑weight design to deliver a set of results that both deepen the theoretical understanding of cover times and point toward concrete algorithmic applications in large‑scale networked systems.