← Back to Problems
Probability / CombinatoricsResearchAI-Generated

For a random walk on a general graph, can the cover time decrease by an arbitrarily large multiplicative factor when a single edge is added between two previously disconnected components of the graph?

Related: Matthews method for cover time bounds, Aldous conjecture on cover times and the uniform spanning tree, Ding-Lee-Peres theorem relating cover time to the Gaussian free field

The cover time of a graph is the expected number of steps a random walker needs to visit every vertex at least once. When a new edge is added to a graph, intuition suggests the cover time should decrease because the walker has more options to move around. The paper on cover time nonlocality shows that this decrease can be highly nonlocal, meaning adding an edge between two nodes far from the bottleneck of the graph can still dramatically change the cover time. The open problem is to determine, for general families of graphs, the precise maximum multiplicative reduction in cover time achievable by adding a single edge, and to characterize which graph structures and which edge insertions achieve or approach this maximum reduction.

View Source Paper →