2012/05/25 by Frank den Hollander, Hollander, Frank den, Alex A. Opoku +1
Mathematics · #60F10 #60G50 #60K35 #82D60 #FOS: Mathematics #Probability (math.PR) #math.PR #msc:60F10 #msc:60G50 #msc:60K35 #msc:82D60
paper · pdf · doi:10.48550/arxiv.1205.5594
This paper has been withdrawn due to an error in the proof of Theorem 1.1
arxiv created 2012/06/08 · arxiv updated 2012/06/11
In [3] the radius of convergence of the generating function of the collision local time of two independent copies of an irreducible, symmetric and transient random walk on Zd, d ≥ 1, was studied. Two versions were considered: z1, the radius of convergence when one walk is averaged over and the other is kept fixed; z2, the radius of convergence when both walks are averaged over. While z2 can be easily computed, no explicit expression is available for z1. In [3] it was conjectured that z1 > z2 under a weak regularity assumption on the random walk. However, this gap was only proved for strongly transient random walk. Subsequently, in [1], [4] and [5] the gap was proved for a subclass of random walks that are transient but not strongly transient. In the present note we settle the full conjecture. The proof is based on a comparison of variational formulas for z1 and z2 derived in [3]. Under additional weak regularity assumptions on the random walk, we derive an upper bound for z1. We further show that the gap persists when the random walk has exponential waiting times. The gaps imply the presence of an intermediate phase for a model of two random walks with a pinning interaction. As explained in [3], they also imply the presence of intermediate phases in three classes of interacting stochastic systems: coupled branching processes, interacting diffusions, and directed polymers with bulk disorder.