Disorder relevance for the random walk pinning model in dimension 3 (Q629810)

The paper extends previous results of the authors on the continuous time random walk pinning model [Ann. Inst. Henri Poincaré, Probab. Stat. 46, No. 2, 414--441 (2010; Zbl 1206.60087)]. The representation of the problem as a weighted renewal process in random environment involves Gibbs weights at an inverse temperature \(\beta\). At some critical \(\beta _c \geq 0\), the model undergoes a localization-delocalization transition. The question of whether \(\beta _c\) differs from the critical point \(\beta _c^{\text{ann}}\) of the annealed model is translated to that of disorder (ir)relevance. Previous proofs of disorder irrelevance in dimensions \(d=1\), \(d=2\) and disorder relevance in \(d\geq 4\) are here supplemented by a proof of disorder relevance in the critical dimension \(d=3\) if \(\beta _c -\beta _c^{\text{ann}}\) is of a suitable order. The proof employs coarse graining and fractional moment techniques. A continuous time version of Doney's local limit theorem [loc. cit.] is proved here for renewal processes with infinite mean.