High moments of 2d directed polymers up to quasi-criticality
Clément Cosco, Shuta Nakajima
TL;DR
This work analyzes two-dimensional directed polymers in random environments under subcritical and quasi-critical scaling, deriving upper bounds for the q-th moments of the partition function up to a diverging threshold q_N. The authors develop a diagrammatic expansion and a refined induction scheme that separates pairwise from higher-order intersections, enabling Gaussian-type moment bounds to hold beyond previously known thresholds and for general weight distributions. They also prove Gaussian-rate large deviation results for log W_N in both regimes and provide a moment-formula framework that extends to non-Gaussian weights. The results sharpen our understanding of high-moment behavior and its connections to extreme-value statistics and log-correlated structures in 2D random polymers, with potential implications for critical and near-critical phenomena. The methodology blends diagrammatic expansions, truncations, long-jump inductions, and careful probabilistic-combinatorial control to bridge subcritical, quasi-critical, and higher-q regimes.
Abstract
We consider two-dimensional directed polymers in random environment in the sub-critical regime and in the quasi-critical regime introduced recently by Caravenna, Cottini and Rossi, arXiv:2307.02453v1. For $q\leq q_N$ with $q_N\to\infty$ diverging at a suitable rate with the size of the system, we obtain upper bound estimates on the $q$-moment of the partition function for general environments. In the sub-critical regime, our results improve the $q_N$-threshold obtained for Gaussian environment in Cosco, Zeitouni, Comm. Math. Phys (2023). As a corollary, we derive large deviation estimates with a Gaussian rate function.
