Abstract:We consider large deviations of the cover time of the discrete torus $(\mathbb{Z}/N\mathbb{Z})^d$, $d \geq 3$ by simple random walk. We prove a lower bound on the probability that the cover time is smaller than $\gamma\in (0,1)$ times its expected value, with exponents matching the upper bound from [Goodman-den Hollander, Probab. Theory Related Fields (2014)] and [Comets-Gallesco-Popov-Vachkovskaia, Electron. J. Probab. (2013)]. Moreover, we derive sharp asymptotics for $\gamma \in (\frac{d+2}{2d},1)$. The strong coupling of the random walk on the torus and random interlacements developed in a recent work [Prévost-Rodriguez-Sousi, <a class="link-https" data-arxiv-id="2309.03192" href="https://arxiv.org/abs/2309.03192">arXiv:2309.03192</a>] serves as an important ingredient in the proofs.
What problem does this paper attempt to address?
This paper aims to solve the problem of large - scale deviation of the covering time of simple random walks on high - dimensional discrete tori ($d\geq3$). Specifically, the author studies the large - deviation behavior of the probability that the covering time $C_N$ is less than its expected value $\gamma\cdot t_{cov}$, where $\gamma\in(0, 1)$.
### Specific problems the paper attempts to solve:
1. **Lower bound estimation**:
- The author proves that when $\gamma\in(\frac{d + 2}{2d}, 1)$, the exact asymptotic form of the probability that the covering time $C_N$ is less than $\gamma$ times the expected value. The specific result is:
\[
\lim_{N\rightarrow\infty}\frac{\log P(C_N\leq\gamma t_{cov})}{N^d(1 - \gamma)}=- 1
\]
- This result shows that in high - dimensional cases, small - scale deviations of the covering time have exponential decay characteristics.
2. **Lower bound for the full range $\gamma\in(0, 1)$**:
- For a wider range of $\gamma\in(0, 1)$, the author gives a weaker but still meaningful lower - bound estimate:
\[
P(C_N\leq\gamma t_{cov})\geq\exp(-C\cdot N^d(1 - \gamma))
\]
where $C>0$ is a constant.
3. **Upper bound estimation**:
- The author also provides a rough upper - bound estimate for large - scale deviations of the covering time:
\[
P(C_N\leq\gamma t_{cov})\leq\exp(-N^d(1 - \gamma)+o(1))
\]
- Combining the upper - and lower - bound estimates, the following asymptotic form can be obtained:
\[
P(C_N\leq\gamma t_{cov})=\exp(-N^d(1 - \gamma)+o(1))
\]
4. **Upper - shift probability**:
- For the case of $\gamma>1$, the author also studies the probability that the covering time exceeds $\gamma$ times the expected value:
\[
P(C_N\geq\gamma t_{cov})=(1 + o(1))N^{-d(\gamma - 1)}
\]
### Research methods and tools:
- **Random interlacements model**: The author uses the random interlacements model to analyze the trajectories of random walks. In particular, through the strong coupling theorem, the random walks on the discrete torus are coupled with random interlacements.
- **Soft local time technique**: The "soft local time" technique is introduced to decouple the trajectories of random walks, so as to better estimate the deviation probability of the covering time.
- **Phased strategy**: The probability of large - scale deviation of the covering time is analyzed and estimated step by step through a four - stage strategy, including free walks, partitioning regions, surgical insertion of loops, etc.
### Summary:
By combining the random interlacements model and the soft local time technique, this paper successfully solves the problem of large - scale deviation of the covering time of simple random walks on high - dimensional discrete tori and provides accurate asymptotic estimates. This not only deepens the understanding of the covering time of random walks but also provides a theoretical basis for further research in related fields.