Table of Contents
Fetching ...

Almost Sure Convergence of Liouville First Passage Percolation

Charles Devlin VI

Abstract

Liouville first passage percolation (LFPP) with parameter $ξ> 0$ is the family of random distance functions (metrics) $(D_h^ε)_{ε> 0}$ on $\mathbb{C}$ obtained heuristically by integrating $e^{ξh}$ along paths, where $h$ is a variant of the Gaussian free field. There is a critical value $ξ_{\text{crit}} \approx 0.41$ such that for $ξ\in (0, ξ_{\text{crit}})$, appropriately rescaled LFPP converges in probability uniformly on compact subsets of $\mathbb{C}$ to a limiting metric $D_h$ on $γ$-Liouville quantum gravity with $γ= γ(ξ) \in (0,2)$. We show that the convergence is almost sure, giving an affirmative answer to a question posed by Gwynne and Miller (2019).

Almost Sure Convergence of Liouville First Passage Percolation

Abstract

Liouville first passage percolation (LFPP) with parameter is the family of random distance functions (metrics) on obtained heuristically by integrating along paths, where is a variant of the Gaussian free field. There is a critical value such that for , appropriately rescaled LFPP converges in probability uniformly on compact subsets of to a limiting metric on -Liouville quantum gravity with . We show that the convergence is almost sure, giving an affirmative answer to a question posed by Gwynne and Miller (2019).
Paper Structure (14 sections, 20 theorems, 142 equations, 2 figures)

This paper contains 14 sections, 20 theorems, 142 equations, 2 figures.

Key Result

Theorem 1.1

Almost surely, $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} D_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} D_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} D_{-NoValue-}^{-NoValue-} } \to D_h$ uniformly on compact

Figures (2)

  • Figure 1: Illustration of Definition \ref{['defn:AnnulusEvents']}. Condition 1 says that if the red path is a $D_h$-geodesic, then $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} } (u,v) \leq \frac{r \mathfrak{a}_{{\epsilon}}^{-1}}{r^{\xi Q} \mathfrak{a}_{{\epsilon}/r}^{-1}}(1 + \delta) D_h(u,v; \mathbb{A}_{r/2, 2r}(z))$, and analogously with $D_h$ and $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} }$ swapped. Condition 2 says that if the $D_h$-distance from $u$ to $v$ is larger than the $D_h$-distance from $u$ to the boundary of the gray annulus, then the red path cannot be a $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} }$-geodesic, and analogously with $D_h$ and $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} }$ swapped. Condition 3 says there is a path (shown in orange) around the cyan annulus with $D_h$-length at most $A$ times the $D_h$-distance across said annulus, and analogously with $D_h$ and $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} }$ swapped.
  • Figure 2: Illustration of the proof of Proposition \ref{['prop:MainIterationArgument']}. Pictured is one of the annuli $\mathbb{A}_{\alpha r_{j-1}, r_{j-1}}(x_{j-1})$ such that $E_{r_{j-1}, \epsilon}(x_{j-1})$ occurs. The path in red is the $D_h$-geodesic $P$. Time $t_{j}$ is the first time after time $t_{j-1}$ that $P$ exits $B_{r_{j-1}}(x_{j-1})$, and $s_j$ is the last time before $t_j$ that $P$ leaves $B_{\alpha r_{j-1}}(x_{j-1})$. The segment $P|_{[s_j, t_j]}$ is a $D_h$-geodesic in $\mathbb{A}_{\alpha r_{j-1}, r_{j-1}}(x_{j-1})$, so condition 1 in Definition \ref{['defn:AnnulusEvents']} says $\IfNoValueTF{-NoValue-} { \IfNoValueTF{-NoValue-} { \mathfrak{a}_{\epsilon}^{-1} \hat{D}_h^{\epsilon} } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_h^{-NoValue-} } } { \mathfrak{a}_{-NoValue-}^{-1} \hat{D}_{-NoValue-}^{-NoValue-} } (P(s_j), P(t_j)) \leq \frac{r_{j-1} \mathfrak{a}_{\epsilon}^{-1}}{r_{j-1}^{\xi Q} \mathfrak{a}_{\epsilon/r_{j-1}}^{-1}} (1 + \delta) (t_j - s_j)$. The orange path is from condition 3 in Definition \ref{['defn:AnnulusEvents']}, and has $D_h$-length at most $A D_h(\partial B_{\alpha r_{j-1}}(x_{j-1}), \partial B_{r_{j-1}}(x_{j-1})) \leq A (t_j - s_j)$. Since $P$ crosses the orange path before time $t_{j-1}$ and after time $s_j$, it follows that $s_j - t_{j-1} \leq A(t_j - s_j)$. We use the latter to show that a positive proportion of $P$ is comprised of the segments $P|_{[s_j, t_j]}$.

Theorems & Definitions (39)

  • Theorem 1.1
  • Remark 1.2
  • Theorem 1.3
  • Theorem 1.4
  • Lemma 2.1
  • Lemma 2.2
  • proof
  • Proposition 2.3
  • Proposition 3.1
  • Definition 3.2
  • ...and 29 more