🔢 mathematics

Two-sided infinite self-avoiding walk in high dimensions

This paper constructs a two-sided infinite self-avoiding walk in dimensions $d \geq 5$ to prove pattern theorems and establish the existence of infinite SAWs via ergodicity and coupling techniques, achieving these results without relying on the lace expansion.

Original authors: Maarten Markering

Published 2026-10-07

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Maarten Markering

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a traveler wandering through a vast, infinite grid of city blocks, taking one step at a time. This traveler has a single, strict rule: they can never visit the same intersection twice. If a path requires stepping on a corner they have already seen, that path is simply invalid and cannot be taken. This simple constraint creates a complex, winding journey known in mathematics as a self-avoiding walk. While the rule is easy to state, predicting the traveler's long-term behavior is notoriously difficult. In low dimensions, like a flat sheet of paper or a three-dimensional room, it remains an open mathematical question whether such a traveler can continue forever without getting trapped, making it hard to say what happens after an infinite amount of time. However, in higher dimensions, where there is more room to move, the traveler can keep going forever. For decades, mathematicians have struggled to describe the statistical properties of this endless journey, particularly how often the traveler repeats certain patterns of movement.

A new study by Maarten Markering from the University of Cambridge finally provides a clear picture of this infinite journey in high-dimensional spaces, specifically for dimensions five and above. The researcher constructed a mathematical model of a two-sided infinite walk, which can be thought of as a path that extends infinitely in both directions, rather than just starting at a point and moving forward. By creating this two-sided version, Markering was able to prove that the traveler's path eventually settles into a predictable rhythm. The study demonstrates that as the walk gets longer, the frequency of any specific pattern of turns and steps stabilizes. In other words, if you watch the traveler for a long enough time, the proportion of times they make a specific sequence of moves will converge to a fixed number, just as a coin flip eventually settles to a 50-50 split between heads and tails. This result confirms that the infinite walk is "ergodic," meaning that its long-term behavior is uniform and independent of where it started, a property that had been difficult to prove without relying on extremely complex and difficult-to-apply mathematical tools.

The significance of this work lies in how it was achieved. Previous proofs of similar results for high-dimensional walks relied on a sophisticated and intricate method called the lace expansion, which is often described as a heavy hammer that is difficult to wield. Markering's approach, by contrast, is more direct and relies only on the basic statistics of how many paths exist and how likely they are to hit specific points. By avoiding the lace expansion, the study offers a new, cleaner proof for the existence of the infinite walk in dimensions five and above, a fact that was already established by Hara and Slade using the lace expansion. Furthermore, the researcher used this new understanding to propose a different way of defining the infinite walk itself, based on a property called the domain Markov property, which essentially says that the future path depends only on the current position and the history of where the traveler has already been, not on the specific route taken to get there. This alternative definition is shown to be equivalent to the standard one in high dimensions, providing a new foundation for future research.

The paper also addresses a lingering question about whether these infinite walks exist in lower dimensions, such as three or four. While the study does not prove their existence in those lower dimensions, it establishes a clear framework for what such a walk would look like if it did exist. The researcher proposes a conjecture that a unique, stable path might exist even in three or four dimensions, provided it satisfies certain symmetry and memory-less properties. Although the author is unable to prove this conjecture in lower dimensions, they provide some supporting evidence. The results confirm that in dimensions five and higher, the infinite walk is not just a theoretical possibility but a well-defined object with stable statistical properties. This means that for high-dimensional spaces, we can now confidently say that the traveler's path will not get stuck in a chaotic loop of randomness but will instead follow a consistent statistical law. The findings suggest that the complex, tangled nature of the walk eventually smooths out, allowing the traveler to explore the grid in a way that is both free and governed by simple, underlying rules. This clarity opens the door to understanding other similar systems in physics and mathematics, such as the behavior of long polymer chains or the structure of complex networks, where the rule of not revisiting the same point is crucial.

By constructing the two-sided infinite walk, the researcher was able to treat the path as a whole, rather than just a sequence of steps growing from a single point. This perspective allowed for the use of a technique called coupling, where two different walks are imagined to be running side by side. The study shows that if you let these two walks run long enough, they will eventually merge and follow the exact same path, regardless of how they started. This merging behavior is the key to proving that the walk is ergodic. It implies that the memory of the starting point fades away as the walk progresses, and the traveler's future becomes independent of their past. This is a profound insight because it means that the infinite walk has a kind of "forgetfulness" built into its structure, ensuring that no matter how the journey begins, it will eventually settle into a standard, predictable pattern of movement.

The study also provides a new way to calculate the likelihood of specific patterns appearing in the walk. It shows that the probability of a pattern occurring is directly related to how many ways the rest of the walk can continue without hitting itself. This relationship allows mathematicians to predict the frequency of any pattern with high precision. For example, if a specific sequence of turns is very likely to cause the walker to get trapped, it will appear less often in the long run. Conversely, patterns that allow the walker to move freely will appear more frequently. This balance between constraint and freedom is what gives the self-avoiding walk its unique character, and the new proofs clarify exactly how this balance is maintained in high dimensions.

In the broader context of mathematical physics, this work helps bridge the gap between simple models and complex reality. The self-avoiding walk is a fundamental model for understanding how long molecules, like DNA or synthetic polymers, behave in space. These molecules cannot occupy the same space twice, just like the traveler in the grid. By proving that these walks have stable, predictable properties in high dimensions, the study offers a more solid theoretical foundation for understanding the behavior of such molecules. While the real world is three-dimensional, and the behavior there is still not fully understood, the high-dimensional results provide a crucial benchmark. They show that when there is enough space, the system behaves in a regular, predictable way, which helps scientists understand what changes when the space becomes more constrained.

The paper concludes by posing a conjecture about the existence of these infinite walks in lower dimensions, suggesting that a unique, stable path might exist even in three or four dimensions, provided it satisfies certain symmetry and memory-less properties. While this remains unproven for lower dimensions, the methods developed in this study offer a new path forward. The researcher has shown that by focusing on the fundamental properties of the walk rather than relying on complex expansion techniques, it is possible to uncover deep truths about its behavior. This approach not only solves a specific problem in high dimensions but also provides a toolkit that could be applied to other difficult problems in probability and statistical physics. The result is a clearer, more intuitive understanding of how order emerges from the chaos of a path that refuses to cross its own tracks.

Technical Summary: Two-sided infinite self-avoiding walk in high dimensions

Problem Statement
The paper addresses fundamental questions regarding the self-avoiding walk (SAW) on the integer lattice $\mathbb{Z}^d$ for dimensions $d \ge 5$ . While the existence of the one-sided infinite SAW ( $\eta_\infty$ ) was established by Lawler using lace expansion techniques, many of its properties remain poorly understood, particularly regarding ergodicity and the convergence of pattern densities. Furthermore, the existence of infinite SAW in low dimensions ( $2 \le d \le 4$ ) remains an open problem. The paper seeks to construct a two-sided infinite SAW ( $\hat{\eta}$ ), prove its ergodicity with respect to the shift operator, and establish strong limit theorems (pattern theorems) for both finite and infinite SAWs without relying on the lace expansion.

Methodology
The primary methodological innovation of this work is the avoidance of the lace expansion, which has historically been the standard tool for analyzing high-dimensional SAWs. Instead, the proofs rely exclusively on:

Asymptotics of SAW counts: The known asymptotic behavior of the number of self-avoiding paths $c_n \sim A\mu^n$ and the two-point function $G(x) \sim a\|x\|^{2-d}$ for $d \ge 5$ .
Coupling Techniques: The construction of specific couplings between SAWs conditioned on different initial patterns.
Domain Markov Property (DMP): Utilizing the property that, conditional on a finite prefix, the remainder of an infinite SAW behaves like an infinite SAW starting from the endpoint, conditioned to escape the prefix.

The core technical achievement is a multi-stage coupling argument. To prove ergodicity, the author constructs a coupling between two infinite SAWs starting with different finite patterns. By iteratively sampling independent SAW segments and checking for "escape" (non-intersection with the existing path), the author demonstrates that the two walks eventually coincide with high probability. This relies on hitting estimates showing that in $d \ge 5$ , the probability of intersection between distant parts of SAWs decays sufficiently fast.

Key Contributions and Results

Construction of Two-Sided Infinite SAW ( $\hat{\eta}$ ):
The paper defines $\hat{\eta}$ as a bi-infinite path formed by two independent one-sided infinite SAWs ( $\eta_\infty^1, \eta_\infty^2$ ) conditioned to not intersect after time 0. The author proves that this object is well-defined (the non-intersection probability is strictly positive for $d \ge 5$ ) and that it arises as the weak limit of:
- Finite two-sided SAWs ( $\eta_{m,n}$ ) as lengths $m, n \to \infty$ .
- Shifted one-sided infinite SAWs ( $T^m \eta_\infty$ ) as $m \to \infty$ .
Ergodicity of Infinite SAW:
The paper proves that the one-sided infinite SAW $\eta_\infty$ is ergodic with respect to the shift operator $T$ . This is established by showing that events depending on the walk at time $0$ and time $m$ become asymptotically independent as $m \to \infty$ . The proof utilizes the coupling technique to show that the total variation distance between shifted walks starting from different initial conditions tends to zero.
Pattern Theorems and Limit Laws:
The author proves that for any finite pattern $\zeta$ , the fraction of times $\zeta$ occurs in a SAW converges to a deterministic limit. Specifically:
- For finite SAW ( $\eta_n$ ): The fraction of occurrences $\pi_\zeta(n, \eta_n)$ converges in probability to $P(\hat{\eta}[0, k] = \zeta)$ .
- For infinite SAW ( $\eta_\infty$ ): The fraction of occurrences converges almost surely to the same limit.
  This result generalizes Kesten's Pattern Theorem (which provided lower bounds on the density of patterns) to a precise law of large numbers for $d \ge 5$ .
Alternative Characterization and Existence Proof:
The paper proposes a conjecture that there exists a unique symmetric probability measure on infinite paths satisfying the Domain Markov Property (DMP). For $d \ge 5$ , the author proves this conjecture, providing a new proof for the existence of the infinite SAW that does not invoke the lace expansion. This characterization defines the infinite SAW as the unique measure satisfying the DMP and specific escape probabilities.

Significance and Claims
The paper claims to provide a new, self-contained framework for analyzing high-dimensional SAWs that bypasses the technically demanding lace expansion. By relying only on the asymptotic behavior of $c_n$ and $G(x)$ , the results suggest that if similar mean-field behaviors are established for other models (like weakly self-avoiding walks or spread-out percolation) without the lace expansion, the results of this paper would immediately extend to those models.

The construction of the two-sided infinite SAW serves as a central tool, allowing the author to interpret the infinite SAW as a limit "seen from the middle." This perspective facilitates the proof of ergodicity and the derivation of strong limit theorems. The paper modestly notes that while it resolves these questions for $d \ge 5$ , the existence of infinite SAW in dimensions $2 \le d \le 4$ remains open, though the proposed DMP-based characterization offers a potential avenue for future investigation.

The work also highlights that the convergence of pattern densities is a consequence of the asymptotic independence of distant segments of the walk, a property rigorously established here via coupling rather than assumed.

More like this