Diagonal boundary conditions in critical loop models

This paper utilizes analytic bootstrap methods to define and characterize diagonal boundaries in critical loop models via a complex parameter, deriving explicit formulas for disc correlation functions and demonstrating that specific parameter values yield discrete spectra of degenerate representations, while also providing a lattice interpretation where loops cannot terminate or change weight upon touching such boundaries.

The Gist

Imagine a vast, infinite floor covered in a giant, tangled web of non-intersecting rubber bands. In the world of physics, this is a "loop model." These loops aren't just random; they represent the behavior of things like polymers (long chain molecules) or the paths taken by water spreading through soil (percolation). When these systems are at a "critical" point—meaning they are perfectly balanced between order and chaos—they become incredibly beautiful and mathematically rich.

This paper is about what happens when you put a wall around this floor of loops. Specifically, the authors are figuring out the rules for how these loops behave when they hit a special kind of wall called a "diagonal boundary."

Here is the breakdown of their discovery, using everyday analogies:

1. The Two Types of Walls

Imagine you are walking a dog on a leash (the loop) in a park. You approach a fence (the boundary).

• Non-Diagonal Walls: These are like a fence with a gate. The dog can run through the gate, or the leash can change length or color when it touches the fence. In physics terms, the loop can "end" on the wall or change its properties.
• Diagonal Walls (The Focus of this Paper): These are like a solid, magical wall. The dog cannot end its walk on the wall, and the leash cannot change its length or color when it touches the wall. The loop must simply bounce off or slide along it, keeping its "identity" intact.

The authors call these "diagonal" because, in the complex math behind the scenes, they only interact with specific, "symmetric" types of fields (like a mirror image of itself).

2. The "Recipe" for the Wall

The authors wanted to know: If I build this special diagonal wall, what are the rules?

They used a method called the "Bootstrap" (think of it as pulling yourself up by your own bootstraps). Instead of building the wall from scratch with bricks, they started with the rules of the loops themselves and asked, "What kind of wall is mathematically possible?"

They found that every diagonal wall is defined by just one number (a complex parameter, $\sigma$).

• Analogy: Think of this number as a "volume knob" or a "dial" on the wall. Turning the dial changes how the loops interact with the wall, but the wall remains a "diagonal" wall.
• They discovered that for most settings of this dial, the wall is "continuous" (smooth and fluid). But for specific, discrete settings (like turning the dial to exact integer numbers), the wall becomes "discrete" (rigid and specific).

3. The "Legs" of the Loops

In these models, loops are often visualized as having "legs" sticking out of them (like a spider with legs).

• The Big Discovery: The authors proved that on a diagonal wall, loops can never lose a leg.
• Analogy: Imagine a spider walking on a wall. If it's a diagonal wall, the spider can walk along it, or it can gain extra legs (maybe 2, 4, or 6 more), but it can never lose a leg. It can never stop walking and just "stick" to the wall as a dead end.
• This is a strict rule: The number of legs is conserved or increases by even numbers. It can never decrease. This explains why the loops can't "end" on the wall—they would have to lose legs to do so, which is forbidden.

4. The Math Magic (The "Recipe Book")

The authors didn't just guess these rules; they wrote down the exact mathematical "recipes" (formulas) for how likely it is to find loops in certain positions on a circular floor (a "disc").

• They calculated the probability of finding one loop (1-point function) and two loops (2-point function) near the wall.
• They found that for the "discrete" walls (the rigid ones), the math simplifies beautifully, and the possible states of the system become a finite, countable list, much like the notes on a piano scale, rather than a continuous slide.

5. Checking the Work

To make sure their "recipes" were correct, they used two methods:

1. Analytic Math: They checked if the formulas made sense with the laws of symmetry (Crossing Symmetry). It's like checking if a puzzle piece fits perfectly from two different angles.
2. Computer Simulation: They built a digital version of the loop model on a computer and ran millions of simulations. The results matched their formulas perfectly, down to tiny decimal places.

Summary

In short, this paper defines a specific, rigid type of boundary for a complex system of tangled loops. They found that:

1. These walls are controlled by a single "dial."
2. On these walls, loops cannot end or lose their "legs"; they can only slide or gain legs.
3. They provided the exact mathematical formulas to predict how these loops behave near the wall.
4. They showed how to build these walls in real-world lattice models (like grids of atoms) using specific mathematical tools called "Jones-Wenzl projectors."

The paper is a fundamental step in understanding how complex systems behave when they hit a boundary that respects their internal symmetry, solving a long-standing puzzle in the physics of critical phenomena.

Technical Summary

Technical Summary: Diagonal Boundary Conditions in Critical Loop Models

Problem Statement
Critical loop models in two dimensions describe phenomena such as percolation and polymer physics. While bulk properties of these models are accessible via integrable lattice techniques and conformal field theory (CFT) methods, the inclusion of boundaries remains largely unsolved. Specifically, the existence of operator product expansions (OPEs) and the structural properties of boundary conditions in critical loop models are not fully understood. Previous work has established specific boundary observables (e.g., Cardy's and Schramm's probabilities), but a general framework for families of boundary conditions and their associated correlation functions is lacking. This work addresses the classification of boundary conditions and the computation of disc correlation functions within the analytic bootstrap framework.

Methodology
The authors employ analytic bootstrap methods, focusing on the constraints imposed by crossing symmetry and the existence of degenerate fields.

• Degenerate Fields: The analysis relies on the degenerate bulk field $V^d_{\langle 1,2 \rangle}$, whose OPEs with non-degenerate fields are known. This allows for the derivation of shift equations for correlation functions.
• Boundary Classification: The paper distinguishes between "diagonal" and "non-diagonal" boundary conditions based on the bulk-to-boundary OPE of the degenerate field. A diagonal boundary is defined by a non-zero coefficient $\mu$ coupling the degenerate field to the boundary identity, implying that only diagonal or degenerate fields have non-zero disc 1-point functions.
• Field Renormalization: To simplify the shift equations, the authors perform specific field renormalizations. This ensures that the coefficients in the shift equations for disc 1-point functions are trivial (i.e., equal to 1), facilitating the derivation of explicit solutions.
• Modular Bootstrap: The boundary spectrum is deduced using the modular covariance of the annulus partition function. By transforming the partition function from the bulk channel to the boundary channel, the authors determine conditions under which the spectrum is discrete.
• Numerical Bootstrap: For discrete boundary conditions, the authors adapt numerical bootstrap techniques to solve the crossing symmetry equations for disc 2-point functions. They truncate the spectrum and solve linear systems to verify the consistency of their analytic conjectures.

Key Contributions and Results

1. Definition and Characterization of Diagonal Boundaries:
   The authors define diagonal boundaries as those coupling only to diagonal fields. They show that such boundaries are characterized by a single complex parameter $\sigma$. The disc 1-point function for a diagonal field $V_P$ is derived as:
   $$\langle V_P \rangle_\sigma = \sin(4\pi\sigma P)$$
   This solution formally coincides with disc 1-point functions in Liouville theory with $c \le 1$, despite loop models existing in the domain $\Re c < 13$.

2. Discrete vs. Continuous Boundaries:
   A criterion is proposed to distinguish discrete from continuous boundaries. A boundary is discrete if and only if the parameter $\sigma$ takes values in the set $\{P(0, S) \mid S \in \mathbb{N}^*\}$. For discrete boundaries, the boundary spectrum consists of degenerate representations $\psi^d_{\langle r,s \rangle}$ with specific constraints on the Kac indices.

3. Bulk 2-Point Functions on the Disc:
   The authors derive an explicit formula for the bulk 2-point function $\langle V_{(r_1,s_1)} V_{(r_2,s_2)} \rangle_\sigma$ in the bulk channel. This is expressed as an infinite sum of Virasoro conformal blocks weighted by bulk OPE coefficients and the disc 1-point functions. The bulk OPE coefficients are conjectured to coincide with known 3-point structure constants of the bulk CFT, modified by the field renormalization factors.

4. Boundary Channel Decomposition and Numerical Verification:
   The bulk channel decomposition is tested against the boundary channel decomposition using numerical bootstrap methods.

   • Uniqueness: For discrete boundaries, the crossing symmetry equations admit a unique solution.
   • Spectrum Reduction: The boundary channel spectrum is found to be a subset of the full boundary spectrum, specifically $\{ \psi^d_{\langle r,s \rangle} \mid r \in 2\max(r_1, r_2) + 1 + 2\mathbb{N}, s \in \{1, 3, \dots, 2S-1\} \}$.
   • Vanishing Structure Constants: Many boundary structure constants $D_k$ are found to vanish, consistent with the conjectured bulk-to-boundary OPEs where the number of legs is conserved or increased by even numbers, but never decreased.

5. Lattice Interpretation:
   The paper sketches the interpretation of these results in lattice loop models.

   • Diagonal Boundaries: Correspond to boundaries that preserve the global symmetry of the loop model. In the dense phase, these are constructed using Jones–Wenzl projectors in the transfer matrix. In the dilute phase, they correspond to "special" boundary conditions where monomer weights are tuned.
   • Physical Constraints: A loop cannot end on a diagonal boundary, nor can it change its weight upon touching it. In bulk-to-boundary OPEs, the number of legs (associated with the first Kac index) can be conserved or increase by even numbers, but cannot decrease.

Significance and Claims
The paper claims to provide the first analytic determination of disc 2-point functions for critical loop models with diagonal boundaries, expressed as infinite combinations of conformal blocks. It establishes that diagonal boundaries are solvable within the bootstrap framework, yielding results analogous to Liouville theory. The authors emphasize that while the bulk OPEs in critical loop models are not fully known, the specific structure of diagonal boundaries allows for the computation of disc 2-point functions without requiring the full bulk OPE.

The work modestly notes that the interpretation of these results in lattice models is a sketch, with a more complete analysis reserved for future work. It also highlights that non-diagonal boundaries remain a significant challenge due to the lack of shift equations constraining the disc 1-point functions for non-diagonal fields and the complexity of their combinatorial structures. The paper concludes that the space of diagonal boundaries is well-understood via the parameter $\sigma$, but solving the full critical loop model on the disc requires further understanding of boundary fields and their structure constants.