Tensor renormalization group approach to the $O(2)$ models via symmetry-twisted partition functions

This paper demonstrates that symmetry-twisted partition functions computed via the tensor renormalization group framework effectively detect spontaneous symmetry breaking in three-dimensional $O(2)$ models, determine the BKT transition in two dimensions, and successfully identify phase transitions in generalized two-dimensional $O(2)$ models.

The Gist

Imagine you are trying to understand how a crowd of people behaves in a giant, invisible room. Sometimes, everyone stands still and faces random directions (chaos). Other times, they all suddenly decide to face the same way (order). In physics, we call these "phases," and the moment they switch from one to the other is a "phase transition."

This paper is about a new, super-smart way to predict exactly when these switches happen, especially for systems that are notoriously difficult to study with traditional computer methods.

Here is the breakdown using simple analogies:

1. The Problem: The "Sign Problem" and the Old Way

Traditionally, physicists use a method called Monte Carlo to simulate these systems. Imagine trying to guess the average mood of a crowd by asking random people. It works well most of the time. However, in certain complex situations (like quantum mechanics), the math gets so weird that the "moods" cancel each other out, making the calculation impossible. This is called the "sign problem." It's like trying to count a crowd where half the people are invisible and the other half are ghosts; you can't get a real number.

2. The New Tool: Tensor Networks (The Lego Approach)

The authors use a different tool called Tensor Renormalization Group (TRG). Instead of asking random people, imagine the whole room is built out of Lego blocks. Each block represents a tiny piece of the system.

• The Magic: You can snap these blocks together in a specific pattern to see the big picture without needing to ask every single person.
• The Benefit: This method doesn't get confused by the "sign problem." It works even when the math is complex and weird.

3. The Secret Sauce: "Symmetry-Twisted" Partition Functions

The core innovation of this paper is a trick they call "Symmetry-Twisting."

Imagine the room has a rule: "Everyone must face North."

• The Normal State: Everyone faces North.
• The Twist: Now, imagine you gently twist the room itself. You tell the people on the East wall, "You must face East," but you tell the people on the West wall, "You must face West," while the middle people try to compromise.

In physics, this is called a twist.

• If the system is chaotic (high temperature), the people don't care about the twist. They just face random directions. The "cost" of the twist is high, and the system resists.
• If the system is ordered (low temperature), the people are so coordinated that they can easily adjust to the twist. The system flows with the twist.

By measuring how much the system "resists" or "accepts" this twist, the authors can tell if the system is ordered or chaotic. This measurement is called the Helicity Modulus (or superfluid density), which is essentially a measure of how "stiff" the order is.

4. What They Discovered

The team applied this "twist" trick to three different scenarios:

A. The 3D Model (The Big Switch)

In a 3D world, there is a specific temperature where the system suddenly snaps from chaos to order.

• The Result: They used their twist method to find this exact temperature. It's like finding the exact moment a pot of water starts boiling. They also calculated how "sharp" that switch is (a number called the critical exponent), which matches perfectly with other known theories.

B. The 2D Model (The BKT Transition)

In a 2D world (like a flat sheet), things are weirmer. The system never fully "freezes" into a perfect order, but it enters a special state called the BKT phase (named after Berezinskii, Kosterlitz, and Thouless).

• The Analogy: Imagine a dance floor. In the chaotic phase, everyone dances randomly. In the BKT phase, everyone dances in pairs, holding hands, but the pairs keep breaking and reforming.
• The Result: The "twist" method allowed them to detect the exact moment the pairs start forming. They found the "universal jump"—a sudden change in the system's stiffness that signals this transition.

C. The Generalized Model (The Shape-Shifter)

Finally, they looked at a more complex version where the "people" can align in specific patterns (like a square or a triangle) rather than just a circle.

• The Result: They found that by changing the angle of their "twist," they could detect two different transitions:
  1. From chaos to a "nematic" phase (where people align in a specific direction but not a specific spot).
  2. From that nematic phase to a fully ordered "ferromagnetic" phase.
• Why it matters: Previous methods struggled to see the second transition clearly. The "twist" method saw it perfectly, acting like a high-powered flashlight in a dark room.

The Big Picture

Think of this paper as inventing a new thermometer for the quantum world.

• Old thermometers (Monte Carlo) break when the physics gets too complex.
• This new thermometer (Symmetry-Twisted TRG) works everywhere.
• By "twisting" the rules of the game, the authors can peek inside the system and see exactly when and how it changes its mind, solving problems that have been stuck for a long time.

This approach is faster, more accurate, and opens the door to studying even stranger materials and quantum states in the future.

Technical Summary

1. Problem Statement

The paper addresses the challenge of detecting phase transitions associated with the spontaneous breaking of continuous global symmetries (specifically $O(2)$ or $U(1)$) in lattice field theories.

• Limitations of Existing Methods: Traditional Monte Carlo (MC) methods often struggle with the "sign problem" in complex actions. Furthermore, standard order parameters for continuous symmetry breaking (like magnetization) vanish in the thermodynamic limit due to the Mermin-Wagner theorem in 2D or require careful finite-size scaling in 3D.
• Limitations of Tensor Networks: While Tensor Renormalization Group (TRG) methods avoid the sign problem, extracting low-energy symmetry realization data has historically relied on the Gu–Wen ratio (a ratio of partition functions on different torus sizes). However, the Gu–Wen ratio is primarily effective for discrete symmetry breaking. Applying it to continuous symmetries is less direct.
• Goal: The authors aim to utilize symmetry-twisted partition functions within the TRG framework to serve as robust order parameters for continuous symmetry breaking, specifically targeting the $O(2)$ model in 2D and 3D, and its generalized variants.

2. Methodology

The authors employ the Tensor Renormalization Group (TRG) method, a real-space renormalization group approach using tensor networks.

• Symmetry-Twisted Partition Functions ($Z_{g_0}$):

  • Instead of standard periodic boundary conditions, the authors introduce a background gauge field $A$ characterized by a holonomy $g_0 \in G$ (where $G=U(1)$) along one cycle of the torus.
  • The twisted partition function is defined as $Z_{g_0} = Z[A=(1, \dots, 1, g_0)]$.
  • Theoretical Basis: In the thermodynamic limit, the ratio $Z_{g_0}/Z_1$ (where $Z_1$ is the untwisted partition function) acts as an order parameter:
    • Symmetric Phase: $Z_{g_0}/Z_1 = 1$ for all $g_0$.
    • Symmetry-Broken Phase: $Z_{g_0}/Z_1 = 0$ if $g_0$ is not in the unbroken subgroup $H$, and $1$ if $g_0 \in H$.
  • This creates a discontinuity at the critical point, allowing for precise detection of phase transitions.

• Implementation in TRG:

  • The fundamental tensors are constructed using a truncated character expansion to ensure finite bond dimensions while preserving global symmetry.
  • Symmetry Blocking: The TRG algorithms (specifically Anisotropic TRG for 3D and Bond-Weighted TRG for 2D) are extended to incorporate symmetry blocking, allowing the twist to be imposed directly on the tensor network contraction.
  • Helicity Modulus: In 2D, where spontaneous symmetry breaking does not occur (BKT transition), the ratio $Z_\alpha/Z_0$ is analytically related to the helicity modulus ($\Upsilon_\alpha$), which serves as the order parameter for the BKT transition.

• Models Studied:

  1. Standard 3D $O(2)$ Model: Investigating spontaneous symmetry breaking at finite temperature.
  2. Standard 2D $O(2)$ Model: Investigating the Berezinskii–Kosterlitz–Thouless (BKT) transition.
  3. Generalized 2D $O(2)$ Model: A model with Hamiltonian $H = -\Delta \sum \cos(\Delta \theta) - (1-\Delta) \sum \cos(q \Delta \theta)$, featuring ferromagnetic, nematic, and paramagnetic phases.

3. Key Contributions

1. Extension of Gu–Wen Logic to Continuous Symmetries: The paper demonstrates that symmetry-twisted partition functions can effectively detect the spontaneous breaking of continuous $U(1)$ symmetry, a task where the standard Gu–Wen ratio is less direct.
2. Direct Access to Helicity Modulus: The method provides a direct route to calculating the helicity modulus in 2D systems without needing external symmetry-breaking fields, which is computationally advantageous.
3. Application to Generalized Models: The approach is successfully applied to a generalized $O(2)$ model with $q=2$, distinguishing between ferromagnetic-nematic and nematic-paramagnetic transitions using specific twist angles.

4. Key Results

A. Three-Dimensional $O(2)$ Model

• Critical Temperature ($T_c$): By analyzing the ratio $Z_\pi/Z_0$ (twist angle $\alpha=\pi$) across various lattice sizes ($L$), the authors identified a scale-invariant crossing point.
• Finite-Size Scaling: Performing a scaling analysis yielded:
  • $T_c = 2.2017(2)$
  • Critical exponent $\nu = 0.663(33)$
• Significance: This represents the first precise determination of the critical exponent $\nu$ for the 3D $O(2)$ model using TRG, with results consistent with recent Monte Carlo studies.

B. Two-Dimensional $O(2)$ Model (BKT Transition)

• Helicity Modulus: The ratio $Z_\alpha/Z_0$ was used to compute the helicity modulus $\Upsilon_\alpha$.
• Universal Jump: The authors observed the universal jump in $\Upsilon$ predicted by the Nelson-Kosterlitz (NK) criterion.
• Transition Point: By extrapolating the crossing temperatures of $\Upsilon_{\pi/6}$ with the NK formula ($2T/\pi$) to the thermodynamic limit ($L \to \infty$), they determined:
  • $T_{BKT} = 0.8927(5)$
• Significance: This result is in excellent agreement with previous literature, validating the twisted partition function as a reliable probe for BKT transitions.

C. Generalized 2D $O(2)$ Model ($q=2$)

• Phase Identification:
  • Ferromagnetic-Nematic Transition: Detected via a sharp jump in $Z_\pi/Z_0$ (twist $\alpha=\pi$), indicating $Z_2$ symmetry breaking.
  • Nematic-Paramagnetic Transition: Detected via the smooth variation of $Z_\alpha/Z_0$ for non-integer multiples of $\pi$ (e.g., $\alpha=\pi/6$), indicating a BKT transition.
• Critical Points:
  • Ferromagnetic-Nematic ($T_c$): $0.35253(7)$ (significantly more precise than previous specific heat analyses).
  • Nematic-Paramagnetic ($T_{BKT}$): $0.7581(4)$.
• Methodological Advantage: The BKT transition was identified without introducing an explicit external symmetry-breaking field, reducing computational cost and improving accuracy compared to previous TRG studies.

5. Significance and Outlook

• Robustness: The study establishes symmetry-twisted partition functions as a powerful, universal tool for identifying phase transitions in systems with continuous symmetries, overcoming limitations of traditional order parameters in tensor network simulations.
• Efficiency: By avoiding the need for external fields to probe BKT transitions, the method reduces computational overhead.
• Future Directions: The authors plan to extend this analysis to generalized $O(2)$ models with $q > 4$ (where unusual ferromagnetic phases are expected) and to investigate these models in three dimensions, a regime not yet explored via TRG.

In conclusion, this paper successfully bridges the gap between tensor network methods and the study of continuous symmetry breaking, providing high-precision critical data for $O(2)$ models that complements and, in some cases, surpasses existing Monte Carlo results.