⚛️ lattice

Non-stabilizerness and entanglement in $(2+1)$ -dimensional SU(2) lattice gauge theory using tensor networks

Using matrix product states to study $(2+1)$ -dimensional SU(2) lattice gauge theory, this paper proves a stronger lower bound for non-local magic and demonstrates that this intrinsic, encoding-independent quantity serves as a more robust and computationally efficient probe of gauge-matter delocalization than traditional entanglement measures.

Original authors: Raghav G. Jha, Jaber I. Taher, Muhammad Asaduzzaman, Goksu C. Toga, Bojko N. Bakalov, Alexander F. Kemper

Published 2026-09-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Raghav G. Jha, Jaber I. Taher, Muhammad Asaduzzaman, Goksu C. Toga, Bojko N. Bakalov, Alexander F. Kemper

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

The universe is built on a set of fundamental forces that govern how particles interact, from the glue holding atomic nuclei together to the light emitted by distant stars. Physicists describe these interactions using quantum field theories, which are mathematical frameworks that treat particles as excitations in underlying fields. While these theories are incredibly successful at predicting what happens in the lab, solving their equations for complex situations is often impossible for even the most powerful classical computers. This is because the quantum states involved can be incredibly intricate, containing a type of complexity that goes beyond simple entanglement. Entanglement, a phenomenon where particles remain connected regardless of distance, is well understood and can sometimes be simulated efficiently. However, there is another resource required to describe these states fully, often called "magic" in the language of quantum information. This magic represents the specific kind of quantum complexity that prevents a state from being easily simulated on a classical computer, acting as a measure of how far a system is from a simple, predictable configuration. Understanding how much magic is needed to describe the ground state of a physical system is crucial for knowing what resources future quantum computers will require to simulate nature itself.

In a recent study, researchers set out to map this hidden complexity within a specific model of particle physics: a two-dimensional grid of space where particles interact with a force field, known as an SU(2) lattice gauge theory. They focused on the lowest energy state of this system, known as the ground state, and asked how the amount of magic changes as they adjusted the strength of the interaction between the particles and the force field. To do this, they used a powerful computational technique called tensor networks, which breaks down complex quantum states into smaller, manageable pieces that can be processed by a computer. They worked with a simplified version of the theory where the force-carrying particles, or gluons, are restricted to their simplest possible forms, a setup known as the hardcore-gluon approximation. This allowed them to simulate systems up to six by six grid points, a size large enough to reveal important patterns but small enough to be handled with current methods.

The team measured three different aspects of this quantum complexity. First, they looked at the total amount of magic in the entire system. Second, they isolated the "non-local" magic, which is the specific complexity that arises only from the connections between different parts of the grid, stripping away any local quirks that depend on how the computer represents the data. Third, they calculated a mathematical lower bound for this non-local magic, a minimum threshold that the complexity must exceed. By varying the strength of the interaction, they watched how these numbers changed. They found that as the interaction strength was tuned, the system underwent a distinct transition. At strong interactions, the particles and the force field remained tightly bound to their specific locations on the grid. As the interaction weakened, the system shifted into a new phase where the particles and the field became delocalized, spreading out and ordering themselves across the grid. This shift, known as the gauge-matter delocalization crossover, happened around a specific interaction strength value of approximately one.

What made this discovery particularly significant was the behavior of the different measurements. The researchers found that the non-local magic acted as a much sharper and more sensitive detector of this transition than the total magic or the standard measure of entanglement. While the total magic and entanglement required very large computational resources to show a clear signal, the non-local magic revealed the transition clearly even when the computer's memory was limited. This suggests that the specific type of complexity tied to the connections between different parts of the system is the most reliable indicator of this physical change. Furthermore, the team proved a new mathematical relationship showing that this non-local magic is always bounded from below by a property of the system's internal structure, confirming that their measurements were consistent with fundamental theoretical limits.

The study also highlighted a subtle but important distinction in how these quantum properties are measured. Because the way a computer translates physical particles into digital bits is not unique, some measurements can change depending on the translation method used. The researchers demonstrated that the non-local magic and their lower bound are immune to this translation choice, making them intrinsic properties of the physical state itself, whereas the total magic can vary based on the computer's setup. This robustness makes the non-local measure a preferred tool for understanding the physics of these systems. The results suggest that even with a simplified model, the system captures the essential physics of the transition, and that as the model is refined to more closely resemble the real world, the amount of magic required to describe the system will likely grow even larger. This implies that simulating the full complexity of nature will demand substantial quantum resources, reinforcing the idea that quantum computers will be essential for exploring the deepest layers of physical reality.

Technical Summary: Non-stabilizerness and Entanglement in (2 + 1)-dimensional SU(2) Lattice Gauge Theory

Problem Statement
The simulation of quantum field theories, particularly non-Abelian lattice gauge theories, on quantum computers requires a precise understanding of the quantum resources necessary to represent and evolve physical states. While entanglement entropy has long served as a primary diagnostic for classical simulability, it is insufficient on its own, as highly entangled stabilizer states remain efficiently simulable classically. Non-stabilizerness, or "magic," has emerged as a complementary resource that quantifies the deviation of a state from the stabilizer polytope. However, numerical studies of magic in non-Abelian gauge theories have been restricted to (1 + 1) dimensions. This work addresses the gap by investigating the ground state of (2 + 1)-dimensional SU(2) lattice gauge theory with matter, a regime where genuine magnetic (plaquette) dynamics and a gauge–matter delocalization crossover occur. The study aims to characterize the non-stabilizerness of these states and determine how these resources scale as the system approaches the continuum limit.

Methodology
The authors formulate the SU(2) Hamiltonian lattice gauge theory in the dressed-site basis within the hardcore-gluon truncation (retaining only the $j=0$ and $j=1/2$ irreducible representations). In this framework, gauge degrees of freedom are absorbed into the sites by explicitly solving Gauss's law, converting link constraints into local site constraints and link-matching conditions. The system is restricted to the zero baryon-number sector ( $b=0$ ) on square lattices with open boundary conditions, up to a size of $6 \times 6$ .

The ground states are approximated using Matrix Product States (MPS) with a one-dimensional ordering (column-major) of the two-dimensional lattice. The study computes three specific facets of magic and entanglement:

Second-order Stabilizer Rényi Entropy (SRE), $M_2$ : A measure of total non-stabilizerness.
Non-local Magic, $M^{NL}_2$ : The component of magic tied to correlations across a bipartition, isolated by minimizing over local unitaries ( $U_A \otimes U_B$ ).
Anti-flatness ( $F$ ) and a Lower Bound: A measure of the deviation of the entanglement spectrum from a flat distribution, used to derive a lower bound for non-local magic.

To compute these, the authors employ Pauli sampling for the full SRE on smaller lattices and utilize the Schmidt reference state of the central cut for non-local quantities. They also calculate standard gauge-invariant observables, including electric energy density, normalized magnetic plaquette, and gauge-invariant entanglement entropy.

Key Contributions

Theoretical Inequality: The authors prove a stronger form of the "sandwich relation" for non-local magic. They establish that for arbitrary Schmidt bases and rank, the inequality $-\log_2(1 - 4F) \leq M^{NL}_2 \leq M_2$ holds. This resolves an open problem regarding the maximal lower bound for non-local magic, improving upon previous results that were limited to specific spectral types or ranks.
Encoding Invariance: The paper demonstrates a critical structural distinction between full SRE and non-local quantities. The full SRE ( $M_2$ ) depends on the specific qubit encoding of the gauge-invariant local Hilbert space, which is non-unique. In contrast, the non-local magic ( $M^{NL}_2$ ) and the anti-flatness ( $F$ ) are invariant under site-local re-encodings. This makes non-local quantities intrinsic to the physical state and bipartition, rendering them the preferred physical diagnostics.
Numerical Implementation: The study extends tensor network simulations of magic to (2 + 1) dimensions, utilizing MPS with bond dimensions up to $\chi = 128$ .

Results

Gauge-Matter Delocalization Crossover: The non-local magic serves as a sharp probe for the crossover from a localized state (at strong coupling, large $g^2$ ) to a flux-ordered/delocalized state (at weak coupling, $g^2 \to 0$ ). The peak of non-local magic occurs at $g^2 \approx 1$ , consistent with the crossover identified in previous untruncated studies.
Convergence and Sensitivity: Non-local magic retains a clear signal at bond dimensions ( $\chi$ ) significantly lower than those required to converge the ground state energy or the full SRE. For instance, on a $6 \times 6$ lattice, the peak of non-local magic shifts from $g^2 \approx 0.8$ at $\chi=64$ to $g^2 \approx 1$ at $\chi=128$ , aligning with the expected crossover point. In contrast, the full SRE and entanglement entropy require higher $\chi$ for convergence.
Validation of Bounds: The numerical results satisfy the newly proven lower bound $-\log_2(1 - 4F) \leq M^{NL}_2$ across the entire range of coupling strengths. As the system approaches the strong-coupling limit ( $g^2 \to \infty$ ), the non-local magic approaches the lower bound, consistent with the state becoming more stabilizer-like.
Entanglement Structure: The gauge-invariant entanglement entropy is decomposed into classical, quantum, and edge contributions. The edge contribution increases with lattice size due to more severed bonds, while the quantum component shows strong dependence on bond dimension but remains small relative to other contributions.

Significance
The paper claims that non-local magic is a more efficient and physically robust diagnostic for the gauge–matter delocalization crossover than the full SRE or standard entanglement entropy, particularly in regimes where full convergence of the ground state is computationally expensive. By proving the general lower bound and establishing the encoding invariance of non-local quantities, the work provides a rigorous framework for quantifying quantum resources in higher-dimensional gauge theories.

The authors note that while the hardcore truncation ( $j_{max}=1/2$ ) captures the qualitative physics of the crossover, it is insufficient for the continuum limit. They posit that future work extending these methods to higher irreducible representations will be necessary to test whether the ground state of the continuum limit possesses the divergent amount of magic predicted by theoretical arguments based on Type III von Neumann algebras. The study reinforces the intersection of quantum information and quantum field theory, highlighting the role of non-stabilizerness in characterizing the complexity of physical states.

Technical Summary: Non-stabilizerness and Entanglement in (2 + 1)-dimensional SU(2) Lattice Gauge Theory

More like this