⚛️ high-energy theory

Remarks on non-invertible symmetries on a tensor product Hilbert space in 1+1 dimensions

This paper proposes an index for non-invertible symmetry operators in 1+1 dimensions that generalizes the Gross-Nesme-Vogts-Werner index, demonstrating that such symmetries on lattice tensor product Hilbert spaces are constrained to weakly integral fusion categories and can be systematically described using topological injective matrix product operators (MPOs).

Original authors: Kansei Inamura

Published 2026-02-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kansei Inamura

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 the universe of quantum physics as a giant, intricate Lego set. For decades, physicists have used "symmetries" to understand how these Lego pieces fit together. Think of a symmetry like a rule that says, "If you rotate this whole structure, it looks exactly the same."

Traditionally, these rules were invertible. This means if you apply the rule (rotate), you can always undo it (rotate back). It's like a reversible door: you can walk in, and you can walk out.

But recently, physicists discovered a new, weirder kind of symmetry called non-invertible symmetry. Imagine a rule that says, "If you rotate this structure, it turns into a different structure that you can't simply rotate back to get the original." It's like a magical transformation where you turn a cup into a saucer, but there's no magic spell to turn the saucer back into a cup. You've lost information.

This paper, written by Kansei Inamura, tries to figure out how these "one-way" symmetries can exist in a world built from discrete blocks (like a computer grid or a lattice of atoms).

Here is the breakdown of the paper's main ideas using simple analogies:

1. The Problem: The "Lego Grid" Constraint

In the real world (or in quantum field theory), these weird symmetries are described by complex math called "Fusion Categories." But when we try to build them on a computer or a physical lattice (a grid of Lego bricks), things get tricky.

The grid has a strict rule: The total number of Lego bricks must stay the same. You can't just magically create or destroy bricks.

The Old Rule: If a symmetry is "invertible" (reversible), it fits perfectly on the grid.
The New Problem: If a symmetry is "non-invertible" (one-way), it often requires the grid to change its size or shape in a way that breaks the rules.

The paper asks: Can these weird, one-way symmetries actually exist on a standard Lego grid, or do they break the universe's building codes?

2. The Solution: A "Finger Counting" Index

To solve this, the author invents a new tool called an Index.

Think of the Index like a finger-counting system for these symmetries.

When you apply a symmetry, it might stretch the grid a little bit or shift things around.
The Index measures exactly how much it stretches or shifts.
If the Index is a "clean" number (like 1, 2, or 3), the symmetry fits on the grid.
If the Index is a "messy" number (like the square root of 2), it usually means the symmetry doesn't fit on a standard grid unless we allow some special tricks.

The author generalizes an old counting method (the GNVW index) used for reversible symmetries and adapts it for these new, weird one-way symmetries.

3. The Big Discovery: "Weakly Integral" Symmetries

The paper proves a fascinating constraint. It says that for these non-invertible symmetries to exist on a grid, they must belong to a special club called "Weakly Integral Fusion Categories."

The Analogy:
Imagine you are trying to tile a floor with tiles of different sizes.

Integral Symmetries: You have tiles of size 1, 2, and 3. You can perfectly cover the floor.
Non-Integral Symmetries: You have a tile of size $\sqrt{2}$ (about 1.414). You can't cover a standard floor perfectly with these; you'll always have gaps or overlaps.
The Paper's Finding: The author shows that if you allow the tiles to "mix" with the floor's translation (sliding the whole floor over by one inch), you can make the $\sqrt{2}$ tiles work, BUT only if the total area of all your tiles adds up to a whole number.

In physics terms: You can have these weird symmetries on a grid, but only if the "total size" of the symmetry is a whole number. If it's not, the universe (the grid) rejects it.

4. The Tool: Tensor Networks (The "Magic Blueprint")

To prove this, the author uses a method called Tensor Networks.

Analogy: Imagine a massive, interconnected web of strings and knots. Each knot is a tiny piece of the quantum system.
The author proposes a specific type of knot called a "Topological Injective MPO" (Matrix Product Operator).
Think of this as a specialized blueprint for building these symmetries. It's a way of drawing the symmetry so that it automatically respects the rules of the Lego grid.
The paper shows that if you build your symmetry using these specific blueprints, the "Finger Counting" (Index) always works out correctly, and the symmetry fits the grid.

5. The "Zipper" Conditions

The paper introduces some technical conditions called the "Broken Zipper" and "Two-Sided Zipper."

Analogy: Imagine trying to zip up a jacket where the teeth are slightly bent.
The "Zipper Condition" checks if the symmetry operations can "zip" together smoothly without getting stuck.
The author shows that for the symmetries they studied (like the famous Kramers-Wannier duality, which swaps hot and cold in magnets), these zippers work perfectly. This proves that these specific symmetries are valid and can exist on a grid.

Summary: What does this mean for us?

This paper is a "rulebook update" for quantum physicists.

It confirms a suspicion: We can't just put any weird symmetry on a computer grid. There are strict mathematical limits.
It provides a test: If you want to build a quantum computer or a new material with these symmetries, you can use the author's "Index" to check if it's possible.
It bridges the gap: It connects the abstract, beautiful math of "Fusion Categories" with the gritty reality of building things out of atoms and qubits.

In a nutshell: The paper says, "Weird, one-way symmetries are real, but they are picky. They will only live on a grid if their 'total size' is a whole number. If they try to live there otherwise, the grid will break."

1. Problem Statement

The paper addresses a fundamental question in the study of generalized symmetries in quantum many-body systems: Which fusion category symmetries can be realized on a lattice with a tensor product Hilbert space?

Context: In relativistic Quantum Field Theory (QFT), non-invertible symmetries are described by topological defects forming fusion categories. However, on a lattice, the Hilbert space is a tensor product of finite-dimensional local spaces ( $H = \bigotimes H_i$ ). This structure imposes strict constraints.
The Conflict: It was recently proven that a fusion category $\mathcal{C}$ can be realized exactly on a tensor product Hilbert space only if $\mathcal{C}$ is integral (all quantum dimensions are integers). However, many physically relevant non-integral categories (e.g., the Ising fusion category) appear to be realizable if one allows the symmetry operators to mix with Quantum Cellular Automata (QCAs), such as lattice translations.
The Conjecture: A recent conjecture posits that even with mixing with QCAs, only weakly integral fusion categories (where the total dimension is an integer) can be realized on a tensor product Hilbert space.
The Gap: Proving this conjecture requires a precise definition of non-invertible symmetry operators on a lattice and an understanding of their "index" (a topological invariant generalizing the GNVW index for QCAs). Specifically, it is unknown whether the index is homogeneous (i.e., whether all fusion channels of two symmetry operators share the same index).

2. Methodology

The author employs two complementary approaches to tackle this problem:

A. Axiomatic Approach (Physical Assumptions)

The paper first assumes the existence of defect Hilbert spaces ( $H^L_X, H^R_X$ ) associated with symmetry operators $D_X$ . These spaces are obtained by locally modifying the original Hilbert space around a defect. The author postulates four physical assumptions:

Existence: Defect Hilbert spaces exist and are independent of defect position (up to isomorphism).
Product: The product of symmetry operators corresponds to the fusion of defects.
Sum: The sum of operators corresponds to the direct sum of defect Hilbert spaces.
Unitarity: The Hermitian conjugate corresponds to the defect with opposite orientation.

Based on these, the author defines Left and Right Dimensions ($ldim, rdim$) as the ratios of the defect Hilbert space dimension to the total Hilbert space dimension. From these, an Index is defined:
$\text{ind}(D_X) = \sqrt{\frac{ldim(D_X)}{rdim(D_X)}}$
and a Lattice Quantum Dimension:
$qdim_{lat}(D_X) = \sqrt{ldim(D_X) \cdot rdim(D_X)}$

B. Tensor Network Approach

To provide a concrete formulation without relying solely on abstract assumptions, the author introduces a specific class of tensor network operators: Topological Injective Matrix Product Operators (MPOs).

Definition: An injective MPO tensor $O$ is "topological" if it satisfies specific algebraic relations involving the left and right fixed points ( $\Lambda_l, \Lambda_r$ ) of its transfer matrix. These relations ensure the operator behaves like a topological defect.
Construction: The author constructs defect Hilbert spaces and "movement operators" (unitary operators that shift the defect) directly from the MPO tensors.
Sufficient Conditions: The paper introduces Fusion and Splitting Tensors for the fusion channels of MPOs. It proposes two algebraic conditions:
1. Broken Zipper Condition: A weaker version of the standard zipper condition, allowing the fusion tensor to commute with the MPO up to a shift.
2. Two-Sided Zipper Condition: A condition relating the fusion/splitting tensors to the fixed points of the transfer matrices.

3. Key Contributions and Results

A. Generalization of the GNVW Index

The paper successfully generalizes the Gross-Nesme-Vogts-Werner (GNVW) index (originally for invertible QCAs) to non-invertible symmetry operators.

For invertible symmetries (QCAs), $qdim_{lat} = 1$ and the index equals the GNVW index.
For non-invertible symmetries, the index can be non-trivial (e.g., $\sqrt{2}$ for the Kramers-Wannier duality).

B. The Integrality Constraint (Axiomatic)

Using the axiomatic approach, the author proves:

Exact Realization: If a fusion category is realized exactly (without QCA mixing), it must be integral. This recovers the mathematical result of Evans and Jones [26].
QCA Mixing: If the fusion rules are realized up to QCAs (mixing allowed), and if the index is homogeneous (all fusion channels of $D_X D_Y$ $D_{X} D_{Y}$ have the same index), then the fusion category must be weakly integral.
- Logic: Homogeneity implies $qdim_{lat}$ is a one-dimensional representation of the fusion rules. Since $qdim_{lat}$ is rational (derived from finite dimensions), the quantum dimensions must be square roots of integers, making the total dimension an integer.

C. Tensor Network Realization and Sufficient Conditions

The paper provides a concrete framework to test the homogeneity conjecture:

Topological Injective MPOs: These operators naturally include invertible symmetries, non-anomalous fusion category symmetries, and Kramers-Wannier dualities.
Homogeneity Proof: The author proves that if the Broken Zipper and Two-Sided Zipper conditions are satisfied by the fusion/splitting tensors of the MPOs, then the index is homogeneous.
Verification: The paper explicitly constructs these tensors and verifies the conditions for:
- Invertible symmetries (MPUs).
- Non-anomalous fusion categories (Representations of Hopf algebras).
- Kramers-Wannier duality for $\mathbb{Z}_2$ and $\mathbb{Z}_N$ gauging.
- In all these examples, the index is homogeneous, and the resulting categories are weakly integral.

D. Defect Hilbert Spaces and Sequential Circuits

The author constructs explicit defect Hilbert spaces for topological injective MPOs, showing they match the axiomatic definition.
It is shown that topological injective MPOs admit sequential quantum circuit representations, where local unitary gates act as "movement operators" for the defects. This connects the abstract index theory to recent proposals on sequential circuits as generalized symmetries.

4. Significance and Outlook

Support for the Conjecture: The paper provides strong evidence for the conjecture that only weakly integral fusion categories can be realized on tensor product Hilbert spaces (even with QCA mixing). While the general proof of index homogeneity remains open, the fact that it holds for all known examples (including non-invertible dualities) suggests it is a universal property.
Bridge between Fields: It bridges the gap between abstract fusion category theory, operator algebra (QCAs), and tensor network methods (MPOs).
New Tools: The introduction of "Topological Injective MPOs" and the "Broken/Two-Sided Zipper conditions" provides a rigorous algebraic toolkit for analyzing non-invertible symmetries on lattices without relying on continuum limits.
Open Question: The paper poses a critical question: Do all topological injective MPOs satisfy the zipper conditions? If yes, the conjecture is proven. If no, there may exist exotic non-weakly-integral symmetries realizable on lattices, which would have profound implications for the classification of gapped phases.

In summary, Inamura establishes a robust index theory for non-invertible symmetries on lattices, demonstrating that the interplay between lattice constraints and fusion rules strongly restricts realizable symmetries to the class of weakly integral fusion categories, provided the index is homogeneous—a condition verified for all major known examples.