⚛️ quantum physics

Regular language quantum states

This paper introduces "regular language states," a family of quantum many-body states constructed from regular formal languages that encompass physically relevant states like GHZ, W, and Dicke states, and establishes a theoretical framework using matrix product states and tensor networks to efficiently characterize, recognize, and determine the equivalence and shift-invariance of these states.

Original authors: Marta Florido-Llinàs, Álvaro M. Alhambra, David Pérez-García, J. Ignacio Cirac

Published 2026-03-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Marta Florido-Llinàs, Álvaro M. Alhambra, David Pérez-García, J. Ignacio Cirac

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 you are trying to describe a massive, complex library of books. Some books are simple stories, others are intricate novels, and some are just lists of words. In the world of quantum physics, scientists deal with "quantum states," which are like the descriptions of these books, but instead of words, they are made of qubits (quantum bits) that can be in many states at once.

This paper introduces a new way to organize and understand a specific, very important family of these quantum states. The authors call them Regular Language States (RLS).

Here is the breakdown of what they did, using simple analogies:

1. The Library of Rules (Regular Languages)

In computer science, there is a concept called a "Regular Language." Think of this as a set of rules for writing sentences.

The Analogy: Imagine a rule that says, "You can write any number of zeros, followed by exactly one '1', followed by any number of zeros." (e.g., 000100, 1, 0010000).
This rule is simple. A computer can check if a sentence follows this rule very quickly because it doesn't need to remember a huge amount of history; it just needs to follow a simple flowchart.
The authors realized that many famous quantum states (like the GHZ state, W state, or Dicke state) are actually just the quantum version of these simple rule-based sentences. They are superpositions (quantum mixtures) of all the possible sentences that follow a specific simple rule.

2. The Flowchart (Finite Automata)

How do we describe these rules? In computer science, we use a Finite Automaton.

The Analogy: Think of a flowchart or a maze with a few rooms. You start at the entrance. Every time you see a "0," you move to a specific room. Every time you see a "1," you move to another. If you end up in a "Green Room" (an accepting state) after reading the whole sentence, the sentence is valid.
The paper shows that every one of these "Regular Language States" can be built directly from such a flowchart.
The Magic: Because the flowchart is small and simple, the resulting quantum state is also "compact." Even if the quantum system has millions of particles, you can describe it using a small, fixed-size set of instructions (called a Matrix Product State or MPS). This is like having a tiny recipe that can bake a cake of any size.

3. The "Unambiguous" Rulebook

The authors found a special trick. They realized that to make the quantum math work perfectly, the flowchart must be unambiguous.

The Analogy: Imagine a maze where, for a specific path of instructions, there is only one way to get from the start to the finish. If there were two different ways to get there for the same sentence, the math would get messy (the coefficients wouldn't be clean).
They developed a new mathematical test (using tools from both computer science and quantum physics) to check if a flowchart is "unambiguous" without having to build the whole thing. This helps physicists quickly know if a set of rules creates a valid, clean quantum state.

4. The "Canonical Form" (The Unique ID Card)

One of the biggest headaches in quantum physics is: "Are these two quantum states actually the same, just dressed differently?"

The Analogy: Imagine you have two descriptions of a city. One says "Go North, then East." The other says "Go East, then North." They might describe the same destination, but the instructions look different.
The authors created a "Canonical Form." Think of this as a unique ID card or a standardized passport for these quantum states.
They proved that if you convert any "Regular Language State" into this specific ID format, you can instantly tell if two states are equivalent.
The Twist: They also discovered something surprising. In standard quantum physics, if you change the "basis" (the way you look at the state) using local operations, the "size" of the description (bond dimension) usually stays the same. But for these Regular Language States, you can change the state in a way that makes the description smaller or larger, even though the physics hasn't fundamentally changed. It's like rewriting a novel in a different language that happens to be more concise.

5. Why Does This Matter?

Efficiency: It gives scientists a new, faster way to design and simulate quantum computers. Instead of guessing how to build a complex state, they can just write down a simple "regular expression" (like a search engine query) and let the math generate the quantum state.
New Tools: It bridges the gap between Computer Science (how we process text) and Quantum Physics (how nature works). It suggests that the tools we use to write compilers and search engines can actually help us solve problems in quantum gravity and condensed matter physics.
Future Circuits: The authors suggest that because these states have a "flowchart" structure, we can build "compilers" specifically for them. This could make preparing these states on a real quantum computer much faster and less error-prone.

Summary

The paper says: "Let's treat quantum states like sentences in a simple language. If we use the rules of computer science (flowcharts and regular expressions) to build them, we get a powerful, efficient, and easy-to-analyze family of quantum states that includes many of the most important ones we already know."

It's a new lens that turns complex quantum mysteries into manageable, rule-based puzzles.

1. Problem Statement

The identification of physically relevant families of many-body quantum states that admit a concise theoretical description is crucial for advancing quantum science. While families like stabilizer states, Gaussian states, and standard Tensor Network States (specifically Matrix Product States or MPS) have been extensively studied, there is a gap in the theoretical framework for a specific class of states that naturally arise in quantum information and condensed matter physics: those defined by Regular Languages (RL).

Standard MPS theory relies on tools like the canonical form and the fundamental theorem (which relates Local Unitary (LU) equivalence to bond-level transformations). However, these tools often fail for states derived from non-deterministic finite automata or those with open boundary conditions (OBC) that do not fit the translationally invariant (TI) periodic boundary condition (PBC) assumptions of standard MPS. The authors aim to bridge the gap between formal language theory (computer science) and quantum many-body physics to create a unified framework for these states.

2. Methodology

The authors introduce Regular Language States (RLS), defined as the uniform superposition of all words of length $N$ belonging to a specific regular language $L$ . They leverage the duality between formal languages and finite automata to construct a quantum mechanical description.

Definition of RLS: For a regular language $L$ over an alphabet $\Sigma$ , the state $|L_N\rangle$ is $\sum_{w \in L \cap \Sigma^N} |w\rangle$ .
Automata-to-MPS Mapping: The authors utilize Unambiguous Finite Automata (UFA)—automata where every accepted word has exactly one accepting path. They prove that any RLS generated by a UFA can be exactly represented as an MPS with binary entries and a constant bond dimension $D$ equal to the number of states in the automaton.
Algebraic Characterization: Instead of relying solely on automata construction, they develop algebraic criteria using the MPS algebra (the algebra generated by the MPS tensors) to determine:
- Whether a given MPS corresponds to an RLS (checking for unambiguity).
- Whether an RLS is shift-invariant (translationally invariant).
Canonical Decomposition: To address the lack of a standard canonical form for RLS (due to open boundaries and non-uniformity), they introduce a canonical decomposition for regular languages. This involves separating symbols into those that appear infinitely often ( $\Sigma_\infty$ ) and those with bounded occurrences ( $\Sigma_f$ ).
LU-Equivalence Analysis: Using this canonical form, they derive a fundamental theorem for the LU-equivalence of sparse RLS (languages where the number of words scales polynomially with $N$ ).

3. Key Contributions

A. Formal Definition and MPS Representation

The paper establishes that RLS are a subset of MPS with constant bond dimension.

Theorem 1: Any RLS admits an MPS description with binary entries and bond dimension $D = |Q|$ (number of automaton states). The tensors are derived directly from the automaton's transition function.
Efficiency: This provides an efficient criterion to recognize RLS. The authors provide an $O(dD^3)$ algorithm (where $d$ is alphabet size) to check if a binary MPS corresponds to an RLS by verifying conditions on the tensor algebra, avoiding the expensive construction of product automata.

B. Shift-Invariance and Tensor Networks

The authors address the problem of determining if an RLS is translationally invariant (TI).

Corollary 4: They prove that a RL is shift-invariant if and only if a specific commutator condition holds within the MPS algebra: $\langle v_l | [a, b] | v_r \rangle = 0$ for all $a, b$ in the algebra generated by the MPS tensors.
Significance: This allows checking shift-invariance in $O(dD^3)$ time, a significant improvement over naive automata-based methods which scale exponentially ( $O(dD^2 4^D)$ ).

C. Canonical Form and Fundamental Theorem

The paper introduces a new canonical form tailored to RLS, overcoming the limitations of standard MPS canonical forms (which assume TI and PBC).

Canonical Decomposition (Theorem 5): Any RL can be uniquely decomposed into pairs of languages $(L^m_j, X^m_j)$ , separating the "infinite" structure from the "finite" bounded symbols.
Fundamental Theorem of Sparse RLS (Theorem 7): For sparse RLS, two states are LU-equivalent ( $|L_2\rangle = U^{\otimes N} |L_1\rangle$ $∣ L_{2} ⟩ = U^{\otimes N} ∣ L_{1} ⟩$ ) if and only if:
1. The infinite parts of their canonical decompositions are related by a permutation (relabeling) of the alphabet.
2. The finite parts are related by a local unitary transformation acting on the bounded symbols.
Key Insight: Unlike standard TI MPS where LU equivalence preserves the bond dimension, LU equivalence for RLS can change the minimal bond dimension (i.e., the size of the minimal DFA). This reveals that bond dimension is not a robust invariant for RLS under local unitaries.

4. Results and Examples

Physical Relevance: The framework naturally encompasses famous quantum states:
- GHZ states: Correspond to $0^* \cup 1^*$ .
- W-states: Correspond to $0^*10^*$ (exactly one '1').
- Dicke states: Correspond to languages with a fixed number of excitations (e.g., $0^*10^*10^*$ ).
- Domain Wall Excitations: States representing excited states in the transverse-field Ising model.
LU-Equivalence Counter-Example: The authors demonstrate that two RLS can be LU-equivalent yet possess minimal DFAs of different sizes (e.g., bond dimensions 5 vs. 4). This highlights that the "complexity" of an RLS is not solely captured by the bond dimension of a TI representation.
Algorithmic Efficiency: The paper provides polynomial-time algorithms for:
- Checking if an MPS is an RLS.
- Checking if an RLS is shift-invariant.
- Determining LU-equivalence for sparse RLS.

5. Significance and Outlook

Cross-Disciplinary Unification: The work successfully merges concepts from formal language theory (automata, regular expressions) with quantum many-body physics (MPS, entanglement, LU equivalence).
New Tools for Quantum Physics: It provides a rigorous framework to classify and analyze a broad class of quantum states that were previously difficult to handle with standard MPS tools due to their open-boundary nature and non-uniform structure.
Applications:
- State Preparation: The automaton structure of RLS can be exploited to design more efficient quantum circuits for state preparation compared to general-purpose compilers.
- Phase Classification: The fundamental theorem offers a pathway to classify phases of matter defined by regular languages.
- Machine Learning: The connection to Weighted Finite Automata (WFA) suggests potential applications in improving the interpretability of black-box machine learning models.
Future Directions: The authors suggest extending this framework to 2D (using 2D Online Tessellation Automata and PEPS) and to broader language classes like Context-Free Languages, though they note that decidability issues arise in 2D. They also hint at generalizing these results to "Algebraic RLS" for uniform MPS with arbitrary boundary matrices.

In summary, this paper establishes Regular Language States as a distinct and physically relevant family of quantum states, providing the necessary mathematical machinery (canonical forms, equivalence theorems, and efficient algorithms) to study them rigorously within the tensor network framework.