Clifford symmetries in quantum many-body systems
Charlie Nation, Rick P. A. Simon, Shreya Banerjee, Francesco Martini, Alessandro Ricottone, Federico Cerisola, Luca Dellantonio · 2026
Read Clifford symmetries in quantum many-body systems by Charlie Nation, Rick P. A. Simon, Shreya Banerjee, Francesco Martini, Alessandro Ricottone, Federico Cerisola, Luca Dellantonio (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-18966.txt.
Introduction
Charlie Nation and six colleagues find symmetries of quantum many-body models with an algorithm built on the classically efficient Clifford group. Symmetries are a critical first step toward solving a model, yet finding them usually depends on human insight. Their method represents Hamiltonians as graphs and extracts symmetries for arbitrary many-body systems, demonstrated on random and physical Hamiltonians with up to one thousand qubits. The approach not only automates a hard search but can deepen understanding of the models it analyzes.
Catalog record
| CREATOR | Charlie Nation, Rick P. A. Simon, Shreya Banerjee, Francesco Martini, Alessandro Ricottone, Federico Cerisola, Luca Dellantonio |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34537 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.18966
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Clifford symmetries in quantum many-body systems Charlie Nation, Rick P. A. Simon, Shreya Banerjee, Francesco Martini, Alessandro Ricottone, Federico Cerisola, Luca Dellantonio Abstract Obtaining the symmetries of a model is a critical step towards developing an understanding and ultimately analytically or numerically solving the model. However, finding symmetries is generally extremely complicated, often being the result of insightful thinking. In this work, we complement human ingenuity with an algorithm. We leverage the classically efficient Clifford group to find symmetries for arbitrary many-body Hamiltonians via a graph representation. We demonstrate our method on random and physical Hamiltonians, with instances of up to one thousand qubits and demonstrate how our approach can provide deeper understanding of the model. Introduction:- Thanks to Emmy Noether 53 , symmetry is one of the most studied concepts in science 28 . Recognising a symmetry can lead to analytical solutions and simplifications of the underlying model. This proves exceptionally useful in determining the ground state energy 23 ; 76 ; 64 , simulating the system’s dynamics 67 , thermodynamics 72 ; 18 , as well as aiding description of more exotic behaviours such as Hilbert space fragmentation and quantum many-body scars 55 ; 47 ; 48 ; 38 . However, finding symmetries can be extremely challenging, and once found they are not always easy to exploit . In practice, symmetries in quantum many-body systems are obtained by writing the Hamiltonian, applying the method of ‘thinking really hard’, and spotting a transformation that leaves the Hamiltonian invariant 77 ; 78 . This method proves suitable when this transformation is simple (or the thinker is a genius 9 ), however many symmetries require highly non-local operations that cannot be easily spotted 56 ; 12 . In recent years, algorithms 10 ; 69 ; 29 ; 14 were developed to tackle this problem. These approaches are classically efficient in the number of qubits and can deterministically find and exploit all Pauli symmetries, i.e., the Pauli operators that commute with the Hamiltonian. Here, ‘exploit’ means that they also determine the reference frame in which the Hamiltonian is block diagonal, and provide effective operators describing each block 29 , as in Fig. 1 . Figure 1: \hat{H} in Fig. 2 (left) and \hat{H}_{\rm sym}=\hat{H}_{1}+\hat{H}_{2} (right) written in the eigenbasis of the symmetry \hat{S}_{\rm min} found by our algorithm (main text). Opacity and colour are norm and phase, see colour bars. However, while Pauli symmetries are generally non-local, their defining property (commutation with all Paulis in the Hamiltonian) makes them also easy to spot. Therefore, while the algorithmic approaches in Refs. 10 ; 69 ; 29 ; 14 can greatly speed up the simulation pipeline 30 ; 57 ; 20 ; 64 , they do not provide fundamentally new insight into physical models. Few works address more complex (non-Pauli) symmetries, for example via quantum computation 39 , or an algebraic approach 47 . Such methods, however, require explicit building of the Hamiltonian in Hilbert space, and/or rely on solving a challenging optimization problem. Therefore, they suffer from prohibitive scaling with system size (at least on classical computers) and are not applicable for models with hundreds or even dozens of qubits. As of today, there is no algorithmic approach that can reliably find non-Pauli symmetries in large systems. This work fills this gap. We provide an algorithm that finds Clifford symmetries (as they can be classically and efficiently represented via the symplectic formalism 3 ). Our approach relies on graph automorphisms (GA) 11 ; 16 ; 36 , which are well studied in graph theory. GA can be solved 44 ; 5 in quasi-polynomial time, and optimised search strategies prove extremely useful in practice, enabling Clifford symmetries to be found for large system sizes ( 1000 qubits in the examples below). Second, our algorithm also exploits the symmetry, see Fig. 1 . This is done by diagonalizing the symmetry itself. As diagonalizing Clifford circuits is generally np-hard 21 , we (classically and efficiently) find the reference frame in which the symmetry is ‘concentrated’ in the minimal number of interacting qubits, such that numerical diagonalization becomes feasible. We first introduce the key background on Pauli and Clifford groups, which we use to efficiently represent the Hamiltonians, symmetries, and all other operations (Fig. 2 ). We then describe how we find Clifford symmetries via a GA (Fig. 2 ), and then decompose said symmetries to a minimal ‘qubit cost’. Next, we explain how we obtain the effective operators describing each block in the decomposed Hamiltonian (back to Fig. 1 ), before showing results on canonical models (Fig. 3 ): random Hamiltonians with injected symmetries and both spin and Fermionic lattice models 19 ; 71 ; 37 ; 60 ; 70 ; 22 ; 8 . Technical details (including generalization to qudits and open systems) and our code can be found in Refs. 51 and 49 , respectively. Background:- The subset of quantum operations that are classically efficient is known as the Clifford group 17 ; 3 ; 33 ; 6 . It is vital for quantum error correction 26 ; 25 ; 27 ; 61 , benchmark and synthesis of quantum circuits/algorithms 58 ; 46 , characterization of ‘quantumness’ (equiv. ‘nonstablizerness’ or ‘magic’) 68 ; 40 ; 7 , and offers a powerful tool for classical pre-computation for both quantum and classical algorithms 59 ; 35 ; 15 . A possible basis of the Clifford group 33 consists of the SUM, Hadamard, Phase and Pauli gates, which can be classically performed on n qubits with polynomial complexity in n 3 ; 24 . The core idea is to represent Pauli strings as 2n -dimensional vectors rather than as 2^{n}\times 2^{n} matrices in Hilbert space, such that Clifford gates can be implemented as a product of 2n\times 2n matrices on these vectors 3 . In this work, we consider an n qubit many-body Hamiltonian written as a sum of Pauli strings, and give a generalization to both mixed dimension qudits and open systems in Ref. 51 . A Pauli string \hat{P}_{i}=\hat{\sigma}^{(i)}_{1}\otimes\hat{\sigma}^{(i)}_{2}\otimes\cdots\otimes\hat{\sigma}^{(i)}_{n} (where \hat{\sigma}^{(i)}_{\mu}\in\{\hat{\mathbb{1}},\hat{X},\hat{Y},\hat{Z}\} , \mu=1,\dots,n , are the canonical Pauli operators) can be represented by a 2n dimensional vector \vec{p}_{i}=[\vec{x}_{i}\,|\,\vec{z}_{i}] , with \vec{x}_{i},\;\vec{z}_{i}\in\mathbb{Z}^{n}_{2} , denoting powers of \hat{X} and \hat{Z} on each site 17 ; 33 selected from the field of integers \bmod{2} . We thus have \hat{P}_{i}=e^{\frac{{\rm i}\pi\eta_{i}}{2}}\bigotimes_{\mu=1}^{n}\hat{X}^{x_{i,\mu}}\hat{Z}^{z_{i,\mu}}\equiv e^{\frac{{\rm i}\pi\eta_{i}}{2}}\hat{P}(\vec{p}_{i}) , with \eta_{i}\in\mathbb{Z}_{4} a discretized phase and \hat{P}(\vec{p}_{i}) the operator corresponding to \vec{p}_{i} . In the following we will refer to both \hat{P}_{i} and \vec{p}_{i} as Paulis (omitting ‘string’) when clear, though notably the latter has no phase information. The input Hamiltonian on n qubits can be expressed as an operator \hat{H} or a tableau \underline{H} [see Fig. 2 (a)] interchangeably, \hat{H}=\sum_{i}^{M}c_{i}e^{\frac{{\rm i}\pi\eta_{i}}{2}}\hat{P}(\vec{p_{i}})\quad\text{and}\quad\underline{H}=\left[\begin{array}[]{c|c|c}c_{1}&\vec{p}_{1}&\eta_{1}\\ \vdots&\vdots&\vdots\\ c_{M}&\vec{p}_{M}&\eta_{M}\end{array}\right]. (1) This sum of M Paulis can have differing coefficients c_{i}\in\mathbb{C} and phases \eta_{i}\in\mathbb{Z}_{4} . In general c_{i} may absorb any phase, however we track discrete phases separately, as \eta_{i}=1 is the minimal finite phase increment on action of a Clifford gate 33 . The tableau representation \underline{H} in Eq. ( 1 ) is thus a M\times(2n+2) matrix 3 ; 33 . We denote the M\times 2n matrix of Pauli powers as \underline{p} , with rows \vec{p}_{i} , and the coefficient