Abstract
We develop a theoretical framework for $S_n$-equivariant convolutional quantum circuits with SU$(d)$-symmetry, building on and significantly generalizing Jordan's Permutational Quantum Computing (PQC) formalism based on Schur-Weyl duality connecting both SU$(d)$ and $S_n$ actions on qudits. In particular, we utilize the Okounkov-Vershik approach to prove Harrow's statement (Ph.D. Thesis 2005 p.160) on the equivalence between $\operatorname{SU}(d)$ and $S_n$ irrep bases and to establish the $S_n$-equivariant Convolutional Quantum Alternating Ans"atze ($S_n$-CQA) using Young-Jucys-Murphy (YJM) elements. We prove that $S_n$-CQA is able to generate any unitary in any given $S_n$ irrep sector, which may serve as a universal model for a wide array of quantum machine learning problems with the presence of SU($d$) symmetry. Our method provides another way to prove the universality of Quantum Approximate Optimization Algorithm (QAOA) and verifies that 4-local SU($d$) symmetric unitaries are sufficient to build generic SU($d$) symmetric quantum circuits up to relative phase factors. We present numerical simulations to showcase the effectiveness of the ans"atze to find the ground state energy of the $J_1$--$J_2$ antiferromagnetic Heisenberg model on the rectangular and Kagome lattices. Our work provides the first application of the celebrated Okounkov-Vershik's $S_n$ representation theory to quantum physics and machine learning, from which to propose quantum variational ans"atze that strongly suggests to be classically intractable tailored towards a specific optimization problem.
Introduction
In quantum machine learning, the challenge of learning quantum states in complex systems, such as frustrated magnets, is often hindered by the vastness of the Hilbert space and the intricate sign structure of wavefunctions. Classical models, including Convolutional Neural Networks (CNNs), are powerful due to their equivariance, but translating this to quantum spin systems is non-trivial. Many quantum systems possess global symmetries, such as rotation ($SU(d)$) or permutation ($S_n$), which standard Variational Quantum Eigensolvers (VQEs) often fail to respect, leading to "barren plateaus" or convergence into incorrect energy sectors. This research introduces a rigorous mathematical framework for building quantum circuits that are inherently symmetry-aware, leveraging representation theory to achieve a super-exponential advantage over classical methods.
Theoretical Framework: $S_n$ Representation Theory and YJM Elements
The core of the proposed approach lies in Schur-Weyl Duality, a mathematical bridge connecting the permutation group $S_n$ (acting on particle labels) with the unitary group $SU(d)$ (acting on internal states). By utilizing the Okounkov-Vershik approach to $S_n$ representation theory, the authors develop the $S_n$-equivariant Convolutional Quantum Alternating Ansatz ($S_n$-CQA).
A key component of this framework is the use of Young-Jucys-Murphy (YJM) elements. These operators are strictly diagonal in the Schur basis, allowing for efficient navigation within specific symmetry-protected sectors. The structure of these sectors is governed by the branching rules of the symmetric group, which can be visualized via a Bratteli diagram.
The $S_n$-CQA ansatz alternates between a problem Hamiltonian and a "mixer" Hamiltonian constructed from YJM elements. This construction ensures that the quantum state remains within the desired symmetry sector throughout the optimization process. The basis for these sectors is defined by Standard Young Tableaux (SYT), which correspond to unique paths through the Bratteli diagram.
Furthermore, the authors rigorously prove that while 2-local symmetric gates are insufficient for higher-dimensional systems ($d \ge 3$), 4-local $SU(d)$-symmetric unitaries are sufficient to construct generic symmetric circuits.
Numerical Simulations and Results
The $S_n$-CQA ansatz was evaluated using numerical simulations of the $J_1-J_2$ antiferromagnetic Heisenberg model on $3 \times 4$ rectangular and 12-spin Kagome lattices. This model is a standard "hard problem" in condensed matter physics due to its frustrated nature. Compared to classical benchmarks like Restricted Boltzmann Machines (RBM) and Group equivariant CNNs (GCNN), the $S_n$-CQA demonstrated:
- High Precision: Achieving energy errors as low as $10^{-4}$ to $10^{-5}$ relative to exact diagonalization in frustrated regions where classical models struggle.
- Robust Convergence: Unlike RBMs and GCNNs, which showed oscillating or stagnant behavior, the $S_n$-CQA converged smoothly to the ground state.
- Efficiency: The model requires only $O(pn^2)$ parameters for $p$ layers, making it highly efficient for Near-term Intermediate-Scale Quantum (NISQ) devices.
Key Insights and Future Outlook
By forcing the quantum circuit to stay within the correct symmetry sector (e.g., total spin $J=0$), the search space is reduced polynomially. This approach bypasses the "sign problem" that typically causes classical neural networks to fail in frustrated magnets. Moreover, while calculating Fourier coefficients over $S_n$ is a factorial-time problem classically ($O(n!)$), the $S_n$-CQA can simulate these operations in polynomial time, suggesting a massive theoretical advantage.
Future work will focus on addressing the "phase factor problem" between different symmetry sectors and mitigating noise-induced barren plateaus on real quantum hardware. The framework's potential for Quantum Adiabatic Optimization also remains an exciting avenue for further research.
Summary updated