betaFeedback

WorksPhysics and Astronomy

Simulating quantum electrodynamics in 2+1 dimensions with qubits and qumodes

Abstract

We develop a hybrid qubit-qumode framework for simulating quantum electrodynamics in 2+1 dimensions. In this approach, fermionic matter fields are represented by qubits, while U(1) gauge fields are encoded in continuous-variable bosonic modes whose canonical quadratures capture the electric and vector-potential components of the theory. To reconcile the non-compact phase space of the qumodes with the compact U(1) gauge symmetry, we introduce and compare two complementary constraint-enforcement strategies: (i) a squeezing-based projection that confines qumode states to the unit circle through an effective modification of the inner product, and (ii) a method that dynamically enforces compactness via a penalty Hamiltonian term. We construct the corresponding hybrid Hamiltonian, derive its decomposition into experimentally accessible qubit-qumode gates, and analyze its spectrum in the analytically tractable single-plaquette limit. The hybrid formulation reproduces the correct gauge-invariant dynamics and provides a scalable route toward simulating Abelian lattice gauge theories coupled to fermionic matter on near-term hybrid quantum architectures. Ground-state preparation and convergence are demonstrated using a continuous-variable extension of the Quantum Imaginary Time Evolution (QITE) algorithm, establishing a general framework for hybrid discrete-continuous quantum simulations of lattice gauge theories.

Introduction

The real-time dynamics of lattice gauge theories represent one of the most compelling frontiers in quantum simulation, offering access to strongly coupled regimes that remain intractable for classical computation [1, 2, 3, 4]. The foundational work of Jordan, Lee, and Preskill demonstrated that scattering processes in scalar field theories can be simulated efficiently with quantum computers, providing a concrete pathway for exploring quantum field dynamics using controllable quantum systems [5]. Building on the Hamiltonian formulation introduced by Kogut and Susskind [6], a variety of quantum algorithms have been proposed for studying Abelian and non-Abelian gauge theories, confinement phenomena, and particle production. Extending these methods to real-time dynamics of quantum electrodynamics (QED) and quantum chromodynamics (QCD) remains a key goal for both high-energy, nuclear, and condensed-matter physics.

Simulating such theories on quantum hardware requires representing both fermionic and gauge-boson degrees of freedom—objects of fundamentally different character. Fermionic fields are naturally discrete and can be mapped efficiently to qubits via transformations such as Jordan–Wigner or Bravyi–Kitaev [7, 8]. Gauge fields, by contrast, are continuous variables living on group manifolds, e.g., U(1)U(1), SU(2)SU(2), typically represented in qubit-based simulations by truncating their Hilbert spaces to finite-dimensional subspaces [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]. While systematic, such truncations often introduce significant resource overhead and discretization errors that grow near the continuum limit. Continuous-variable (CV) quantum systems, whose observables are described by canonical quadratures of bosonic modes [21], offer a natural alternative for encoding gauge fields without discretization [22].

A broad spectrum of approaches to quantum simulation of lattice gauge theories has emerged over the past decade. Early digital implementations using trapped ions and superconducting qubits have demonstrated real-time dynamics and string breaking in 1+11+1-dimensional gauge models [11, 13], establishing the viability of Hamiltonian-based digital quantum simulation. Analog and hybrid schemes have been explored in cold-atom systems [23] and in trapped-ion and circuit-QED architectures capable of realizing gauge-invariant couplings [24]. Continuous-variable (CV) quantum information has recently gained attention as a natural framework for representing bosonic fields without truncation: studies of the O(3)O(3) model [25], scalar field theory [26, 27, 28, 29, 30], SU(2)SU(2) lattice gauge theory [31], quantum electrodynamics and related field theories [32, 33, 34], Z2Z_2-Higgs [35, 36, 37, 38, 39] and Yukawa [40]. These developments motivate the present work, which integrates discrete qubits for fermionic matter with CV qumodes for gauge fields into a unified hybrid architecture, bridging discrete and continuous paradigms.

Recent advances in experimental architectures now enable hybrid quantum platforms that coherently couple discrete and continuous degrees of freedom, including trapped ions, superconducting circuits, and cavity-QED systems [41, 42, 43, 44]. In trapped-ion devices, qubits are encoded in the internal electronic states of ions, while motional modes act as bosonic qumodes [45, 46]; in superconducting circuits, nonlinear Josephson junctions serve as qubits coupled to harmonic resonators that realize qumodes [47]. Such hybrid systems offer access to universal sets of gates combining qubit and qumode operations, including Gaussian gates, non-Gaussian phase gates, and conditional displacements [42]. These developments open a promising avenue for hybrid discrete-continuous quantum simulation of field theories, where the discrete and continuous sectors can be encoded natively in the hardware.

In this work, we introduce a hybrid qubit–qumode approach to simulating quantum electrodynamics in 2+12+1 dimensions, a paradigmatic Abelian lattice gauge theory that captures key features of gauge-field dynamics in higher dimensions while remaining computationally tractable. In our formulation, fermionic matter fields are represented by qubits, and the U(1)\mathrm{U}(1) gauge degrees of freedom are encoded in pairs of qumodes. This construction enables a continuous-variable representation of the gauge fields that preserves gauge invariance without explicit truncation. To reconcile the non-compact nature of qumodes with the compact U(1)\mathrm{U}(1) gauge symmetry, we introduce two complementary constraint-enforcement schemes: (i) a squeezing-based projection that restricts qumode states to the unit circle through an effective modification of the inner product, and (ii) a method that dynamically enforces compactness by adding corresponding penalty terms to the Hamiltonian.

We demonstrate both approaches in the single-plaquette model, where analytical and numerical results confirm convergence to the correct U(1)\mathrm{U}(1) spectrum. The hybrid Hamiltonian is further decomposed into elementary qubit–qumode gates compatible with experimental realizations, enabling direct implementation on near-term hybrid platforms. We also employ a continuous-variable generalization of the Quantum Imaginary Time Evolution (QITE) algorithm for ground-state preparation [48] [49], establishing a scalable route toward hybrid simulation of interacting gauge–fermion systems.

The framework developed here extends naturally to larger lattices and to non-Abelian gauge theories, where the combination of discrete and continuous resources may provide an efficient and physically transparent encoding. Beyond applications in high-energy and nuclear physics, similar techniques can be adapted to model fermion–boson interactions in quantum chemistry, condensed-matter, and polaritonic systems [50, 51].

The hybrid qubit–qumode representation developed here is complementary to previous digital or continuous-variable simulations, as it preserves the full infinite-dimensional Hilbert space of the gauge field without resorting to truncation and embeds the compact U(1)\mathrm{U}(1) structure directly at the operator level. This allows gauge invariance to be maintained exactly while retaining the experimental accessibility of hybrid architectures, providing a distinct alternative to qubit-only encodings of lattice gauge theories.

The remainder of this paper is organized as follows. § 2 introduces the notation and hybrid gate set. § 3 formulates the lattice Hamiltonian for U(1)\mathrm{U}(1) gauge theory in 2+12+1 dimensions, and § 4 describes its quantum implementation with qubits and qumodes. § 5 presents numerical results for a single plaquette. We conclude in § 6 with an outlook on scalability and applications of hybrid quantum simulations.

Notation and qubit-qumode gates

Hybrid quantum simulation combines discrete-variable (DV) qubits with continuous-variable (CV) bosonic modes (“qumodes” or harmonic oscillators) to form a unified computational basis. Throughout this work, we denote by q^\hat{q} and p^\hat{p} the canonical quadrature operators of a qumode satisfying

[q^,p^]=i(1)[\hat{q},\hat{p}] = i \tag*{(1)}

with units such that ℏ=1\hbar= 1. Each qumode thus realizes an infinite-dimensional Hilbert space HCV\mathcal{H}_{\mathrm{CV}}, spanned by the eigenstates ∣q⟩|q\rangle or ∣p⟩|p\rangle of the corresponding quadrature operators.

Qubits are represented by two-level systems with computational basis states ∣0⟩|0\rangle and ∣1⟩|1\rangle. We denote the Pauli operators by σ∈{X,Y,Z}\sigma\in\{X,Y,Z\}, and the ladder combinations

X±=12(X±iY).(2)X^{\pm} = \frac{1}{2}(X \pm iY). \tag*{(2)}

Multi-qubit operators are expressed as tensor products of Pauli strings, P=σi1⊗σi2⊗⋯P = \sigma_{i_1} \otimes\sigma_{i_2} \otimes\cdots, which form a complete operator basis for the discrete sector HDV\mathcal{H}_{\mathrm{DV}}.

The combined system of nqn_q qubits and ncn_c qumodes resides in

Hhybrid=HDV⊗HCV.(3)\mathcal{H}_{\mathrm{hybrid}} = \mathcal{H}_{\mathrm{DV}} \otimes\mathcal{H}_{\mathrm{CV}}. \tag*{(3)}

Operators acting on the hybrid space may couple qubits and qumodes. A general hybrid operator takes the form

O^=P⊗f(q^,p^),(4)\hat{O} = P \otimes f(\hat{q},\hat{p}), \tag*{(4)}

where PP is a tensor product of Pauli matrices acting on qubits and f(q^,p^)f(\hat{q},\hat{p}) is a polynomial or analytic function of the quadratures. The elementary commutation relations between these operators govern the construction of gate sets for hybrid quantum simulation.

Following [21], universal CV computation is generated by Gaussian gates (displacements, rotations, squeezing) together with a single non-Gaussian operation such as the cubic-phase gate,

V(γ)=eiγ3q^3,(5)V(\gamma) = e^{i\frac{\gamma}{3}\hat{q}^{3}}, \tag*{(5)}

as well as the two-qumode beam splitter gate

BS(z)=e−za^†b^+z∗a^b^†,(6)\mathrm{BS}(z) = e^{-z\hat{a}^{\dagger}\hat{b}+z^{*}\hat{a}\hat{b}^{\dagger}}, \tag*{(6)}

where a^,b^\hat{a},\hat{b} refer to the bosonic lowering operators of two different qumodes. When combined with universal qubit gates {Rx,Ry,Rz,CNOT}\{R_x,R_y,R_z,\mathrm{CNOT}\}, hybrid universality can be achieved by introducing at least one entangling qubit–qumode operation. A convenient choice, experimentally realizable in trapped-ion and superconducting-circuit platforms, is the conditional displacement gate

CD(α)=e(αa^†−α∗a^)Z,(7)\mathrm{CD}(\alpha) = e^{(\alpha\hat{a}^{\dagger}-\alpha^{*}\hat{a})Z}, \tag*{(7)}

where a^=12(q^+ip^)\hat{a} = \frac{1}{\sqrt{2}}(\hat{q}+i\hat{p}) is the bosonic annihilation operator and the Pauli matrix ZZ acts on a control qubit.

Together, the set

Ghybrid={Rx,y,z,CNOT,D(ξ),S(z),V(γ),BS(z),CD(α)},(8)\mathcal{G}_{\mathrm{hybrid}}=\{R_{x,y,z},\mathrm{CNOT},D(\xi),S(z),V(\gamma),BS(z),CD(\alpha)\}, \tag*{(8)}

is universal for hybrid computations [42]. Note that this is not a minimal universal gate set since, for example, the cubic phase gate can be constructed from conditional displacements and qubit gates. In addition, different choices of universal gate sets are possible. For example, the conditional displacement gate CD(α)CD(\alpha) in Eq. (8) can be replaced with the red sideband/Jaynes-Cummings gate listed in Table 1 to achieve a universal hybrid gate set. See in particular Ref. [42] for more details. All elements of the Hamiltonians studied in this work can be expressed as combinations of these gates or their infinitesimal generators. In our simulation of 2+1-dimensional quantum electrodynamics, fermionic matter fields are encoded in qubits via a Jordan–Wigner transformation, while gauge fields residing on lattice links are represented by qumodes. Our simulation protocol employs several additional gates. While these additional gates can be decomposed into the elementary operations listed in Eq. (8), we list all relevant gates used in this work in Table 1.

Gate typeOperationShortOperator
QubitPauli operatorsσi\sigma^{i}
RotationRi(θ)R_i(\theta)eiθσi/2e^{i\theta\sigma_i/2}
Controlled NOTCNOTeiπ4(I1−Z1)(I2−X2)e^{i\frac{\pi}{4}(I_1-Z_1)(I_2-X_2)}
QumodeRotationR(θ)R(\theta)eiθa^†a^e^{i\theta\hat{a}^{\dagger}\hat{a}}
FourierFeiπ2a^†a^e^{i\frac{\pi}{2}\hat{a}^{\dagger}\hat{a}}
DisplacementD(z)D(z)eza^†−z∗a^e^{z\hat{a}^{\dagger}-z^{*}\hat{a}}
Single-mode squeezingS(z)S(z)e(z∗a^a^−za^†a^†)/2e^{(z^{*}\hat{a}\hat{a}-z\hat{a}^{\dagger}\hat{a}^{\dagger})/2}
Beam splitterBS(z)BS(z)eza^b^†−z∗a^†b^e^{z\hat{a}\hat{b}^{\dagger}-z^{*}\hat{a}^{\dagger}\hat{b}}
KerrK(z)K(z)eiz(a^†a^)2e^{iz(\hat{a}^{\dagger}\hat{a})^{2}}
Cross-KerrCK(z)CK(z)eiza^†a^b^†b^e^{iz\hat{a}^{\dagger}\hat{a}\hat{b}^{\dagger}\hat{b}}
Quadratic phaseP(θ)P(\theta)e−iθ2q^2e^{-i\frac{\theta}{2}\hat{q}^{2}}
Cubic phaseV(θ)V(\theta)eiθ3q^3e^{i\frac{\theta}{3}\hat{q}^{3}}
HybridRed sideband/Jaynes CummingsRSB(z)RSB(z)eiza^X++iz∗a^†X−e^{iz\hat{a}X^{+}+iz^{*}\hat{a}^{\dagger}X^{-}}
Blue sideband/Anti-Jaynes CummingsBSB(z)BSB(z)eiza^†X++iz∗a^X−e^{iz\hat{a}^{\dagger}X^{+}+iz^{*}\hat{a}X^{-}}
Controlled rotationCR(θ)CR(\theta)eiθZa^†a^e^{i\theta Z\hat{a}^{\dagger}\hat{a}}
Controlled displacementCD(z)CD(z)eZ(za^†−z∗a^)e^{Z(z\hat{a}^{\dagger}-z^{*}\hat{a})}
Controlled squeezingCS(z)CS(z)eZ(z∗a^a^−za^†a^†)/2e^{Z(z^{*}\hat{a}\hat{a}-z\hat{a}^{\dagger}\hat{a}^{\dagger})/2}
Controlled beam splitterCBS(z)CBS(z)eZ(za^†b^−z∗a^b^†)e^{Z(z\hat{a}^{\dagger}\hat{b}-z^{*}\hat{a}\hat{b}^{\dagger})}

Table 1. Hybrid qubit/qumode gates for the simulation of QED in 2+1 dimensions. The gates are parametrized in terms of a complex variable z=θeiϕz=\theta e^{i\phi} with θ≥0\theta\ge0 and ϕ∈[0,2π)\phi\in[0,2\pi).

The gates above are accessible in several hybrid architectures:

  • Trapped ions and neutral atoms: Qubits can be realized in terms of different electronic states or specifically hyperfine splittings of states. Qumodes can be realized in terms of vibrational modes of the ions or atoms. For example, for trapped ions, collective vibrational modes of the ion chain are particularly useful to realize qumodes due to their resilience to environmental noise. Conditional displacements can be realized by spin-dependent forces. See Refs. [52–59].

  • Superconducting circuits: Qubits are encoded in the nonlinear degrees of freedom of Josephson junctions, such as transmons or fluxonium qubits, while qumodes are realized in terms of electromagnetic modes of microwave resonators or cavities. The nonlinearity of the Josephson element enables tunable qubit–qumode couplings, including cross-Kerr interactions and conditional displacements that serve as building blocks for hybrid gates. See Refs. [42, 47, 62] [60, 61].

  • Cavity QED and optomechanical systems: These platforms realize light-matter interactions through radiation-pressure or dispersive couplings. In cavity-QED systems, a two-level system interacts with a cavity photon mode, forming a qubit/qumode system. Optomechanical systems couple a cavity field to a mechanical oscillator, providing a purely continuous-variable system. See Refs. [63, 64, 65].

These physical primitives motivate the gate-level decompositions developed later, but are not yet tied to a specific field-theoretic mapping. Depending on the hardware platform, qubits or qumodes may exhibit longer or shorter coherence times. The connectivity, native interaction strengths, and the set of experimentally accessible gates can also vary significantly, which can impact the construction of hybrid quantum algorithms.

The definitions above provide the mathematical and operational background for hybrid computations. In § 3, we turn to the independent physics of the U(1) gauge theory in 2+1 dimensions, deriving its Hamiltonian formulation, symmetries, and spectrum without reference to any encoding. The hybrid representation of these operators and their simulation on quantum hardware will then be introduced in § 4.

U(1) lattice gauge theory with fermions in 2+1 dimensions

Before constructing quantum simulation protocols, we first summarize the physics of lattice quantum electrodynamics (QED) in 2+1 dimensions. In this section, we describe the Hamiltonian formulation of an Abelian U(1) gauge theory coupled to fermionic matter [6], including its electric and magnetic components, local gauge symmetry, and Gauss’s law constraints. We derive the reduction of degrees of freedom imposed by gauge invariance, analyze the exactly solvable single-plaquette case, and obtain the normal-mode spectrum for extended lattices. These results establish the physical quantities (energy scales, excitation gaps, and constraint structures) that will be encoded and simulated using the hybrid qubit–qumode framework introduced in § 4.

Hamiltonian formulation

The Hamiltonian approach to lattice gauge theory provides a natural framework for describing the real-time dynamics of gauge fields coupled to fermionic matter. For an Abelian U(1) gauge theory discretized on a two-dimensional spatial lattice with spacing aa, the Hamiltonian separates into electric and magnetic components [6]

HE=g22∑nEn,i2,HB=−12g2a2∑n(Pn+Pn†−2)(9)H_E = \frac{g^2}{2}\sum_n E_{n,i}^2,\qquad H_B = -\frac{1}{2g^2a^2}\sum_n\left(P_n+P_n^\dagger-2\right) \tag*{(9)}

where n≡(nx,ny)n \equiv(n_x,n_y) labels lattice sites, i∈{x,y}i \in\{x,y\} denotes link directions, gg is the gauge coupling with dimension [g]=1/2[g]=1/2, and aa is the lattice spacing with dimension [a]=−1[a]=-1. We will work with units where a=1a=1.

The operator En,iE_{n,i} represents the lattice electric field on the link emanating from site nn in direction ii, and its conjugate link variable

Un,i=eiχn,i∈U(1)(10)U_{n,i}=e^{i\chi_{n,i}}\in U(1) \tag*{(10)}

encodes the vector potential. They satisfy the canonical commutation relation

[En,i,Un′,j]=δnn′δijUn,j(11)[E_{n,i},U_{n',j}]=\delta_{nn'}\delta_{ij}U_{n,j} \tag*{(11)}

which mirrors the continuum relation between the electric field and vector potential. The magnetic part of the Hamiltonian is written in terms of oriented products of link variables around each plaquette,

Pn≡Un,xUn+ex,yUn+ey,x†Un,y†(12)P_n\equiv U_{n,x}U_{n+e_x,y}U^\dagger_{n+e_y,x}U^\dagger_{n,y} \tag*{(12)}

so that HBH_B reproduces the lattice curl of the vector potential. In the continuum limit, it approaches the familiar expression HB→12∫d2xB2H_B\to\frac{1}{2}\int d^2x B^2, with B=∂xAy−∂yAxB=\partial_xA_y-\partial_yA_x.

Charged fermionic matter fields residing on lattice sites can be incorporated through the Kogut–Susskind (staggered) formulation [6]. Their addition entails two more contributions to the Hamiltonian:

HM=m0∑n(−1)nx+nyΨn†Ψn,HK=12∑n∑i∈{x,y}[Ψn†Un,ei†Ψn+ei+h.c.](13)H_M=m_0\sum_n(-1)^{n_x+n_y}\Psi_n^\dagger\Psi_n,\qquad H_K=\frac{1}{2}\sum_n\sum_{i\in\{x,y\}}\left[\Psi_n^\dagger U^\dagger_{n,e_i}\Psi_{n+e_i}+\mathrm{h.c.}\right] \tag*{(13)}

where m0m_0 is the bare mass of the fermionic field Ψn\Psi_n.

The full Hamiltonian is therefore

H=HE+HB+HM+HK.(14)H=H_E+H_B+H_M+H_K. \tag*{(14)}

This model, quantum electrodynamics in 2+1 dimensions, retains the essential interplay between electric and magnetic fields, and includes massive charged fermions that can form bound states or particle–antiparticle pairs. In the continuum limit, m0,g→0m_0,g\to0, and the number of plaquettes diverges. The Hamiltonian in the continuum is labeled by a single dimensionless parameter, m0/g2m_0/g^2.

The full Hamiltonian (14) is gauge invariant, commuting with the generators GnG_n of local gauge transformations,

Gn≡∑i∈{x,y}(En,i−En+ei,i)−Qn,(15)G_n \equiv\sum_{i\in\{x,y\}} \left(E_{n,i} - E_{n+e_i,i}\right) - Q_n, \tag*{(15)}

where QnQ_n is the charge operator at lattice site nn. In the staggered fermion formulation, it is given by

Qn=Ψn†Ψn−1−(−1)nx+ny21.(16)Q_n = \Psi_n^\dagger\Psi_n - \frac{1-(-1)^{n_x+n_y}}{2}\mathbb{1}. \tag*{(16)}

The Hilbert space of physical states is restricted by imposing the Gauss law constraints,

Gn∣phys⟩=0.(17)G_n \lvert\mathrm{phys}\rangle= 0. \tag*{(17)}

The total charge vanishes for systems with open boundaries and neutral backgrounds,

Qtotal=∑nQn=0.(18)Q_{\mathrm{total}} = \sum_n Q_n = 0. \tag*{(18)}

It is possible to add non-dynamical background charges QnQ_n by modifying Gauss’s law to Gn∣phys⟩=Qn∣phys⟩G_n \lvert\mathrm{phys}\rangle= Q_n \lvert\mathrm{phys}\rangle, thus altering the physical content of the system in the continuum. Here, we will set all background charges to zero. It is straightforward to add them to our discussion.

The constraints drastically reduce the number of independent degrees of freedom. For a lattice of (2N+1)×(2N+1)(2N+1) \times(2N+1) plaquettes, the total number of links is 4(N+1)(2N+1)4(N+1)(2N+1), while there are 4(N+1)2−14(N+1)^2-1 independent Gauss-law conditions. Hence, the number of physical gauge degrees of freedom equals (2N+1)2(2N+1)^2, matching the number of plaquettes – one dynamical variable per plaquette corresponding to the magnetic flux through that cell.

We emphasize that all results in this Section are derived independently of any quantum encoding; the expressions for the Hamiltonian, Gauss-law constraints, and plaquette structure provide the physical foundation upon which the hybrid representation of Section 4 will act.

Single plaquette

The simplest system we can study is a single plaquette consisting of four links and four vertices in a chargeless background. For simplicity, we label the four vertices as (0,0)≡1(0,0)\equiv1, (0,1)≡2(0,1)\equiv2, (1,1)≡3(1,1)\equiv3 and (1,0)≡4(1,0)\equiv4. The electric field and link variables are relabeled according to Figure 1. Gauss’s law at the four vertices imposes the constraints

Figure 1. One plaquette system for the pure U(1)U(1) gauge theory before (left) and after (right) the selection of the dynamical degrees of freedom.

E12+E14=Q1,E23−E12=Q2,−E23−E43=Q3,E43−E14=Q4.(19)\begin{aligned} E_{12}+E_{14} &= Q_1,\\ E_{23}-E_{12} &= Q_2,\\ -E_{23}-E_{43} &= Q_3,\\ E_{43}-E_{14} &= Q_4. \tag*{(19)} \end{aligned}

Notice that they are compatible only if the total charge vanishes, see Eq. (18),

Qtotal=Q1+Q2+Q3+Q4=0.(20)Q_{\mathrm{total}} = Q_1+Q_2+Q_3+Q_4=0. \tag*{(20)}

Thus, there is only one independent gauge degree of freedom. Selecting the pair (E23,U23)(E_{23}, U_{23}) to be the dynamical gauge degree of freedom of the single plaquette system, we arrive at the electric and magnetic parts of the Hamiltonian,

HE=g22[E2+(E−Q2)2+(E+Q3)2+(E−Q1−Q2)2],HB=−12g2(U+U†−2)=1g2(1−cos⁡χ).(21)\begin{aligned} H_E &= \frac{g^2}{2}\left[E^2+(E-Q_2)^2+(E+Q_3)^2+(E-Q_1-Q_2)^2\right],\\ H_B &= -\frac{1}{2g^2}\left(U+U^\dagger-2\right)=\frac{1}{g^2}(1-\cos\chi). \tag*{(21)} \end{aligned}

Here we simplified the notation by denoting E≡E23E\equiv E_{23} and U≡U23=eiχU\equiv U_{23}=e^{i\chi}. The two additional contributions from fermions are

HM=m0∑i=14(−1)1+iΨi†Ψi,H_M=m_0\sum_{i=1}^{4}(-1)^{1+i}\Psi_i^\dagger\Psi_i,
HK=12(Ψ1†Ψ2+Ψ1†Ψ4+Ψ2UΨ3†+Ψ4Ψ3†)+h.c.(22)H_K=\frac{1}{2}\left(\Psi_1^\dagger\Psi_2+\Psi_1^\dagger\Psi_4+\Psi_2U\Psi_3^\dagger+\Psi_4\Psi_3^\dagger\right)+\mathrm{h.c.} \tag*{(22)}

Fermions live in a 16-dimensional Hilbert space. However, the requirement of a vanishing total charge restricts the physical space to 6 dimensions.

In the static charge limit (large m0m_0), the fermionic Hamiltonian is dominated by HMH_M. The fermionic Hilbert space is spanned by 16 basis vectors which can be obtained by repeatedly acting with modes Ψi,Ψi†\Psi_i,\Psi_i^\dagger (i=1,2,3,4i=1,2,3,4) on the vacuum state ∣Ωf⟩|\Omega_f\rangle which is annihilated by Ψ1,3,Ψ2,4†\Psi_{1,3}, \Psi^\dagger_{2,4}. There are 6 states with Qtotal=0Q_{\mathrm{total}}=0,

∣v1⟩≡∣Ωf⟩,∣v2⟩≡Ψ2Ψ3†∣Ωf⟩,∣v3⟩≡Ψ3†Ψ4∣Ωf⟩,=∣vvvv⟩=∣ve−e+v⟩=∣vve+e−⟩(23)\begin{aligned} |v_1\rangle&\equiv|\Omega_f\rangle, & |v_2\rangle&\equiv\Psi_2\Psi^\dagger_3|\Omega_f\rangle, & |v_3\rangle&\equiv\Psi^\dagger_3\Psi_4|\Omega_f\rangle, \\ &=|vvvv\rangle& &=|ve^-e^+v\rangle& &=|vv e^+e^-\rangle \tag*{(23)} \end{aligned}
∣v4⟩≡Ψ1†Ψ2∣Ωf⟩,∣v5⟩≡Ψ1†Ψ4∣Ωf⟩,∣v6⟩≡Ψ1†Ψ2Ψ3†Ψ4∣Ωf⟩,=∣e+e−vv⟩=∣e+vve−⟩=∣e+e−e+e−⟩.\begin{aligned} |v_4\rangle&\equiv\Psi^\dagger_1\Psi_2|\Omega_f\rangle, & |v_5\rangle&\equiv\Psi^\dagger_1\Psi_4|\Omega_f\rangle, & |v_6\rangle&\equiv\Psi^\dagger_1\Psi_2\Psi^\dagger_3\Psi_4|\Omega_f\rangle, \\ &=|e^+e^-vv\rangle& &=|e^+vve^-\rangle& &=|e^+e^-e^+e^-\rangle. \end{aligned}

Their energies, the eigenvalues of HMH_M, are (−2m0,0,2m0)(-2m_0,0,2m_0), with a four-fold degenerate zero eigenvalue corresponding to a particle-antiparticle pair. Their charges are

Q1,3∣e+⟩=∣e+⟩,Q1,2,3,4∣v⟩=0,Q2,4∣e−⟩=−∣e−⟩,(24)Q_{1,3}|e^+\rangle=|e^+\rangle,\qquad Q_{1,2,3,4}|v\rangle=0,\qquad Q_{2,4}|e^-\rangle=-|e^-\rangle, \tag*{(24)}

in accord with the staggered fermion prescription.

The full Hamiltonian

H=HE+HB+HM+HK,(25)H=H_E+H_B+H_M+H_K, \tag*{(25)}

contains a term which is linear in EE, which is of the form

−g2AE,A=Q1+2Q2−Q3.(26)-g^2AE,\qquad A=Q_1+2Q_2-Q_3. \tag*{(26)}

It can be eliminated by performing the unitary transformation

U=ei4Ax,(27)U=e^{\frac{i}{4}Ax}, \tag*{(27)}

which commutes with the charges and does not affect HMH_M. After some algebra, we obtain

H~≡U†HU=H~g+H~f+H~K,(28)\widetilde{H}\equiv U^\dagger H U=\widetilde{H}_g+\widetilde{H}_f+\widetilde{H}_K, \tag*{(28)}

where

H~g=2g2E2+1g2(1−cos⁡χ),H~f=HM+g28(3(Q1+Q3)−4Q2Q4−Q1Q3),H~K=12ei4x(Ψ1†Ψ2+Ψ4†Ψ1+Ψ3Ψ2†+Ψ4Ψ3†)+h.c.(29)\begin{aligned} \widetilde{H}_g&=2g^2E^2+\frac{1}{g^2}(1-\cos\chi),\\ \widetilde{H}_f&=H_M+\frac{g^2}{8}\left(3(Q_1+Q_3)-4Q_2Q_4-Q_1Q_3\right),\\ \widetilde{H}_K&=\frac{1}{2}e^{\frac{i}{4}x}\left(\Psi^\dagger_1\Psi_2+\Psi^\dagger_4\Psi_1+\Psi_3\Psi^\dagger_2+\Psi_4\Psi^\dagger_3\right)+\text{h.c.} \tag*{(29)} \end{aligned}

The eigenstates of the gauge part of the Hamiltonian, H~g\widetilde{H}_g,

H~g∣Φn,g⟩=En,g∣Φn,g⟩,(30)\widetilde{H}_g|\Phi_{n,g}\rangle=E_{n,g}|\Phi_{n,g}\rangle, \tag*{(30)}

can be written analytically in terms of Mathieu functions (n=0,1,2,…n = 0, 1, 2, \ldots)

En,g={g22an(−1g4)+1g2,n even,g22bn+1(−1g4)+1g2,n odd,⟷Φn,g(χ)∝{cen(χ2,−1g4),n even,sen+1(χ2,−1g4),n odd.(31)E_{n,g} = \begin{cases} \frac{g^2}{2}a_n\left(-\frac{1}{g^4}\right)+\frac{1}{g^2}, & n\ \text{even},\\ \frac{g^2}{2}b_{n+1}\left(-\frac{1}{g^4}\right)+\frac{1}{g^2}, & n\ \text{odd}, \end{cases} \quad\longleftrightarrow\quad \Phi_{n,g}(\chi)\propto \begin{cases} \mathrm{ce}_n\left(\frac{\chi}{2},-\frac{1}{g^4}\right), & n\ \text{even},\\ \mathrm{se}_{n+1}\left(\frac{\chi}{2},-\frac{1}{g^4}\right), & n\ \text{odd}. \end{cases} \tag*{(31)}

The eigenstates of the static fermionic part, H~f\widetilde{H}_f, are given by Eq. (23),

H~f∣vi⟩=Ei,f∣vi⟩,(32)\widetilde{H}_f|v_i\rangle=E_{i,f}|v_i\rangle, \tag*{(32)}

where

E1,f=−2m0,E2,f=E3,f=E4,f=E5,f=3g28,E6,f=2m0+g28.(33)E_{1,f}=-2m_0,\quad E_{2,f}=E_{3,f}=E_{4,f}=E_{5,f}=\frac{3g^2}{8},\quad E_{6,f}=2m_0+\frac{g^2}{8}. \tag*{(33)}

In the static limit, we can ignore H~K\widetilde{H}_K. We deduce the zeroth-order energy levels

En,i(0)=En,g+Ei,f⟷∣Ψn,i(0)⟩=U∣Φn,g⟩⊗∣vi⟩.(34)E_{n,i}^{(0)}=E_{n,g}+E_{i,f}\quad\longleftrightarrow\quad|\Psi_{n,i}^{(0)}\rangle=U|\Phi_{n,g}\rangle\otimes|v_i\rangle. \tag*{(34)}

The ground state is

∣Ψ0,1(0)⟩=∫−ππdχ ce0(χ2,−1g4)∣χ⟩⊗∣vvvv⟩,(35)|\Psi_{0,1}^{(0)}\rangle=\int_{-\pi}^{\pi}d\chi\,\mathrm{ce}_0\left(\frac{\chi}{2},-\frac{1}{g^4}\right)|\chi\rangle\otimes|vvvv\rangle, \tag*{(35)}

and its energy is E0,1(0)=g22a0(1/g4)+1/g2−2m0E_{0,1}^{(0)}=\frac{g^2}{2}a_0(1/g^4)+1/g^2-2m_0. The first excited state is

∣Ψ1,1(0)⟩=∫−ππdχ se2(χ2,−1g4)∣χ⟩⊗∣vvvv⟩,(36)|\Psi_{1,1}^{(0)}\rangle=\int_{-\pi}^{\pi}d\chi\,\mathrm{se}_2\left(\frac{\chi}{2},-\frac{1}{g^4}\right)|\chi\rangle\otimes|vvvv\rangle, \tag*{(36)}

with energy E1,1(0)=g22b2(1g4)+1g2−2m0E_{1,1}^{(0)}=\frac{g^2}{2}b_2\left(\frac{1}{g^4}\right)+\frac{1}{g^2}-2m_0. We obtain an energy gap of ΔE=E1,1(0)−E0,1(0)∼O(1)\Delta E=E_{1,1}^{(0)}-E_{0,1}^{(0)}\sim\mathcal{O}(1).

Away from the static limit, we can treat HKH_K as a perturbation which amounts to an expansion in m0−1m_0^{-1}. For the ground state, first-order perturbation produces no shift in energy. At the second order, we obtain the energy level

E0=E0,1(0)−∑(n,i)≠(0,1)∣⟨Ψ0,1(0)|HK|Ψn,i(0)⟩∣2En,i(0)−E0,1(0)+O(m0−2).(37)E_0=E_{0,1}^{(0)}-\sum_{(n,i)\ne(0,1)}\frac{\left|\left\langle\Psi_{0,1}^{(0)}\middle|H_K\middle|\Psi_{n,i}^{(0)}\right\rangle\right|^2}{E_{n,i}^{(0)}-E_{0,1}^{(0)}}+\mathcal{O}(m_0^{-2}). \tag*{(37)}

After some algebra, we obtain

E0=−2m0+g22a0(1g4)+1g2−12m0+O(m0−2).(38)E_0=-2m_0+\frac{g^2}{2}a_0\left(\frac{1}{g^4}\right)+\frac{1}{g^2}-\frac{1}{2m_0}+\mathcal{O}(m_0^{-2}). \tag*{(38)}

A comparison of the perturbative results and exact diagonalization (ED) is shown in Figure 2. As expected, the two results agree in the large mass region.

Figure 2. Comparison of the perturbative analysis (PT) in Eq. (38) with exact diagonalization (ED). We show the ground state energy (left) and the energy gap (right) for different values of the mass m0m_0.

Square lattice

Next, we generalize the above results to a square lattice with (2N+1)×(2N+1)(2N+1)\times(2N+1) plaquettes. We consider a lattice with open boundary conditions. This system has 4(N+1)24(N+1)^2 lattice sites and 4(2N+1)(N+1)4(2N+1)(N+1) links. Each link accommodates a gauge degree of freedom. These degrees of freedom are constrained by Gauss’s Law at each lattice site. However, we have 4(N+1)2−14(N+1)^2-1 independent constraints due to the vanishing total charge of the system. Therefore, the number of independent gauge degrees of freedom is 4(2N+1)(N+1)−4(N+1)2+1=(2N+1)24(2N+1)(N+1)-4(N+1)^2+1=(2N+1)^2. Let n=(nx,ny)\mathbf{n}=(n_x,n_y) label a lattice site with nx,ny=0,1,…,2N+1n_x,n_y=0,1,\ldots,2N+1. We will fix the gauge so that all link variables are set to the identity 11 except the horizontal link variables Un,x=eiχn,x∈U(1)U_{\mathbf{n},x}=e^{i\chi_{\mathbf{n},x}}\in U(1) with nx,ny=0,1,…,2Nn_x,n_y=0,1,\ldots,2N and χn,x∈[−π,π)\chi_{\mathbf{n},x}\in[-\pi,\pi), which are chosen as our (2N+1)2(2N+1)^2 independent gauge degrees of freedom. Using Gauss’s law (15), we can express the electric field in terms of the independent fields En,xE_{\mathbf{n},x}, which are conjugate to the link variables,

[En,x,Un′,x]=δnn′Un,x,[En,x,χn′,x]=−iδnn′(39)[E_{\mathbf{n},x},U_{\mathbf{n}',x}]=\delta_{\mathbf{n}\mathbf{n}'}U_{\mathbf{n},x},\qquad[E_{\mathbf{n},x},\chi_{\mathbf{n}',x}]=-i\delta_{\mathbf{n}\mathbf{n}'} \tag*{(39)}

We deduce the electric and magnetic parts of the Hamiltonian,

HE=g2∑n,n′(Hnn′(2)En,xEn′,x+Hnn′(1)QnEn′,x+Hnn′(0)QnQn′),H_E=g^2\sum_{\mathbf{n},\mathbf{n}'}\left(\mathcal{H}^{(2)}_{\mathbf{n}\mathbf{n}'}E_{\mathbf{n},x}E_{\mathbf{n}',x}+\mathcal{H}^{(1)}_{\mathbf{n}\mathbf{n}'}Q_{\mathbf{n}}E_{\mathbf{n}',x}+\mathcal{H}^{(0)}_{\mathbf{n}\mathbf{n}'}Q_{\mathbf{n}}Q_{\mathbf{n}'}\right),
HB=1g2∑n(1−cos⁡(χn,x−χn+ey,x))(40)H_B=\frac{1}{g^2}\sum_{\mathbf{n}}\left(1-\cos(\chi_{\mathbf{n},x}-\chi_{\mathbf{n}+\mathbf{e}_y,x})\right) \tag*{(40)}

For the electric part, HEH_E, we sum over all lattice sites. Since we are working with En,xE_{\mathbf{n},x} for nx,ny<2N+1n_x,n_y<2N+1, we have Hnn′(2)=0\mathcal{H}^{(2)}_{\mathbf{n}\mathbf{n}'}=0, if one of nx,ny,nx′,ny′n_x,n_y,n'_x,n'_y is 2N+12N+1. Similarly, Hnn′(1)=0\mathcal{H}^{(1)}_{\mathbf{n}\mathbf{n}'}=0, if one of nx=2N+1n_x = 2N + 1, or ny=2N+1n_y = 2N + 1. For the magnetic part, HBH_B, the sum runs over lattice sites with nx,ny<2N+1n_x,n_y < 2N + 1.

After gauge fixing, the fermionic contributions are

HM=m0∑n(−1)nx+nyΨn†Ψn,H_M = m_0 \sum_n (-1)^{n_x+n_y} \Psi_n^\dagger\Psi_n ,
HK=12∑nΨn†Un,xΨn+ex+12∑nΨn†Ψn+ey+h.c..(41)H_K = \frac{1}{2} \sum_n \Psi_n^\dagger U_{n,x} \Psi_{n+e_x} + \frac{1}{2} \sum_n \Psi_n^\dagger\Psi_{n+e_y} + \mathrm{h.c.} . \tag*{(41)}

We consider the weak-coupling and heavy-fermion limit, where we can obtain explicit analytic expressions. The weak coupling limit is obtained by rescaling En,x→En,x/gE_{n,x} \to E_{n,x}/g, χn,x→gχn,x\chi_{n,x} \to g\chi_{n,x}. We also expand in 1/m01/m_0. At leading order, the Hamiltonian is

H=Hg(0)+HM+O(1),(42)H = H_g^{(0)} + H_M + \mathcal{O}(1), \tag*{(42)}

where the gauge field contribution is

Hg(0)=∑n,n′Hnn′(2)En,xEn′,x+12∑n(χn,x−χn+ey,x)2.(43)H_g^{(0)} = \sum_{n,n'} \mathcal{H}^{(2)}_{nn'} E_{n,x} E_{n',x} + \frac{1}{2} \sum_n \left(\chi_{n,x} - \chi_{n+e_y,x}\right)^2 . \tag*{(43)}

Thus, the bosonic gauge part consists of (2N+1)2(2N + 1)^2 coupled harmonic oscillators. By diagonalizing the matrix Hnn′(2)\mathcal{H}^{(2)}_{nn'}, we obtain the frequencies of the normal modes,

ωkg=2sin⁡2πkx4(N+1)+2sin⁡2((kx+1)π2+πky4(N+1)),kx,ky=1,…,2N+1.(44)\omega_k^g = 2\sin^2 \frac{\pi k_x}{4(N+1)} + 2\sin^2 \left(\frac{(k_x+1)\pi}{2} + \frac{\pi k_y}{4(N+1)}\right), \qquad k_x,k_y = 1,\ldots,2N+1. \tag*{(44)}

The gap in the bosonic sector is given by the lowest normal frequency, 4sin⁡2π4(N+1)4\sin^2 \frac{\pi}{4(N+1)}, obtained by setting kx=ky=1k_x = k_y = 1,

ΔE=E1−E0=4sin⁡2π4(N+1)+O(1).(45)\Delta E = E_1 - E_0 = 4\sin^2 \frac{\pi}{4(N+1)} + \mathcal{O}(1) . \tag*{(45)}

As expected, the gap vanishes in the continuum limit (N→∞N \to\infty).

For the fermionic part, we obtain modes with frequencies ωf=±m0\omega^f = \pm m_0. We associate positive frequency modes with particles and negative frequency modes with antiparticles.

This completes the field-theoretic description of the U(1) gauge theory coupled to fermionic matter in 2+1 dimensions. The next section translates these dynamics into the language of hybrid quantum simulation, specifying how electric, magnetic, and matter operators are encoded and implemented using qubits and qumodes.

Quantum computation with qubits and qumodes

Having established the field-theoretic structure of U(1) gauge theory coupled to fermions in 2+1 dimensions, we now formulate a representation suitable for quantum simulation on hybrid architectures that combine discrete (qubit) and continuous (qumode) degrees of freedom. We express the full lattice Hamiltonian in terms of experimentally accessible qubit–qumode gates. This formulation will serve as the foundation for the numerical results presented in the subsequent section.

Hybrid representation of lattice operators

Each fermionic site n=(nx,ny)n=(n_x,n_y) is represented by a qubit whose computational basis encodes the local occupation number. In accord with the staggered fermion prescription, the single site state ∣0⟩n\lvert0\rangle_n represents a particle of charge +1+1 for odd nx+nyn_x+n_y and ∣1⟩n\lvert1\rangle_n is the vacuum state for these sites whereas for even nx+nyn_x+n_y, the vacuum is represented by ∣0⟩n\lvert0\rangle_n and an anti-particle with charge −1-1 by ∣1⟩n\lvert1\rangle_n. In this representation, the charge operators become projection operators acting on single qubits,

Qn↦Zn+(−1)nx+ny12.(46)Q_n \mapsto\frac{Z_n+(-1)^{n_x+n_y}\mathbb{1}}{2}. \tag*{(46)}

The fermionic creation and annihilation operators are mapped to qubit operators as

Ψn†↦Xn+,Ψn↦Xn−,(47)\Psi_n^\dagger\mapsto X_n^+,\quad\Psi_n\mapsto X_n^-, \tag*{(47)}

where X±X^\pm is defined in (2), with appropriate Jordan–Wigner strings ensuring fermionic anticommutation relations across sites.

Gauge fields residing on links (n,i)(n,i) with i∈{x,y}i\in\{x,y\} are represented by pairs of qumodes with quadratures qn,i=(qn,i0,qn,i1)\mathbf{q}_{n,i}=(q^0_{n,i},q^1_{n,i}) and pn,i=(pn,i0,pn,i1)\mathbf{p}_{n,i}=(p^0_{n,i},p^1_{n,i}), obeying the commutation rules

[qn,iμ,pn′,jν]=iδnn′δijδμν.(48)[q^\mu_{n,i},p^\nu_{n',j}]=i\delta_{nn'}\delta_{ij}\delta^{\mu\nu}. \tag*{(48)}

To connect the field-theoretic formulation outlined in Section 3 with a representation suitable for quantum simulation, we now promote the classical lattice variables (En,i,Un,i)(E_{n,i},U_{n,i}) to operators acting on hybrid qubit-qumode registers. The mapping below defines the correspondence between the canonical variables of the lattice theory and their quantum counterparts realized on the qumode Hilbert space:

En,i↦Jn,i≡qn,i0pn,i1−qn,i1pn,i0,Un,i↦qn,i0+iqn,i1.(49)E_{n,i}\mapsto J_{n,i}\equiv q^0_{n,i}p^1_{n,i}-q^1_{n,i}p^0_{n,i},\quad U_{n,i}\mapsto q^0_{n,i}+iq^1_{n,i}. \tag*{(49)}

This correspondence identifies the lattice electric field En,iE_{n,i} with the angular momentum operator Jn,iJ_{n,i} in the two-dimensional space of the pair of bosonic modes and the link operator Un,iU_{n,i} with a complexified coordinate. This identification ensures that local gauge transformations act on the hybrid operators in the same way as in the classical theory. Indeed, the unitary implementing gauge transformations is

Ugauge(θ)=∏neiθnGn,Gn=∑i∈{x,y}(Jn,i−Jn+ei,i)−Qn,(50)U_{\mathrm{gauge}}(\theta)=\prod_n e^{i\theta_nG_n},\quad G_n=\sum_{i\in\{x,y\}}(J_{n,i}-J_{n+e_i,i})-Q_n, \tag*{(50)}

cf. Eq. (15). Using the canonical commutation relations of the qumode quadratures (48), we obtain

Ugauge(θ)(qn,i0+iqn,i1)Ugauge†(θ)=ei(θn−θn+ei)(qn,i0+iqn,i1),U_{\mathrm{gauge}}(\theta)(q^0_{n,i}+iq^1_{n,i})U_{\mathrm{gauge}}^\dagger(\theta)=e^{i(\theta_n-\theta_{n+e_i})}(q^0_{n,i}+iq^1_{n,i}),
Ugauge(θ)Jn,iUgauge†(θ)=Jn,i.(51)U_{\mathrm{gauge}}(\theta)J_{n,i}U_{\mathrm{gauge}}^\dagger(\theta)=J_{n,i}. \tag*{(51)}

reproducing the transformation properties of the classical link variables. Thus the dictionary in Eq. (49) not only maps the lattice Hamiltonian to hybrid quantum operators but also guarantees gauge invariance at the operator level, ensuring that all subsequent algorithmic steps act within the gauge-invariant subspace.

By employing two qumodes for each link, we have enlarged the phase space. However, we have not broken gauge invariance because the electric field acts as angular momentum that leaves the unit circle qn,i2=1q_{n,i}^{2}=1 invariant, [Jn,i,qn,i2]=0[J_{n,i},q_{n,i}^{2}]=0. Consequently, qn,i2q_{n,i}^{2} is conserved as it commutes with the Hamiltonian. To restore compactness of the link variables Un,iU_{n,i}, we ought to impose the constraints

qn,i2=(qn,i0)2+(qn,i1)2=1.(52)q_{n,i}^{2}=(q_{n,i}^{0})^{2}+(q_{n,i}^{1})^{2}=1 . \tag*{(52)}

We introduce two complementary methods to restore compactness within this CV representation.

Compactification Method A

We create compact continuous variables through a squeezed-state projection that modifies the inner product in the qumode Hilbert space. Let Λ\Lambda be the effective squeezing parameter (to be defined precisely below). Wavefunctions are suppressed outside the unit circles (52), yielding expectation values that converge to the desired compact-group results as ∼Λ−2\sim\Lambda^{-2}. In the limit Λ→∞\Lambda\to\infty, each qumode effectively wraps onto the unit circle, realizing Un,iU_{n,i} as a true phase operator.

The gauge part of any state that we consider must be of the form

∣ψ(Q)⟩∝∫d2q δ(q2−1)ψ(q)∣q⟩,(53)|\psi^{(Q)}\rangle\propto\int d^{2}q\,\delta(q^{2}-1)\psi(q)|q\rangle, \tag*{(53)}

where the delta function enforces the unitarity constraint in Eq. (52). We denote with the superscript (Q)(Q) any state or operator used in the quantum simulation. To implement this state, we apply a non-Gaussian unitary D(s)\mathcal{D}(s) that entangles each of the system qumodes qμq_{\mu} (μ=0,1\mu=0,1) with an ancilla prepared in the quadrature eigenstate ∣pa=s⟩a|p_{a}=s\rangle_{a}, where ss is a free parameter. This unitary is defined as

Dμ(s)=e−isqμ2qa.(54)D_{\mu}(s)=e^{-isq_{\mu}^{2}q_{a}} . \tag*{(54)}

It can be implemented with beam splitters and cubic phase gates (5) as

Dμ(s)=BSμa† Va(−2s)⋅Vμ(2s)BSμa⋅Va(s)(55)D_{\mu}(s)=\mathrm{BS}_{\mu a}^{\dagger}\,V_{a}(-\sqrt{2}s)\cdot V_{\mu}(\sqrt{2}s)\mathrm{BS}_{\mu a}\cdot V_{a}(s) \tag*{(55)}

The circuit implementing this is shown in Figure 3. Then the circuit that creates ∣ψ(Q)⟩|\psi^{(Q)}\rangle from input ∣ψ⟩=∫d2q ψ(q)∣q⟩|\psi\rangle=\int d^{2}q\,\psi(q)|q\rangle is given in Figure 6.

Circuit implementing the non-Gaussian unitary

Figure 3. Circuit implementing the non-Gaussian unitary Dμ(s)D_{\mu}(s) given in Eq. (54).

Quantum circuit that creates the ground state coupled cluster ansatz

Figure 6. Quantum circuit that creates the ground state coupled cluster ansatz ∣ψ0(α,Λ)⟩|\psi_0(\alpha,\Lambda)\rangle, see Eq. (78), with radial cutoff Λ=g2ser\Lambda= \frac{g}{2}se^r.

In a realistic implementation, the initial state of the ancilla qumode in the quantum circuit of Figure 6 is approximated by a squeezed state S(r)∣0⟩aS(r)|0\rangle_{a}, where S(r)=er2(a2−a†2)S(r)=e^{\frac{r}{2}(a^{2}-a^{\dagger2})} is the squeezing operator. The modified quantum circuit yields the state

∣ψapprox(Q)(Λ)⟩∝∫d2q e−Λ2g2(q2−1)2ψ(q)∣q⟩,Λ=g2ser(56)|\psi_{\mathrm{approx}}^{(Q)}(\Lambda)\rangle\propto\int d^{2}q\,e^{-\frac{\Lambda^{2}}{g^{2}}(q^{2}-1)^{2}}\psi(q)|q\rangle,\quad\Lambda=\frac{g}{2}se^{r} \tag*{(56)}
Quantum circuit creating the state

Figure 4. The quantum circuit that creates the state ∣ψ(Q)⟩|\psi^{(Q)}\rangle, see Eq. (53), from a given input two-qumode state ∣Ψ⟩|\Psi\rangle, where DD is a displacement gate (see Table 1) and D\mathcal{D} is defined in Eq. (54).

which reduces to the desired state ∣ψ(Q)⟩|\psi^{(Q)}\rangle in Eq. (53) in the limit of infinite squeezing, r→∞r \to\infty. The unitarity constraint δ(q2−1)\delta(q^2 - 1) has been traded for a sharply peaked Gaussian.

Compactification Method B

An alternative way of enforcing compactness is by adding a penalty term to the Hamiltonian,

Hμ=μ2∑n,i(qn,i2−1)2.(57)H_{\mu} = \frac{\mu}{2} \sum_{n,i} (q_{n,i}^{2} - 1)^{2}. \tag*{(57)}

The penalty parameter μ>0\mu> 0 controls the trade-off between compactness and numerical stability. For large but finite μ\mu, low-energy states remain confined near the compact submanifold and reproduce the same spectrum as the exact theory up to corrections of order 1/μ1/\mu.

Both strategies are compatible with current CV technology. The first relies on experimentally achievable squeezing operations, while the second can be realized through engineered non-Gaussian CV gates.

As discussed in § 3, for a square lattice with (2N+1)×(2N+1)(2N + 1) \times(2N + 1) plaquettes and open boundary conditions, there are 4(N+1)24(N + 1)^2 fermionic sites, and after gauge fixing, (2N+1)2(2N + 1)^2 independent gauge degrees of freedom. Therefore, we need O(N2)O(N^2) qubits and qumodes to realize the lattice system with DV/CV hybrid quantum hardware. Thus, the size of the quantum system needed scales linearly with the volume of the lattice system we wish to simulate. The gate complexity scales polynomially with lattice size when local interactions are decomposed via first- or second-order Trotterization. Truncating each qumode to deffd_{\mathrm{eff}} basis states yields a finite-dimensional approximation with total Hilbert-space dimension 2(2N+1)2deff4(N+1)(2N+1)2^{(2N+1)^2}d_{\mathrm{eff}}^{4(N+1)(2N+1)}, enabling classical benchmarking for small lattices and near-term quantum hardware implementations for few-plaquette systems.

Since the independent gauge degrees of freedom can be chosen along links in the xx-direction, we will drop the index xx from these degrees of freedom in order to simplify the notation. Each can be simulated with a pair of qumodes qn=(qn0,qn1)\mathbf{q}_n = (q_n^0,q_n^1) with the electric field

En↦Jn=qn0qn1−qn1qn0E_n \mapsto J_n = q_n^0 q_n^1 - q_n^1 q_n^0 and the link Un↦qn0+iqn1U_n \mapsto q_n^0 + i q_n^1 (Eq. (49)). The various parts of the Hamiltonian (Eq. (40)) are implemented as

HE(Q)=g2∑n,n′(Hnn′(2)JnJn′+Hnn′(1)QnJn′+Hnn′(0)QnQn′) ,H_E^{(Q)} = g^2 \sum_{n,n'} \left( \mathcal{H}_{nn'}^{(2)} J_n J_{n'} + \mathcal{H}_{nn'}^{(1)} Q_n J_{n'} + \mathcal{H}_{nn'}^{(0)} Q_n Q_{n'} \right) \, ,
HB(Q)=12g2∑n(qn+ey−qn)2 ,H_B^{(Q)} = \frac{1}{2g^2} \sum_n \left(q_{n+e_y} - q_n\right)^2 \, ,
HM(Q)=m0∑n(−)nx+nyQn ,H_M^{(Q)} = m_0 \sum_n (-)^{n_x+n_y} Q_n \, ,
HK(Q)=12∑n(qn0−iqn1)Xn+Xn+ex−+12∑nPL(n,n+ey)Xn+Xn+ey−+h.c. ,(58)H_K^{(Q)} = \frac{1}{2} \sum_n (q_n^0 - i q_n^1) X_n^+ X_{n+e_x}^- + \frac{1}{2} \sum_n P_L(n,n+e_y) X_n^+ X_{n+e_y}^- + \text{h.c.} \, , \tag*{(58)}

where PL(n,n+ey)=∏n′∈L(n,n+ey)Zn′P_L(n,n+e_y) = \prod_{n'\in L(n,n+e_y)} Z_{n'} is a string of Pauli ZZ matrices along a path joining the sites nn and n+eyn+e_y. Notice that once the constraint ((52)) is imposed, the magnetic Hamiltonian HB(Q)H_B^{(Q)} reduces to the expected expression HBH_B. The form of HB(Q)H_B^{(Q)} we chose ensures numerical stability. For the Jordan-Wigner transformation we chose the path illustrated in Figure 5. This is the most convenient choice that leads to simple two-qubits-one-qumode interactions in the kinetic Hamiltonian. Different paths could entail longer Pauli strings interacting with the bosonic modes. The matrix H(2)\mathcal{H}^{(2)} has diagonal entries (n=n′n=n') equal to 2 and entries corresponding to nearest neighbors equal to −1/2-1/2. H(1)\mathcal{H}^{(1)} and H(0)\mathcal{H}^{(0)} can be obtained by solving Gauss Law at each vertex and choosing an ordered path for both bosons and fermions, independently.

Path for the Jordan-Wigner transformation from fermions to spin degrees of freedom

Figure 5. Path for the Jordan-Wigner transformation from fermions to spin degrees of freedom.

Time evolution

To study the time evolution of the hybrid qubit-qumodes system, we need to engineer the incremental evolution operator by a small time interval Δt\Delta t,

e−iΔtH(Q)≈e−iΔtHE(Q)e−iΔtHB(Q)e−iΔtHM(Q)e−iΔtHK(Q).(59)e^{-i\Delta t H^{(Q)}} \approx e^{-i\Delta t H_{E}^{(Q)}} e^{-i\Delta t H_{B}^{(Q)}} e^{-i\Delta t H_{M}^{(Q)}} e^{-i\Delta t H_{K}^{(Q)}} . \tag*{(59)}

Electric term

The electric contribution in Eq. (59) can be decomposed into factors of five different types,

{e−isJn, e−isJn2e−isJnJn′, e−isZnJn′, e−isZn, e−isZnZn′},(60)\left\{e^{-isJ_n},\ e^{-isJ_n^2}e^{-isJ_nJ_{n'}},\ e^{-isZ_nJ_{n'}},\ e^{-isZ_n},\ e^{-isZ_nZ_{n'}}\right\}, \tag*{(60)}

where n′≠nn' \ne n. The first type of factor is a pure bosonic gate. Suppose that JnJ_n acts on the pair of qubits with quadratures q=(q0,q1)\mathbf{q}=(q^0,q^1), so Jn=J01=q0p1−q1p0J_n=J^{01}=q^0p^1-q^1p^0. Using

J=F0†⋅BS⋅(N0−N1)⋅BS⋅F0,(61)J=F^{0\dagger}\cdot\mathrm{BS}\cdot(N^0-N^1)\cdot\mathrm{BS}\cdot F^0, \tag*{(61)}

where N=12(p2+q2)N=\frac{1}{2}(p^2+q^2) is the number operator, F=eiπ2NF=e^{i\frac{\pi}{2}N} is the Fourier transform operator, and BS represents a balanced beam splitter, we can express it as

e−isJ=F0†⋅BS†⋅R0(−s)⋅R1(s)⋅BS⋅F0.(62)e^{-isJ}=F^{0\dagger}\cdot\mathrm{BS}^{\dagger}\cdot R^0(-s)\cdot R^1(s)\cdot\mathrm{BS}\cdot F^0. \tag*{(62)}

The quantum circuit implementing e−isJe^{-isJ} is

Image not extracted

Figure 14 (PDF p. 19).

Similarly, the second term in (60) can be expressed as

e−isJ2=F0†⋅BS†⋅K0(−s)⋅K1(−s)⋅CK(2s)⋅BS⋅F0(63)e^{-isJ^2}=F^{0\dagger}\cdot\mathrm{BS}^{\dagger}\cdot K^0(-s)\cdot K^1(-s)\cdot\mathrm{CK}(2s)\cdot\mathrm{BS}\cdot F^0 \tag*{(63)}

in terms of single-qumode Kerr gates and two-qumode cross-Kerr gates (see Table 1). The quantum circuit implementing e−isJ2e^{-isJ^2} is

Image not extracted

Figure 15 (PDF p. 19).

The second type of factor is also purely bosonic, but it contains two different JJ operators acting on four qumodes. The quantum circuit implementing eisJn⊗Jn′e^{isJ_n\otimes J_{n'}} on 2 gauge degrees of The third type of factor, e−isZ⊗Je^{-isZ\otimes J}, is a hybrid unitary involving two qumodes on which JJ acts and one qubit on which ZZ acts. It can be expressed as the hybrid qubit-qumode gate CBS(z=−sz=-s) (see Table 1).

Quantum circuit diagram

Figure 16.

The remaining factors in the set (60) involve only Pauli matrices. They are implementable with standard qubit gates.

Magnetic term

The magnetic contribution in Eq. (59) can be decomposed into two different sets of purely bosonic contributions

{e−isq^2,e−isq^,e−is(q^1−q^2)2}.(64)\left\{e^{-is\hat{q}^{2}},e^{-is\hat{q}},e^{-is(\hat{q}_{1}-\hat{q}_{2})^{2}}\right\}. \tag*{(64)}

They are all Gaussian qumode gates that can be implemented with quadratic phase gates, displacements, and balanced beam splitters (see Table 1).

Kinetic term

The fermionic kinetic contribution contains purely qubit gates as well as hybrid gates of the form (see Eq. (58))

eisq^⊗σi⊗σj,(65)e^{is\hat{q}\otimes\sigma_i\otimes\sigma_j}, \tag*{(65)}

They can be implemented with a doubly controlled displacement targeting the qumode with qubits acting as control operators,

CCD⁡(s)=e−isq^⊗Z⊗Z=(1⊗CNOT⁡)⋅(e−isq^⊗Z⊗1)⋅(1⊗CNOT⁡),(66)\operatorname{CCD}(s)=e^{-is\hat{q}\otimes Z\otimes Z}=(1\otimes\operatorname{CNOT})\cdot\left(e^{-is\hat{q}\otimes Z}\otimes1\right)\cdot(1\otimes\operatorname{CNOT}), \tag*{(66)}

where we used the identity Z⊗Z=CNOT⁡⋅(Z⊗1)⋅CNOT⁡Z\otimes Z=\operatorname{CNOT}\cdot(Z\otimes1)\cdot\operatorname{CNOT}. In order to obtain the desired qubit controls we act with appropriate single qubit rotations. For example, for the unitary operator e−isq^⊗X⊗Xe^{-is\hat{q}\otimes X\otimes X}, we obtain

e−isq^⊗X⊗X=(H⊗H)⋅CCD(s)⋅(H⊗H),(67)e^{-is\hat{q}\otimes X\otimes X}=(H\otimes H)\cdot\mathrm{CCD}(s)\cdot(H\otimes H), \tag*{(67)}

where HH is the Hadamard matrix. The quantum circuit implementing the unitary evolution operator (67) is

Quantum circuit implementing the unitary evolution operator $e^{-is\hat{q}\otimes X\otimes X}$

Figure 17 (PDF p. 21).

The unitary evolution associated with the other terms in HK(Q)H_K^{(Q)} in Eq. (58) is implementable with standard qubit gates.

Mass term

Finally, the fermion mass term in the Hamiltonian (58) is implementable with single qubit ZZ rotations.

QITE

One way to obtain the ground state of any given theory is to exploit the rapid decay of the evolution operator evaluated at imaginary times [48] [49],

⟨ψ∣e−2τHH∣ψ⟩⟨ψ∣e−2τH∣ψ⟩=E0+O(e−2τ(E1−E0)),(68)\frac{\langle\psi|e^{-2\tau H}H|\psi\rangle}{\langle\psi|e^{-2\tau H}|\psi\rangle}=E_0+O\left(e^{-2\tau(E_1-E_0)}\right), \tag*{(68)}

where ∣ψ⟩|\psi\rangle is an arbitrary state with non-vanishing overlap with the ground state.

Unlike standard qubit-based QITE algorithms, the present formulation operates on a hybrid Hilbert space incorporating both discrete (qubit) and continuous (qumode) generators. Each imaginary-time update simultaneously adjusts matter amplitudes and gauge-field quadrature distributions while preserving the Gauss-law constraints. To our knowledge, this represents the first implementation of QITE in a hybrid DV–CV gauge-theory setting.

Rotating to the imaginary axis is a non-unitary procedure that necessarily involves a measurement. We introduce two different methods to engineer the necessary gates for QITE.

QITE Method A

For a Trotter step, we need to engineer the non-unitary operator e−ΔτO^e^{-\Delta\tau\hat{O}}, where O^\hat{O} is a Hermitian hybrid operator, e.g., O^=h^⊗P\hat{O}=\hat{h}\otimes P. Here, h^\hat{h} acts on a register of qumodes and PP is a string of Pauli operators acting on qubits.

We introduce an ancilla qubit in the state ∣0⟩a|0\rangle_a and act with the entangling unitary operator eiΔτO^⊗Yae^{i\Delta\tau\hat{O}\otimes Y_a}, where YaY_a is a Pauli matrix acting on the ancilla qubit, followed by a Hadamard gate HaH_a. After measuring the ancilla projecting it onto the ∣0⟩a\lvert0\rangle_a state, we obtain the operator

a⟨0∣HaeiΔτO^⊗Y^a∣0⟩a=12e−ΔτO^+O((Δτ)2),(69){}_a\langle0\lvert H_a e^{i\Delta\tau\hat{O}\otimes\hat{Y}_a}\lvert0\rangle_a = \frac{1}{\sqrt{2}}e^{-\Delta\tau\hat{O}}+\mathcal{O}((\Delta\tau)^2), \tag*{(69)}

The quantum circuit that implements this is

Quantum circuit implementing the Hadamard-gate method

Figure 18.

By repeatedly applying this circuit, we engineer the non-unitary e−τOe^{-\tau O} for finite imaginary time τ\tau.

QITE Method B

Alternatively, the non-unitary e−ΔτO^e^{-\Delta\tau\hat{O}} can be engineered with the aid of an ancilla qumode. An advantage of this method is that Δτ\Delta\tau does not need to be small. However, it can only be applied to Hermitian operators OO that can be brought into the form O^→o^2\hat{O}\to\hat{o}^{2}, where o^\hat{o} is a hybrid Hermitian operator. Assuming that this is the case, i.e.,

O^=U†o^2U(70)\hat{O}=U^{\dagger}\hat{o}^{2}U \tag*{(70)}

where UU is a unitary operator, we introduce an ancilla qumode in the vacuum state ∣0⟩A\lvert0\rangle_A [49], and apply the entangling non-Gaussian unitary gate

ei2Δτo^⊗p^A,(71)e^{i2\sqrt{\Delta\tau}\hat{o}\otimes\hat{p}_A}, \tag*{(71)}

where p^A\hat{p}_A is a quadrature operator acting on the ancilla qumode. We then measure the ancilla qumode with a photon detector. If the detector detects no photon, the measurement acts as a projection onto

A⟨0∣ei2Δτo^⊗p^A∣0⟩A∝e−Δτo^2.(72){}_A\langle0\lvert e^{i2\sqrt{\Delta\tau}\hat{o}\otimes\hat{p}_A}\lvert0\rangle_A \propto e^{-\Delta\tau\hat{o}^{2}}. \tag*{(72)}

After applying the unitary transformation (70), we obtain the desired non-unitary operator,

U†e−Δτo^2U=e−ΔτO^.(73)U^{\dagger}e^{-\Delta\tau\hat{o}^{2}}U=e^{-\Delta\tau\hat{O}}. \tag*{(73)}

The quantum circuit that implements this is

Quantum circuit implementing the ancilla-qumode method

Figure 19.

For example, consider O^=2g2J2\hat{O}=2g^{2}J^{2} which contributes to the electric part of the Hamiltonian. In this case, o^=2gJ\hat{o}=\sqrt{2}gJ and U=1U=1.

The hybrid framework developed in this Section provides general algorithms for the calculation of physical properties, such as the QITE algorithm for preparing ground and low-lying excited states of lattice gauge theories encoded on qubit–qumode architectures. To illustrate its implementation and to benchmark convergence and gauge-invariance preservation, we apply our method to the minimal non-trivial instance of U(1)U(1) quantum electrodynamics in 2+12+1 dimensions, the single-plaquette model, in § 5. This system contains the full structure of the theory, including compact electric and magnetic terms, local Gauss-law constraints, and fermionic matter coupling, yet remains analytically tractable for comparison with exact diagonalization. The single-plaquette problem thus serves as a testbed for validating the hybrid encoding, assessing the effectiveness of the squeezing- and Hamiltonian-penalty compactness schemes, and quantifying the accuracy of hybrid algorithms in reproducing gauge-invariant dynamics.

Results for a single plaquette

To illustrate our hybrid quantum approach introduced in § 4, in this Section we derive explicit results for a lattice consisting of a single plaquette and compare them with analytic expressions and exact diagonalization (ED).

The Hamiltonian

In our hybrid quantum computation, we implement the gauge field with a pair of qumodes of quadratures q=(q0,q1)\boldsymbol{q}=(q^{0},q^{1}) and p=(p0,p1)\boldsymbol{p}=(p^{0},p^{1}) obeying the commutation rules [qμ,pν]=iδμν[q^{\mu},p^{\nu}]=i\delta^{\mu\nu}, and the fermionic field with four qubits. After imposing three of the Gauss Law constraints, the remaining variable EE is implemented as an angular momentum J=q0p1−q1p0J=q^{0}p^{1}-q^{1}p^{0}, with the corresponding link variable implemented as U=q0+iq1U=q^{0}+iq^{1}. This enlarges the phase space, but it does not break gauge invariance because the angular momentum JJ leaves the unit circle q2=1\boldsymbol{q}^{2}=1 invariant, [J,q2]=0[J,\boldsymbol{q}^{2}]=0. As a consequence, q2\boldsymbol{q}^{2} is conserved, i.e. [H,q2]=0[H,\boldsymbol{q}^{2}]=0.

For the fermionic field, we employ the Jordan-Wigner transformation, implementing it via Pauli matrices,

Ψ1↦X1−,Ψ2↦−iZ1X2−,Ψ3↦−Z1Z2X3−,Ψ4↦iZ1Z2Z3X4−,(74)\Psi_{1}\mapsto X_{1}^{-},\qquad\Psi_{2}\mapsto-iZ_{1}X_{2}^{-},\qquad\Psi_{3}\mapsto-Z_{1}Z_{2}X_{3}^{-},\qquad\Psi_{4}\mapsto iZ_{1}Z_{2}Z_{3}X_{4}^{-}, \tag*{(74)}

where X±=12(X±iY)X^{\pm}=\frac{1}{2}(X\pm iY) and the indices indicate the position of the qubit the Pauli matrix acts on. In this representation, the charge operators become projection operators acting on single qubits,

Q1↦Z1+12,Q2↦Z2−12,Q3↦Z3+12,Q4↦Z4−12.(75)Q_{1}\mapsto\frac{Z_{1}+1}{2},\qquad Q_{2}\mapsto\frac{Z_{2}-1}{2},\qquad Q_{3}\mapsto\frac{Z_{3}+1}{2},\qquad Q_{4}\mapsto\frac{Z_{4}-1}{2}. \tag*{(75)}

We model the system with the Hamiltonian

H(Q)=HE(Q)+HB(Q)+HM(Q)+HK(Q)(76)H^{(Q)}=H_{E}^{(Q)}+H_{B}^{(Q)}+H_{M}^{(Q)}+H_{K}^{(Q)} \tag*{(76)}

where

HE(Q)=g22[J2+(J−Q2)2+(J+Q3)2+(J−Q1−Q2)2],HB(Q)=12g2[(q0−1)2+(q1)2],HM(Q)=m0∑i=14(−1)1+iQi,HK(Q)=12(X1+X2+(q0−iq1)X3+X2−+X3+X4−+X1+Z2Z3X4−)+h.c.(77)\begin{aligned} H_E^{(Q)} &= \frac{g^2}{2}\left[J^2+(J-Q_2)^2+(J+Q_3)^2+(J-Q_1-Q_2)^2\right],\\ H_B^{(Q)} &= \frac{1}{2g^2}\left[(q^0-1)^2+(q^1)^2\right],\\ H_M^{(Q)} &= m_0\sum_{i=1}^{4}(-1)^{1+i}Q_i,\\ H_K^{(Q)} &= \frac{1}{2}\left(X_1^+X_2+(q^0-iq^1)X_3^+X_2^-+X_3^+X_4^-+X_1^+Z_2Z_3X_4^-\right)+\mathrm{h.c.} \tag*{(77)} \end{aligned}

and J=q0p1−q1p0J=q^0p^1-q^1p^0.

The remaining Gauss’s Law constraint (20) imposing vanishing total charge will be implemented by only considering states obeying (20). This restricts the 16-dimensional fermionic Hilbert space to a six-dimensional subspace. Thus, it is possible to describe the system with three qubits. However, the representation of fermionic operators becomes cumbersome, resulting in a longer quantum circuit. In the following, we describe the system with four qubits.

Variational ansatz

We consider the pure U(1) gauge theory and the ground state ansatz

∣ψ0(α,Λ)⟩∝∫d2q e−Λ2g2(q2−1)2eαq0e−12q2∣q⟩(78)|\psi_0(\alpha,\Lambda)\rangle\propto\int d^2q\,e^{-\frac{\Lambda^2}{g^2}(q^2-1)^2}e^{\alpha q^0}e^{-\frac{1}{2}q^2}|q\rangle \tag*{(78)}

which can be engineered with the quantum circuit depicted in Figure 6 with the gate parameters r,sr,s chosen so that Λ=g2ser\Lambda=\frac{g}{2}se^r (Eq. (56)). Without fermions, the Hamiltonian operator is given by

Hg(Q)=HE(Q)+HM(Q)(79)H_g^{(Q)}=H_E^{(Q)}+H_M^{(Q)} \tag*{(79)}

with vanishing charges, Qi=0Q_i=0 (i=1,2,3,4i=1,2,3,4). We obtain the energy of the ansatz (78) in terms of an integral over the unphysical radial coordinate R=q2R=\sqrt{q^2} and Bessel functions we obtain by integrating of the angular variable χ=arctan⁡q1q0\chi= \arctan\frac{q^{1}}{q^{0}},

ϵ0(α,Λ)≡⟨ψ0(α,Λ)∣Hg(Q)∣ψ0(α,Λ)⟩=∫0∞dR Re−2Λ2g2(R2−1)2−R2((R2+1)I0(2Rα)+2R(g4α−1)I1(2Rα))2g2∫0∞dR Re−2Λ2g2(R2−1)2−R2I0(2Rα).(80)\begin{aligned} \epsilon_{0}(\alpha,\Lambda) &\equiv\langle\psi_{0}(\alpha,\Lambda) \lvert H_{g}^{(Q)} \rvert\psi_{0}(\alpha,\Lambda)\rangle\\ &= \frac{ \int_{0}^{\infty} \mathrm{d}R\,R e^{-\frac{2\Lambda^{2}}{g^{2}}(R^{2}-1)^{2}-R^{2}}\left((R^{2}+1)I_{0}(2R\alpha)+2R(g^{4}\alpha-1)I_{1}(2R\alpha)\right)}{ 2g^{2}\int_{0}^{\infty}\mathrm{d}R\,R e^{-\frac{2\Lambda^{2}}{g^{2}}(R^{2}-1)^{2}-R^{2}}I_{0}(2R\alpha)} . \tag*{(80)} \end{aligned}

These integrals can be evaluated for large radial cutoff Λ\Lambda by expanding around R=1R=1, and using

∫0∞dR Re−2Λ2g2(R2−1)2(R2−1)N={π2(N−1)!!(2Λ/g)N+1+O(e−2Λ2/g2),N even,O(e−2Λ2/g2),N odd.(81)\int_{0}^{\infty} \mathrm{d}R\,R e^{-\frac{2\Lambda^{2}}{g^{2}}(R^{2}-1)^{2}}(R^{2}-1)^{N} = \begin{cases} \sqrt{\frac{\pi}{2}}\frac{(N-1)!!}{(2\Lambda/g)^{N+1}}+\mathcal{O}(e^{-2\Lambda^{2}/g^{2}}), & N\ \text{even},\\ \mathcal{O}(e^{-2\Lambda^{2}/g^{2}}), & N\ \text{odd}. \end{cases} \tag*{(81)}

We obtain the expansion

ϵ0(α,Λ)=1g2(1−I1(2α)I0(2α))+g2αI1(2α)I0(2α)+O(1Λ2).(82)\epsilon_{0}(\alpha,\Lambda)=\frac{1}{g^{2}}\left(1-\frac{I_{1}(2\alpha)}{I_{0}(2\alpha)}\right)+g^{2}\alpha\frac{I_{1}(2\alpha)}{I_{0}(2\alpha)}+\mathcal{O}\left(\frac{1}{\Lambda^{2}}\right). \tag*{(82)}

In the weak coupling limit, it is minimized for α≈12g2\alpha\approx\frac{1}{2g^{2}}. We deduce the estimate of the ground state energy

E0=1+18Λ2+O(g2,Λ−4).(83)E_{0}=1+\frac{1}{8\Lambda^{2}}+\mathcal{O}(g^{2},\Lambda^{-4}). \tag*{(83)}

Figure 7 shows results obtained by minimizing the energy of the ansatz over the variational parameter α\alpha for various values of the radial cutoff (Λ=1,2,3)(\Lambda=1,2,3) as a function of the coupling constant gg. Very good agreement with exact results is found even for small values of Λ\Lambda.

Figure 7. Ground state energy of a single-plaquette pure-gauge system vs. the coupling constant using a variational method for various values of the radial cutoff Λ\Lambda.

Ground state energy levels asymptote to E0→1E_0 \to1 in the weak coupling limit (g→0g \to0), as expected from the analytical result (83).

Similarly, the expectation value of the constraint (q2=1q^2 = 1) can be shown to vanish as Λ→∞\Lambda\to\infty,

⟨ψ0(α,Λ)|(q2−1)2|ψ0(α,Λ)⟩=∫0∞dR R (R2−1)2e−2Λ2g2(R2−1)2−R2I0(2Rα)∫0∞dR R e−2Λ2g2(R2−1)2−R2I0(2Rα)=g24Λ2+O(1Λ4)(84)\left\langle\psi_0(\alpha,\Lambda) \middle| (q^2-1)^2 \middle| \psi_0(\alpha,\Lambda) \right\rangle= \frac{\int_0^\infty dR\,R\,(R^2-1)^2 e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}I_0(2R\alpha)} {\int_0^\infty dR\,R\,e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}I_0(2R\alpha)} = \frac{g^2}{4\Lambda^2}+O\left(\frac{1}{\Lambda^4}\right) \tag*{(84)}

as desired. Using the ground state ansatz (78), we obtain the estimate of the vacuum expectation value of the plaquette operator,

⟨W[C]⟩=⟨ψ0(α,Λ)|q0|ψ0(α,Λ)⟩=∫0∞dR R2e−2Λ2g2(R2−1)2−R2I1(2Rα)∫0∞dR R e−2Λ2g2(R2−1)2−R2I0(2Rα)(85)\langle W[C]\rangle= \left\langle\psi_0(\alpha,\Lambda) \middle| q^0 \middle| \psi_0(\alpha,\Lambda) \right\rangle= \frac{\int_0^\infty dR\,R^2 e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}I_1(2R\alpha)} {\int_0^\infty dR\,R\,e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}I_0(2R\alpha)} \tag*{(85)}

Using Eq. (81), we obtain the expansion

⟨W[C]⟩=I1(2α)I0(2α)+g2α8Λ2(3(I1(2α)I0(2α))2−2)+O(1Λ4).(86)\langle W[C]\rangle= \frac{I_1(2\alpha)}{I_0(2\alpha)} +\frac{g^2\alpha}{8\Lambda^2}\left(3\left(\frac{I_1(2\alpha)}{I_0(2\alpha)}\right)^2-2\right) +O\left(\frac{1}{\Lambda^4}\right). \tag*{(86)}

In particular, in the weak coupling limit, we have α≈12g2\alpha\approx\frac{1}{2g^2}, therefore,

⟨W[C]⟩=1−g22+116Λ2(1−3g2)+O(g4,Λ−4).(87)\langle W[C]\rangle= 1-\frac{g^2}{2}+\frac{1}{16\Lambda^2}\left(1-3g^2\right)+O\left(g^4,\Lambda^{-4}\right). \tag*{(87)}

Similarly, using the ansatz for the first excited state,

∣ψ1(β,Λ)⟩=q1∣ψ0(β,Λ)⟩,(88)|\psi_1(\beta,\Lambda)\rangle= q^1|\psi_0(\beta,\Lambda)\rangle, \tag*{(88)}

we obtain the estimate of the first photonic energy level,

ϵ1≡⟨ψ(β,Λ)∣H∣ψ1(β,Λ)⟩⟨ψ1(β,Λ)∣ψ1(β,Λ)⟩=∫0∞dR R2e−2Λ2g2(R2−1)2−R2((2+β(1+R2−2g4))I1(2Rβ)−2β(1−3g4)RI0(2Rβ))2βg2∫0∞dR R2e−2Λ2g2(R2−1)2−R2I1(2Rβ)=1g2(1+1β−g4)−(1g2−3g2β)I0(2β)I1(2β)+O(1Λ2).(89)\begin{aligned} \epsilon_1 \equiv\frac{\langle\psi(\beta,\Lambda)|H|\psi_1(\beta,\Lambda)\rangle}{\langle\psi_1(\beta,\Lambda)|\psi_1(\beta,\Lambda)\rangle} \\ &= \frac{\int_0^\infty\mathrm{d}R\,R^2 e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}\left((2+\beta(1+R^2-2g^4))I_1(2R\beta)-2\beta(1-3g^4)R I_0(2R\beta)\right)}{2\beta g^2\int_0^\infty\mathrm{d}R\,R^2 e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2-R^2}I_1(2R\beta)} \\ &= \frac{1}{g^2}\left(1+\frac{1}{\beta}-g^4\right)-\left(\frac{1}{g^2}-3g^2\beta\right)\frac{I_0(2\beta)}{I_1(2\beta)}+\mathcal{O}\left(\frac{1}{\Lambda^2}\right). \tag*{(89)} \end{aligned}

In the weak coupling limit, this is minimized for β≈12g2\beta\approx\frac{1}{2g^2}. We deduce the estimate of the first photonic energy level and the gap,

E1=3+18Λ2+O(g2,Λ−4),ΔE=E1−E0=2+O(g2,Λ−4).(90)E_1 = 3+\frac{1}{8\Lambda^2}+\mathcal{O}(g^2,\Lambda^{-4}), \qquad\Delta E=E_1-E_0=2+\mathcal{O}(g^2,\Lambda^{-4}). \tag*{(90)}

We obtain very good agreement with exact results for both the vacuum expectation value of the Wilson loop ⟨W[C]⟩\langle W[C]\rangle and the first photonic energy level, as shown in Figure 8. In the weak coupling limit (g→0g\to0), the graphs asymptote to the values obtained by analytical calculations (W[C]→1W[C]\to1, in agreement with Eq. (87), and E1→3E_1\to3, in agreement with Eq. (90)).

Image not extracted

Figure 8. Vacuum expectation value of the plaquette operator (left) and first photonic energy level (right) of a single-plaquette pure-gauge system vs. the coupling constant using a variational method for various values of the radial cutoff Λ\Lambda.

Hamiltonian penalty method

As discussed in § 4, a second approach to enforcing the compactification constraint is by adding a penalty term to the Hamiltonian (Eq. (57)),

Hμ=μ2(q^02+q^12−1)2=μ2(R2−1)2(μ>0).(91)H_\mu= \frac{\mu}{2}\left(\hat{q}_0^2+\hat{q}_1^2-1\right)^2 = \frac{\mu}{2}\left(R^2-1\right)^2 \quad(\mu>0). \tag*{(91)}

Clearly, for sufficiently large μ\mu, the ground state energy corresponds to a configuration for which R=1R=1. In order to illustrate the mechanism, we compute the ground state by exactly diagonalizing the Hamiltonian. Since the (unphysical) radial coordinate RR is non-dynamical, it effectively behaves as a parameter of the theory. The case with μ=0\mu=0 corresponds to no constraint being imposed, and the energy corresponding to the physical configuration is not the one with minimal energy. As we increase μ\mu above a certain value, the ground state favors the physical choice R=1R=1. The threshold value of μ\mu for an acceptable penalty term (5.18) in the Hamiltonian of a single plaquette vs. the coupling constant is shown in Figure 9. We set the fermion mass m0=1.5m_0=1.5. Similar results are obtained for different choices of the fermion mass parameter. The threshold value was chosen for ground state energies within 1% of the exact ground state energy and radial coordinate within 5% of the physical value R=1R=1. Interestingly, for all values of the coupling constant, a choice of μ>1.6\mu>1.6 suffices. Moreover, in the weak coupling limit, the penalty term is not needed as the physical choice R=1R=1 is naturally favored.

Image not extracted

Figure 9. The threshold value of μ\mu for an acceptable penalty term (91) in the Hamiltonian of a single plaquette vs. the coupling constant. Fermion mass is set to m0=1.5m_0 = 1.5.

Transition probabilities

Next, we consider the time evolution of the complete single-plaquette system including fermions, and perform the quantum computation of transition probabilities using the quantum circuits introduced in § 4. As an example, we will work with the initial state

∣V⟩≡∣vvvv⟩⊗∣ψ0(α=0,Λ)⟩.(92)|V\rangle\equiv|vvvv\rangle\otimes|\psi_0(\alpha= 0,\Lambda)\rangle. \tag*{(92)}

Here, ∣vvvv⟩|vvvv\rangle is the ground state of free static fermions, Eq. (23), and ∣ψ0⟩|\psi_0\rangle is in the gauge sector and defined in Eq. (78). We compute the survival probability of the state ∣V⟩|V\rangle under time evolution with Hamiltonian H(Q)H^{(Q)}, Eq. (76), with its components given by Eq. (77). We Trotterize the evolution operator as

e−itH(Q)=(e−iΔtHE(Q)e−iΔtHB(Q)e−iΔtHM(Q)e−iΔtHK(Q))t/Δt.(93)e^{-itH^{(Q)}} = \left(e^{-i\Delta t H_E^{(Q)}} e^{-i\Delta t H_B^{(Q)}} e^{-i\Delta t H_M^{(Q)}} e^{-i\Delta t H_K^{(Q)}}\right)^{t/\Delta t}. \tag*{(93)}

The survival probability is obtained by applying the string of gates in (93) to the state ∣V⟩|V\rangle and then measuring the qubits and qumodes projecting back onto the state ∣V⟩|V\rangle. The quantum circuit implementing this is

Image not extracted

Figure 20 (PDF p. 29).

To simulate the survival amplitude, we use polar coordinates for the qumodes (q0=Rcos⁡χq^0 = R\cos\chi, q1=Rsin⁡χq^1 = R\sin\chi) and apply ED (exact diagonalization) in the gauge field basis ⟨χ∣j⟩=12πeijχ\langle\chi|j\rangle= \frac{1}{\sqrt{2\pi}}e^{ij\chi}, j∈Zj \in\mathbb{Z}, where J∣j⟩=j∣j⟩J|j\rangle= j|j\rangle, and the fermionic field basis ∣vi⟩|v_i\rangle (i=1,2,…,6i = 1, 2, \ldots, 6) defined in Eq. (23). We obtain the following expression for the complex amplitude

A(t)≡⟨V∣e−itH(Q)∣V⟩=∫0∞dR R e−R2e−2Λ2g2(R2−1)2⟨j=0∣⊗⟨vvvv∣e−itH(Q)∣vvvv⟩⊗∣j=0⟩∫0∞dR R e−R2e−2Λ2g2(R2−1)2.(94)\begin{aligned} \mathcal{A}(t) \equiv\langle V|e^{-itH^{(Q)}}|V\rangle\\ &= \frac{ \int_{0}^{\infty} \mathrm{d}R\,R\,e^{-R^2}e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2}\langle j=0|\otimes\langle vvvv|e^{-itH^{(Q)}}|vvvv\rangle\otimes|j=0\rangle}{ \int_{0}^{\infty} \mathrm{d}R\,R\,e^{-R^2}e^{-\frac{2\Lambda^2}{g^2}(R^2-1)^2}} . \tag*{(94)} \end{aligned}

The integral can be done by converting it to a Riemann sum and calculating each term in the sum numerically.

In Figure 10, we plot the vacuum survival probability at g−2=10g^{-2}=10 as a function of time, for various values of the radial cutoff Λ\Lambda. We obtain very good agreement with exact results for time t≲4t\lesssim4. Results for finite Λ\Lambda start to diverge from exact results for t≳4t\gtrsim4 where larger values of Λ\Lambda are needed to maintain accuracy.

Image not extracted

Figure 10. Survival probability of the state ∣V⟩|V\rangle, see Eq. (92), as a function of time for different values of the squeezing parameter Λ\Lambda compared with the exact results for a single plaquette.

Ground state using QITE

Another method to compute the ground state energy is by using QITE, as discussed in § 4. To implement the procedure for a single plaquette, it is convenient to start with the initial state given in Eq. (92). For each QITE step, we apply the non-unitary operator

e−ΔτH(Q)≈UQITE≡e−ΔτHE(Q)e−ΔτHB(Q)e−ΔτHM(Q)e−ΔτHK(Q),(95)e^{-\Delta\tau H^{(Q)}} \approx U_{\mathrm{QITE}} \equiv e^{-\Delta\tau H_E^{(Q)}}e^{-\Delta\tau H_B^{(Q)}}e^{-\Delta\tau H_M^{(Q)}}e^{-\Delta\tau H_K^{(Q)}} , \tag*{(95)}

see Eqs. (76) and (77), using the quantum circuits discussed in § 4. After SS QITE steps, we arrive at the estimate of the ground state

∣ψground⟩=(UQITE)S∣V⟩⟨V∣(UQITE)2S∣V⟩,(96)|\psi_{\mathrm{ground}}\rangle= \frac{(U_{\mathrm{QITE}})^S|V\rangle}{\sqrt{\langle V|(U_{\mathrm{QITE}})^{2S}|V\rangle}} , \tag*{(96)}

and the corresponding estimate of the ground state energy

ϵ0=⟨ψground∣H(Q)∣ψground⟩=⟨V∣H(Q)(UQITE)2S∣V⟩⟨V∣(UQITE)2S∣V⟩.(97)\epsilon_0 = \langle\psi_{\mathrm{ground}|H^{(Q)}|\psi_{\mathrm{ground}}\rangle} = \frac{\langle V|H^{(Q)}(U_{\mathrm{QITE}})^{2S}|V\rangle}{\langle V|(U_{\mathrm{QITE}})^{2S}|V\rangle} . \tag*{(97)}

Ideally, we would take Δτ→0\Delta\tau\to0 and the number of steps S→∞S \to\infty. However, due to finite resources, a realistic implementation would require a finite SS and Δτ\Delta\tau chosen carefully to balance the requirement that Δτ\Delta\tau be sufficiently small (due to Trotterization) and the total evolution time SΔτS\Delta\tau be large enough to allow for the decay of the system to a good approximation of the actual ground state.

In Figure 11, we plot the energy of a single plaquette undergoing imaginary time evolution for representative values of the coupling constant and radial cutoff Λ\Lambda, and fermion mass m0=1.5m_0 = 1.5. As expected, smaller imaginary time increments Δτ\Delta\tau require more QITE steps to converge. However, for Δτ=0.25\Delta\tau= 0.25, even after S=4S = 4, the energy is remarkably close to the ground state energy. The best results are obtained for Δτ=0.1\Delta\tau= 0.1 after S=10S = 10 QITE steps. They are shown in Figure 12 for various values of the coupling constant. We obtain good agreement with exact results.

Image not extracted

Figure 11. Energy of a single plaquette under QITE for different choices of the imaginary time increment Δτ\Delta\tau, coupling constant gg and radial cutoff Λ\Lambda. The fermion mass is set at m0=1.5m_0 = 1.5 in both cases. The dashed blue line is the result obtained from Exact Diagonalization (ED).

Image not extracted

Figure 12. Estimate of the ground state energy of a single plaquette vs. the coupling constant using QITE with S=10S = 10 steps and decay time increment Δτ=0.1\Delta\tau= 0.1. Fermion mass is set at m0=1.5m_0 = 1.5. Plots are shown for two different values of the radial cutoff Λ=1,5\Lambda= 1, 5 and ED.

Next, we compute the vacuum expectation value of the chiral condensate,1

OΨˉΨ=14∑i=14(−1)i+1ΨˉiΨi.(98)\mathcal{O}_{\bar{\Psi}\Psi} = \frac{1}{4}\sum_{i=1}^{4}(-1)^{i+1}\bar{\Psi}_i\Psi_i. \tag*{(98)}

Using QITE, we obtain the estimate for the vacuum expectation value,

⟨OΨˉΨ⟩=⟨ψground∣OΨˉΨ∣ψground⟩=⟨V∣(UQITE)SOΨˉΨ(UQITE)S∣V⟩⟨V∣(UQITE)2S∣V⟩(99)\langle\mathcal{O}_{\bar{\Psi}\Psi}\rangle= \langle\psi_{\mathrm{ground}}|\mathcal{O}_{\bar{\Psi}\Psi}|\psi_{\mathrm{ground}}\rangle= \frac{\langle V|(U_{\mathrm{QITE}})^S\mathcal{O}_{\bar{\Psi}\Psi}(U_{\mathrm{QITE}})^S|V\rangle}{\langle V|(U_{\mathrm{QITE}})^{2S}|V\rangle} \tag*{(99)}

Our results are shown in Figure 13. We used QITE with S=10S = 10 steps, decay time increment Δτ=0.1\Delta\tau= 0.1, radial cutoff Λ=1\Lambda= 1, and fermion mass m0=1.5m_0 = 1.5. We obtained good agreement with ED. We In the weak coupling limit, we recover the free fermion chiral condensate, which is given by the analytic expression

Image not extracted

Figure 13. Vacuum expectation value of the chiral condensate for a single plaquette vs. the coupling constant using QITE with S=10S = 10 steps, decay time increment Δτ=0.1\Delta\tau= 0.1, radial cutoff Λ=1\Lambda= 1, and fermion mass m0=1.5m_0 = 1.5, compared with ED and free fermions (Eq. (100)).

⟨OΨˉΨ⟩free fermion ground state=−14(1+m01+m02).(100)\langle\mathcal{O}_{\bar{\Psi}\Psi}\rangle_{\mathrm{free\ fermion\ ground\ state}} = -\frac{1}{4}\left(1+\frac{m_0}{\sqrt{1+m_0^2}}\right). \tag*{(100)}

In the strong coupling limit (g−2≪1g^{-2} \ll1), the electric contribution dominates, and the fermions are effectively static charges. Therefore, we expect the chiral condensate to approximate the expression ((5.27) in the limit $m_0 \to \infty$, i.e., $\langle O_{\bar\psi\psi}\rangle \to -\frac{1}{2}$. This is in agreement with our results depicted in Figure 13`.

Conclusion and outlook

We have formulated and analyzed a hybrid qubit–qumode framework for simulating U(1)U(1) lattice gauge theory coupled to fermionic matter in 2+12+1 dimensions, providing an explicit link between the field-theoretic Hamiltonian and experimentally accessible quantum-hardware operations. By encoding fermionic matter fields in discrete variables (qubits) and gauge fields in continuous bosonic modes (qumodes), the approach captures the full structure of the theory without introducing an artificial truncation of the gauge sector. Two compactness-enforcement strategies were developed: a squeezing-based projection that restricts qumode amplitudes to the compact manifold, and a Hamiltonian penalty term that confines the continuous variables dynamically. Both methods reproduce the correct gauge-invariant spectrum and are compatible with currently available hybrid platforms.

The electric, magnetic, and matter–gauge interaction terms of the lattice Hamiltonian were decomposed into a universal set of qubit–qumode gates composed of Gaussian, non-Gaussian, and conditional displacement operations. This establishes a concrete and hardware-realizable path for simulating Abelian gauge dynamics on hybrid devices such as trapped ions and superconducting circuits. Scaling estimates indicate polynomial growth of resources with lattice size, suggesting that few-plaquette systems lie within the reach of near-term technology. Using a continuous-variable extension of the QITE algorithm, we demonstrated that hybrid imaginary-time dynamics can efficiently prepare ground states and preserve gauge invariance during evolution.

The hybrid architecture presented here merges the strengths of discrete- and continuous-variable computations and opens a path toward scalable simulations of quantum field theories. Future work will extend the present formulation to non-Abelian gauge groups, explore hybrid variational and error-mitigation strategies, and test the methods on emerging hybrid quantum processors. The combination of theoretical rigor and experimental feasibility makes hybrid qubit–qumode simulation a promising paradigm for realizing quantum field dynamics in the laboratory. Future work will also extend this hybrid framework to non-Abelian gauge groups and explore real-time dynamics on fault-tolerant architectures.

Acknowledgments

We would like to thank Matt Grau for helpful discussions. We acknowledge support by DOE ASCR funding under the Quantum Computing Application Teams Program, NSF award DGE-2152168, and DOE Office of Nuclear Physics, Quantum Horizons Program, award DE-SC0023687. The research was supported the DOE, Office of Science, Office of Nuclear Physics, Early Career Program under contract No. DE-SC0024358 and DE-SC0025881. The authors would like to thank Stony Brook Research Computing and Cyberinfrastructure, and the Institute for Advanced Computational Science at Stony Brook University for access to the SeaWulf computing system, made possible by grants from the National Science Foundation (#1531492 and Major Research Instrumentation award #2215987), with matching funds from Empire State Development’s Division of Science, Technology and Innovation (NYSTAR) program (contract C210148).

References

  1. [1]M. C. Banuls, R. Blatt, J. Catani, A. Celi, J. I. Cirac, M. Dalmonte, L. Fallani, et al., “Simulating lattice gauge theories within quantum technologies,” The European physical journal D 74 no. 8, (2020) 165.
  2. [2]Y. Alexeev, D. Bacon, K. R. Brown, R. Calderbank, L. D. Carr, F. T. Chong, B. DeMarco, et al., “Quantum computer systems for scientific discovery,” PRX Quantum 2 (Feb, 2021) 017001. https://link.aps.org/doi/10.1103/PRXQuantum.2.017001.DOI
  3. [3]C. W. Bauer, Z. Davoudi, A. B. Balantekin, T. Bhattacharya, M. Carena, W. A. de Jong, P. Draper, et al., “Quantum simulation for high-energy physics,” PRX Quantum 4 (May, 2023) 027001. https://link.aps.org/doi/10.1103/PRXQuantum.4.027001.DOI
  4. [4]A. Di Meglio, K. Jansen, I. Tavernelli, C. Alexandrou, S. Arunachalam, C. W. Bauer, K. Borras, et al., “Quantum computing for high-energy physics: State of the art and challenges,” PRX Quantum 5 (Aug, 2024) 037001. https://link.aps.org/doi/10.1103/PRXQuantum.5.037001.DOI
  5. [5]S. P. Jordan, K. S. M. Lee, and J. Preskill, “Quantum Algorithms for Quantum Field Theories,” Science 336 (2012) 1130–1133, arXiv:1111.3633 [quant-ph].DOI
  6. [6]J. B. Kogut and L. Susskind, “Hamiltonian Formulation of Wilson’s Lattice Gauge Theories,” Phys. Rev. D 11 (1975) 395–408.DOI
  7. [7]P. Jordan and E. P. Wigner, “About the Pauli exclusion principle,” Z. Phys. 47 (1928) 631–651.
  8. [8]S. B. Bravyi and A. Y. Kitaev, “Fermionic Quantum Computation,” Annals Phys. 298 (2002) 210–226, arXiv:quant-ph/0003137.DOI
  9. [9]S. Chandrasekharan and U. J. Wiese, “Quantum link models: A Discrete approach to gauge theories,” Nucl. Phys. B 492 (1997) 455–474, arXiv:hep-lat/9609042.DOI
  10. [10]E. Zohar, J. I. Cirac, and B. Reznik, “Quantum Simulations of Lattice Gauge Theories using Ultracold Atoms in Optical Lattices,” Rept. Prog. Phys. 79 no. 1, (2016) 014401, arXiv:1503.02312 [quant-ph].DOI
  11. [11]E. A. Martinez, C. A. Muschik, P. Schindler, D. Nigg, A. Erhard, M. Heyl, P. Hauke, M. Dalmonte, T. Monz, P. Zoller, and R. Blatt, “Real-time dynamics of lattice gauge theories with a few-qubit quantum computer,” Nature 534 no. 7608, (Jun, 2016) 516–519. https://doi.org/10.1038/nature18318.
  12. [12]N. Klco, E. F. Dumitrescu, A. J. McCaskey, T. D. Morris, R. C. Pooser, M. Sanz, E. Solano, P. Lougovski, and M. J. Savage, “Quantum-classical computation of schwinger model dynamics using quantum computers,” Phys. Rev. A 98 (Sep, 2018) 032331. https://link.aps.org/doi/10.1103/PhysRevA.98.032331.DOI
  13. [13]C. Kokail, C. Maier, R. van Bijnen, T. Brydges, M. K. Joshi, P. Jurcevic, C. A. Muschik, P. Silvi, R. Blatt, C. F. Roos, and P. Zoller, “Self-verifying variational quantum simulation of lattice models,” Nature 569 no. 7756, (May, 2019) 355–360. https://doi.org/10.1038/s41586-019-1177-4.
  14. [14]A. Alexandru, P. F. Bedaque, S. Harmalkar, H. Lamm, S. Lawrence, and N. C. Warrington, “Gluon Field Digitization for Quantum Computers,” Phys. Rev. D 100 no. 11, (2019) 114501, arXiv:1906.11213 [hep-lat].DOI
  15. [15]J. F. Haase, L. Dellantonio, A. Celi, D. Paulson, A. Kan, K. Jansen, and C. A. Muschik, “A resource efficient approach for quantum and classical simulations of gauge theories in particle physics,” Quantum 5 (2021) 393, arXiv:2006.14160 [quant-ph].DOI
  16. [16]C. W. Bauer and D. M. Grabowska, “Efficient representation for simulating U(1) gauge theories on digital quantum computers at all values of the coupling,” Phys. Rev. D 107 no. 3, (2023) L031503, arXiv:2111.08015 [hep-ph].arxiv.org/abs/2111.08015
  17. [17]A. Ciavarella, N. Klco, and M. J. Savage, “Trailhead for quantum simulation of SU(3) Yang-Mills lattice gauge theory in the local multiplet basis,” Phys. Rev. D 103 no. 9, (2021) 094501, arXiv:2101.10227 [quant-ph].DOI
  18. [18]Z. Davoudi, A. F. Shaw, and J. R. Stryker, “General quantum algorithms for Hamiltonian simulation with applications to a non-Abelian lattice gauge theory,” Quantum 7 (2023) 1213, arXiv:2212.14030 [hep-lat].DOI
  19. [19]S. Grieninger, D. E. Kharzeev, and I. Zahed, “Entanglement in a holographic Schwinger pair with confinement,” Phys. Rev. D 108 no. 8, (2023) 086030, arXiv:2305.07121 [hep-th].DOI
  20. [20]J. C. Halimeh, M. Aidelsburger, F. Grusdt, P. Hauke, and B. Yang, “Cold-atom quantum simulators of gauge theories,” Nature Phys. 21 no. 1, (2025) 25–36, arXiv:2310.12201 [cond-mat.quant-gas].arxiv.org/abs/2310.12201
  21. [21]S. Lloyd and S. L. Braunstein, “Quantum computation over continuous variables,” Phys. Rev. Lett. 82 (1999) 1784–1787, arXiv:quant-ph/9810082.DOI
  22. [22]K. Marshall, R. Pooser, G. Siopsis, and C. Weedbrook, “Quantum simulation of quantum field theory using continuous variables,” Phys. Rev. A 92 no. 6, (2015) 063825, arXiv:1503.08121 [quant-ph].DOI
  23. [23]R. Dasgupta and I. Raychowdhury, “Cold-atom quantum simulator for string and hadron dynamics in non-abelian lattice gauge theory,” Phys. Rev. A 105 (Feb, 2022) 023322. https://link.aps.org/doi/10.1103/PhysRevA.105.023322.DOI
  24. [24]C. Muschik, M. Heyl, E. Martinez, T. Monz, P. Schindler, B. Vogell, M. Dalmonte, P. Hauke, R. Blatt, and P. Zoller, “(U(1) wilson lattice gauge theories in digital quantum simulators,” New Journal of Physics 19 no. 10, (Oct, 2017) 103020. https://doi.org/10.1088/1367-2630/aa89ab.
  25. [25]R. G. Jha, F. Ringer, G. Siopsis, and S. Thompson, “Continuous-variable quantum computation of the O(3) model in 1+1 dimensions,” Phys. Rev. A 109 no. 5, (2024) 052412, arXiv:2310.12512 [quant-ph].DOI
  26. [26]R. A. Briceño, R. G. Edwards, M. Eaton, C. González-Arciniegas, O. Pfister, and G. Siopsis, “Toward coherent quantum computation of scattering amplitudes with a measurement-based photonic quantum processor,” Phys. Rev. Res. 6 no. 4, (2024) 043065, arXiv:2312.12613 [quant-ph].arxiv.org/abs/2312.12613
  27. [27]S. Abel, M. Spannowsky, and S. Williams, “Simulating quantum field theories on continuous-variable quantum computers,” Phys. Rev. A 110 no. 1, (2024) 012607, arXiv:2403.10619 [quant-ph].DOI
  28. [28]S. Abel, M. Spannowsky, and S. Williams, “Qumode Tensor Networks for False Vacuum Decay in Quantum Field Theory,” arXiv:2506.17388 [quant-ph].arxiv.org/abs/2506.17388
  29. [29]S. Abel, M. Spannowsky, and S. Williams, “Real-time scattering processes with continuous-variable quantum computers,” Phys. Rev. A 112 no. 1, (2025) 012614, arXiv:2502.01767 [quant-ph].DOI
  30. [30]N. Gupta, C. D. White, and Z. Davoudi, “Euclidean-Monte-Carlo-informed ground-state preparation for quantum simulation of scalar field theory,” arXiv:2506.02313 [quant-ph].DOI
  31. [31]V. Ale, N. M. Bauer, R. G. Jha, F. Ringer, and G. Siopsis, “Quantum computation of su(2) lattice gauge theory with continuous variables,” Journal of High Energy Physics 2025 no. 6, (Jun, 2025) 84. https://doi.org/10.1007/JHEP06(2025)084.
  32. [32]E. Crane et al., “Hybrid oscillator-qubit quantum processors: Simulating fermions, bosons, and gauge fields,” arXiv:2409.03747 [quant-ph].DOI
  33. [33]M. Miranda-Riaza, P. Fontana, and A. Celi, “Renormalized dual basis for scalable simulations of 2+ 1d compact quantum electrodynamics,” arXiv preprint arXiv:2510.18594 (2025) .
  34. [34]P. Fontana, M. Miranda-Riaza, and A. Celi, “Efficient finite-resource formulation of non-abelian lattice gauge theories beyond one dimension,” Phys. Rev. X 15 (Sep, 2025) 031065. https://link.aps.org/doi/10.1103/k9p6-c649.
  35. [35]S. Saner et al., “Real-time observation of aharonov-bohm interference in a z2 lattice gauge theory on a hybrid qubit-oscillator quantum computer,” arXiv preprint arXiv:2507.19588 (2025) .DOI
  36. [36]S. Varona, S. Saner, O. Băzăvan, G. Araneda, G. Aarts, and A. Bermudez, “Towards quantum computing feynman diagrams in hybrid qubit-oscillator devices,” arXiv preprint arXiv:2411.05092 (2024) .arxiv.org/abs/2411.05092
  37. [37]E. C. Domanti, A. Bermudez, and L. Amico, “Dynamical aharonov-bohm cages and tight meson confinement in a z 2-loop gauge theory,” Communications Physics 8 no. 1, (2025) 357.arxiv.org/abs/2412.12425
  38. [38]O. Băzăvan, S. Saner, E. Tirrito, G. Araneda, R. Srinivas, and A. Bermudez, “Synthetic z 2 gauge theories based on parametric excitations of trapped ions,” Communications Physics 7 no. 1, (2024) 229.DOI
  39. [39]A. Schuckert, S. Kühn, K. C. Smith, E. Crane, and S. M. Girvin, “Constrained many-body phases in a Z₂-Higgs lattice gauge theory,” arXiv:2503.03828 [cond-mat.quant-gas].arxiv.org/abs/2503.03828
  40. [40]A. T. Than, S. V. Kadam, V. Vikramaditya, N. H. Nguyen, X. Liu, Z. Davoudi, A. M. Green, and N. M. Linke, “Observation of quantum-field-theory dynamics on a spin-phonon quantum computer,” arXiv:2509.11477 [quant-ph].arxiv.org/abs/2509.11477
  41. [41]T. J. Stavenger, E. Crane, K. C. Smith, C. T. Kang, S. M. Girvin, and N. Wiebe, “C2QA - Bosonic Qiskit,” in 26th IEEE High Performance Extreme Computing. 9, 2022. arXiv:2209.11153 [quant-ph].
  42. [42]Y. Liu et al., “Hybrid Oscillator-Qubit Quantum Processors: Instruction Set Architectures, Abstract Machine Models, and Applications,” arXiv:2407.10381 [quant-ph].DOI
  43. [43]J. Y. Araz, M. Grau, J. Montgomery, and F. Ringer, “Hybrid quantum simulations with qubits and qumodes on trapped-ion platforms,” Phys. Rev. A 112 no. 1, (2025) 012620, arXiv:2410.07346 [quant-ph].DOI
  44. [44]J. H. Busnaina, Z. Shi, J. M. Alcaine-Cuervo, C. X. Yang, I. Nsanzineza, E. Rico, and C. M. Wilson, “Native three-body interactions in a superconducting lattice gauge quantum simulator,” Phys. Rev. B 112 (Oct, 2025) 134514. https://link.aps.org/doi/10.1103/6prx-zmdz.DOI
  45. [45]D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, “Quantum dynamics of single trapped ions,” Rev. Mod. Phys. 75 (Mar, 2003) 281–324. https://link.aps.org/doi/10.1103/RevModPhys.75.281.DOI
  46. [46]D. J. Wineland, C. Monroe, W. M. Itano, D. Leibfried, B. E. King, and D. M. Meekhof, “Experimental issues in coherent quantum-state manipulation of trapped atomic ions,” 1998. https://arxiv.org/abs/quant-ph/9710025.
  47. [47]A. Blais, A. L. Grimsmo, S. M. Girvin, and A. Wallraff, “Circuit quantum electrodynamics,” Rev. Mod. Phys. 93 no. 2, (2021) 025005, arXiv:2005.12667 [quant-ph].DOI
  48. [48]M. Motta, C. Sun, A. T. K. Tan, M. J. O’Rourke, E. Ye, A. J. Minnich, F. G. S. L. Brandão, and G. K.-L. Chan, “Determining eigenstates and thermal states on a quantum computer using quantum imaginary time evolution,” Nature Physics 16 no. 2, (Feb, 2020) 205–210. https://doi.org/10.1038/s41567-019-0704-4.
  49. [62]dispersive coupling to a qubit,” Nature Phys. 18 no. 12, (2022) 1464–1469, arXiv:2111.06414 [quant-ph].DOI
  50. [63]M. Hatridge et al., “Quantum Back-Action of an Individual Variable-Strength Measurement,” Science 339 no. 6116, (2013) 1226897, arXiv:1903.11732 [quant-ph].
  51. [64]A. P. Higginbotham, P. S. Burns, M. D. Urmey, R. W. Peterson, N. S. Kampel, B. M. Brubaker, G. Smith, K. W. Lehnert, and C. A. Regal, “Harnessing electro-optic correlations in an efficient mechanical converter,” Nature Phys. 14 no. 10, (2018) 1038–1042, arXiv:1712.06535 [quant-ph].DOI
  52. [65]C. Trallero-Herrero and T. C. Weinacht, “Transition from weak- to strong-field coherent control,” Phys. Rev. A 75 (Jun, 2007) 063401. https://link.aps.org/doi/10.1103/PhysRevA.75.063401.
  53. [66]J. E. Drut and T. A. Lähde, “Lattice field theory simulations of graphene,” Phys. Rev. B 79 (Apr, 2009) 165425. https://link.aps.org/doi/10.1103/PhysRevB.79.165425.DOI
  54. [67]Y. Araki, “Chiral symmetry breaking in monolayer graphene by strong coupling expansion of compact and non-compact u(1) lattice gauge theories,” Annals of Physics 326 no. 6, (2011) 1408–1424. https://www.sciencedirect.com/science/article/pii/S0003491611000108.

Paper details

vlnc:W_dz74xhnh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents