UMD-PP-019-03

Towards analog quantum simulations of lattice gauge theories with trapped ions

Zohreh Davoudi,1, 2, ∗ Mohammad Hafezi,3, 4 Christopher Monroe,3, 5 Guido Pagano,3, 5 Alireza Seif,3 and Andrew Shaw1 1Maryland Center for Fundamental Physics and Department of Physics, University of Maryland, College Park, MD 20742, USA 2RIKEN Center for Accelerator-based Sciences, Wako 351-0198, Japan 3Joint Quantum Institute and Department of Physics, University of Maryland, College Park, MD 20742 4Department of Electrical and Computer Engineering and Institute for Research in Electronics and Applied Physics, University of Maryland, College Park, MD 20742 5Joint Center for Quantum Information and Computer Science, University of Maryland, College Park, MD 20742, USA (Dated: August 12, 2019) Gauge field theories play a central role in modern physics and are at the heart of the Standard Model of elementary particles and interactions. Despite significant progress in applying classical computational techniques to simulate gauge theories, it has remained a challenging task to compute the real-time dynamics of systems described by gauge theories. An exciting possibility that has been explored in recent years is the use of highly-controlled quantum systems to simulate, in an analog fashion, properties of a target system whose dynamics are difficult to compute. Engineered atom- laser interactions in a linear crystal of trapped ions offer a wide range of possibilities for quantum simulations of complex physical systems. Here, we devise practical proposals for analog simula- tion of simple lattice gauge theories whose dynamics can be mapped onto spin-spin interactions in any dimension. These include 1+1D quantum electrodynamics, 2+1D Abelian Chern-Simons theory coupled to fermions, and 2+1D pure Z2 gauge theory. The scheme proposed, along with the optimization protocol applied, will have applications beyond the examples presented in this work, and will enable scalable analog quantum simulation of Heisenberg spin models in any number of dimensions and with arbitrary interaction strengths.

I. INTRODUCTION quantum simulation and quantum computation [3–5], to tackle these problems. The invariance of physical systems under local transfor- While the idea of simulating a quantum system using mations of fields leads to fundamental constraints on how another quantum system with a higher level of control matter fields interact, and introduces new bosonic de- dates back to Feynman [6], only the experimental ad- grees of freedom, the gauge fields. Gauge field theories vancements in recent years have enabled powerful and coupled to matter are responsible for a wide range of sizable quantum simulations to become a reality. As in phenomena in nature, and permeate condensed matter, the case of classical computations, digital computations nuclear, and particle physics. In the case of gauge the- on quantum platforms may be the ultimate solution to ories comprising the Standard Model (SM) of particle all computational problems, including quantum simula- physics, progress in perturbative tools has enabled pre- tions of physical systems. However, in the era of noisy dictions for high-energy experiments at the Large Hadron intermediate-scale quantum (NISQ) computing [7], the Collider [1]. Furthermore, progress in non-perturbative number of high-fidelity operations that can be performed tools has led to theoretical input for precision experi- on a device can be highly constrained by the short co- ments in search of violations of fundamental symmetries herence time of the quantum state. As a result, the digi- arXiv:1908.03210v1 [quant-ph] 8 Aug 2019 in nature, and to predicting hadronic excitations and talization of complex dynamics, such as those associated their internal structure [2]. Nonetheless, the computa- with gauge field theories, can be limited to small system tional complexity of such studies grows significantly with sizes and short evolution times. It is therefore important the system size. In the strong-coupling regime, in which to seek alternative approaches in the NISQ era. An in- non-perturbative Monte Carlo sampling of quantum vac- teresting possibility is offered by analog simulations, in uum configurations is a common practice, questions such which the native Hamiltonian of the controlled quantum as the nature of the phase diagram of finite-density sys- system is engineered to be mapped to that of the tar- tems and the real-time dynamics of matter remain unan- get system. The quantum operations are then naturally swered. It is therefore essential to explore a broader set implemented once the system is prepared to evolve ac- of computational approaches, including those based on cording to the desired Hamiltonian. Among the most compelling platforms for analog sim- ulations of quantum systems, including those governed by gauge theories, are cold neutral atoms in optical ∗ The authors’ list is alphabetically ordered. lattices [8–15], optical tweezers [16, 17], and trapped 2 ions [18, 19]. Simple low-dimensional field theories to be achieved in the presence of noise. While fully ana- such as relativistic Dirac fermions, 1+1D1 and 2+1D log proposals exist for simulating simple low-dimensional scalar and fermionic quantum electrodynamics (QED), LGTs [76], none have been implemented so far due to and non-Abelian SU(2) and SO(3) gauge theories have technical limitations. been studied in this context, and proposals exist to It is important to classify gauge field theories of inter- map the desired lattice Hamiltonians (or their approx- est in terms of whether analog simulation of their dynam- imated forms) to that of the engineered Hamiltonian ics is feasible given current technology. It is also essen- of neutral atoms in optical lattices [20–41]. Recent tial to investigate whether fully analog implementations implementations of simple static and dynamical gauge can circumvent the accumulated noise due to digitaliza- theories with neutral atoms in optical lattices [42–47], tion [77], and whether the noise in an analog setup can be however, demonstrate the challenge of simulating more effectively mitigated. Finally, it would be beneficial to as- phenomenologically-relevant gauge theories. Given the sess the practicality of existing ideas, and to develop new current size of controlled quantum systems, only a small proposals for extending the quantum toolkit of trapped number of degrees of freedom can be studied, leading to ions, to enable a one-to-one mapping between the engi- unavoidable truncations in the Hilbert space of a gauge neered Hamiltonian of the ion-laser quantum system to theory that lives in a continuous infinite-volume space- the dynamics of a fermionic system coupled to gauge de- time. Such a limitation is present in other digital and grees of freedom (bosons). This paper is a first step in analog quantum platforms as well. It is nonetheless im- addressing these questions. Here, we focus on identifying portant that theoretical developments in formulating and goals that can be achieved in the near term, by specify- mapping gauge theories for a quantum simulation pro- ing, in detail, practical proposals for a range of accessible ceed alongside the continual experimental progress that gauge theories. aims to significantly improve capabilities and capacities The gauge field theories studied in this paper are exam- of simulating platforms. ples of the theories whose discretized formulations can be Trapped ions provide a pristine platform for quantum 1 mapped entirely to systems with spin- 2 degrees of free- simulations [18]. Given the extremely high level of con- dom. These examples include: i) 1+1D quantum electro- trol enabled by laser-cooled and localized ions confined dynamics (Schwinger model). This model has similari- by electromagnetic fields, exceedingly high fidelities in ties to quantum chromodynamics in 3+1D, including ex- state preparation and measurement, all-to-all entangling hibiting a non-trivial vacuum. ii) 2+1D Abelian Chern- capability enabled through control over the excitations Simons theory coupled to matter fields. This model is an of the motional normal modes, and scalability potential example of a topological gauge theory with applications of such systems, this architecture has become a primary in many areas of physics. iii) 2+1D Z2 gauge theory candidate for digital quantum computations in recent with a non-trivial phase diagram on a lattice, including years [48–62]. A unique feature of the trapped ion ar- exhibiting confinement. We discuss the mapping of these chitecture is that global addressing of the ions using a theories to spin systems, and present experimental pro- few laser beams allows the realization of tunable long- tocols for realizing these interactions in current and near- range spin-spin interactions in the chain. With no need term ion-trap systems. In order to provide a reference for for individual addressability, systems of a few tens of upcoming implementations in the case of the Schwinger ions have been successfully realized, and analog simu- model, detailed examples for 4 and 8 fermion-site theories lations of sizable quantum spin systems are made possi- will be presented. ble [63–71]. More complex quantum many-body systems, such as those described by gauge field theories, require A linear chain of trapped ions is often viewed as a platform for simulating spin- 1 systems in 1+1D. How- either some degree of individual addressing or higher- 2 order spin interactions among different species, as put ever, once such a system is augmented with individual forward in several proposals for simulating the relativis- addressing, it offers far more possibilities for quantum tic Dirac equation [5, 72–74] a quantum field theory of simulations of arbitrary spin systems, including those in scalar fields [75], and 1+1D QED [56, 76]. A milestone in higher dimensions. Such proposals have been put forth quantum simulations of lattice gauge theories (LGTs) us- in Ref. [78], and are explicitly taken advantage of in ing trapped ions was achieved in Ref. [56], where the real- the current work to simulate the dynamics of the LGTs time dynamics of 1+1D QED in a system of four trapped mentioned above. We also demonstrate the accessibil- ions in a linear trap was made possible through a digi- ity of nearly-perfect nearest-neighbor interactions by sim- tal protocol, but the number of operations required for ply controlling the lasers’ phases and intensities on each a Trotterized procedure prevented a long evolution time ion, and demonstrate the sensitivity of the evolution to the imperfections of the engineered Hamiltonian in the case of the Schwinger model. By controlling intensities, phases and frequencies of laser beams addressing each 1 Here and in the following, the first number denotes the space ion, a highly accurate mapping to spin-spin Hamiltoni- dimension, and the second number refers to the time dimension. ans with arbitrary interaction profiles is enabled. An When there is only one number, it is meant to refer to the space important feature of the protocols devised in this work dimension (or the spacetime dimension with a Euclidean metric). is a thorough optimization procedure that maximizes 3 the closeness to the desired Hamiltonian, while simul- tional clarity, further details of the proposed scheme and taneously minimizes errors stemming from residual cou- a number of involved analytical forms will be presented plings to motional excitations. The proposed experimen- in the appendices. tal scheme will have applications beyond the examples The ion-laser interaction Hamiltonian for a system of discussed, and is a general protocol for realizing interest- N trapped ions can be written as [65] ing spin systems described by a Heisenberg Hamiltonian N nL in arbitrary dimensions. (i) −i∆ω(i)t+i∆ϕ(i)+i∆k(i)·∆r(i) H = Ω e L L L The paper is organized as follows. Sec.II includes de- int L i=1 tails of ion-laser Hamiltonian considered in the scheme X LX=1 (i) (i) (i) (i) of this work, and presents the effective Hamiltonian ob- (α0I + α1σx + α2σy + α3σz ) + h.c., (1) × tained, its range of validity, and the associated undesired Index L in Eq. (1) runs over nL pairs of Raman beams. contributions that must be minimized subsequently. The (i) two associated appendicesA andB offer details on a par- ΩL is the Rabi frequency associated with the laser L. (i) ticular experimental platform, and a scheme that elim- ∆ϕL denotes the phase difference between the two lasers inates an unwanted bias term in engineering the effec- (i) in each pair of Raman beams, ∆ωL is the difference in tive Hamiltonian. The full evolution operator is further their angular frequency, namely the beatnote frequency, detailed in AppendixC. Sec.III presents the example (i) and ∆kL is the difference in their momentum k-vector. of the lattice Schwinger model, its purely spin represen- In general, each ion is addressed with multiple pairs of tation, and explicit experimental proposals for simulat- Raman beams individually (hence the superscript (i) on ing 4 and 8 fermion-site theories. The former case is quantities), requiring both amplitude and frequency con- implemented with a single detuning for each set of the trol of the beams. Such individual addressing of the lasers used, while the latter takes advantage of a multi- ions is widely used in digital ion-trap platforms, and can frequency, multi-amplitude scheme, requiring a thorough be ported to analog platforms in upcoming experiments. optimization of interaction couplings. Additional results ∆r(i) denotes the displacement vector of ion i from its on the 8 fermion-site theory are presented in AppendixD. equilibrium position. The Pauli matrices σ(i) act on the The results of the numerical evaluation of the full evo- quasi-spin of ion i, and α0, α1, α2, and α3 are constants lution operator up to the order considered are presented related to the spin-dependent forces on the two states of in another associated Appendix (AppendixE) as well as the qubit [65] and are controlled by the intensity, geome- in Supplemental Material. Sec.IV presents examples of try and polarization of the laser beams, see AppendixB LGTs in higher dimensions and their dual spin repre- for further details. sentation, along with discussions on their amenability to We assume that the confining potential is sufficiently the quantum simulation scheme of this work. We con- stronger along the transverse axes of the trap so that clude in Sec.V by highlighting the differing features of the ions form a 1D crystal in space. With appropri- the scheme presented here compared with the previous ate anharmonic axial confinement forces, the ions can work, the significance of the results obtained, and future be nearly equally-spaced [83, 84], with a typical spac- extensions that may enable addressing a wider class of ing between adjacent ions of a few micrometers. Due to gauge theories. the long-range Coulomb force among the ions and the common trapping potential applied, the motion of the ions can be described in terms of a set collective normal II. 1D CHAIN OF TRAPPED IONS AND modes. Then, ∆r(i) in Eq. (1) can be expressed in terms ENGINEERED EFFECTIVE INTERACTIONS of phononic degrees of freedom, whose excitation energies are quantized in units of the normal-mode frequencies of Consider N ions confined in a radio-frequency Paul the system. For the Hamiltonians of gauge theories con- trap [79]. The “qubit” in this system can be encoded sidered in this work, it is necessary to introduce multiple in two stable internal levels of the ion, denoted in the pairs of bichromatic Raman beams directed at each ion, following as and . These states are separated in en- such that each pair couples to only one set of the three |↑i |↓i ergy by an angular frequency ω0 (with Planck’s constant independent sets of normal modes. Such a scheme can ~ = 1 here and in the rest of the paper). Coherent op- be achieved with N individual beams and three global erations on spin degrees of freedom are realized through beams. Each of the individual beams will have three fre- stimulated Raman transitions using two laser beams with quencies2 that are tuned sufficiently apart such that each a momentum-vector difference ∆k. The physics of ion- frequency will drive the qubit only by pairing with one of laser interactions and the single and two-qubit manipu- the global beams. This setup will allow to tune indepen- lations in an ion trap is well known [19, 63, 65, 80–82]. dently Hamiltonians acting along orthogonal directions However, the involved evolution of the system under mul- tiple pairs of Raman beams, which are needed for engi- neering the Hamiltonians of models considered here, re- quires a few technical novelties, and warrants a dedicated 2 Or three sets of frequencies as required by the multi-frequency discussion which will follow in this section. For presenta- scheme of Sec.III. 4

FIG. 1. A schematic representation of a Raman-beams configuration that induces effective spin-spin interactions in the Heisen- berg model. The N sets of individual beams can be chosen along the (ξX,ˆ ξYˆ , χZˆ) unit vector (2ξ2 +χ2 = 1). Global beams (I), (II), and (III) are then chosen to propagate along (−ξX,ˆ ξYˆ , χZˆ), (ξX,ˆ ξY,ˆ −χZˆ), and (ξX,ˆ −ξYˆ , χZˆ), respectively. These will cause net ∆k vectors compared with the individual beams along the Xˆ, Zˆ, and Yˆ directions, respectively. Chosen values of these parameters for the examples of this work are given in AppendixA.

of the Bloch sphere with negligible undesired cross cou- i.e., ∆kII = ∆kII Zˆ and the transverse normal modes 5 plings as shown below. The chosen directionality of the along the Y direction, i.e., ∆kIII = ∆kIII Yˆ . The cor- A beams can ensure that each global-individual pair will responding normal mode frequencies are denoted as ωm T result in a net k-vector along one of the three orthogo- and ωm. Different superscripts are introduced to dis- nal principal axes of the trap, X,Y and Z, see Fig.1. 3 tinguish the transverse and axial normal modes which Here, X and Y denote the most-confined directions in the have different spectra. Finally, in the Lamb-Dicke regime trap, which will have the same normal-mode spectra for where ∆k(i)∆r(i) 1, and when the laser frequencies symmetric traps commonly used. These will be denoted are chosenh such thati  all transitions except for those near as transverse directions. The least-confined direction is the first sideband transitions6 are far off-resonant, the denoted as Z and is named the axial direction. three sets of Raman-beam pairs at each ion induce the Consider now the ion-laser system in the interaction laser-ion Hamiltonians of the form picture, in which all excitations arising from the free N N Hamiltonian (i) (i) (i) iµ t+i∆ϕ −iµ t+i∆ϕ0 H = iΩ e I I + e I I N ω N 1 I I × 0 (i) T † i=1 m=1 H0 = σz + ωm(amam + )+ X   X 2 2 (i) − T † T (i) i=1 m=1  eη a e iωmt + a eiωmt (α iα )σ + h.c., X X I,m m m 1 − 2 + A † 1 T † 1 ω (b bm + ) + ω (c cm + ) + const. (2)   (3) m m 2 m m 2  T are rotated away by frequencies of the order of ω0, ωm, A 4 † and ωm. am (am) annihilates (creates) a phonon ex- N N (i) (i) citation of the transverse normal mode m with angu- (i) iµ t+i∆ϕ −iµ t+i∆ϕ0 H = iΩ e II II + e II II lar frequency ωT along the X direction of the trap, i.e., II II m i=1 m=1 ∆k = ∆k Xˆ . Similarly, b and c (b† and c† ) are, X   X I I m m m m (i) − A † A (i) eη b e iωmt + b eiωmt (α iα )σ + h.c., respectively, the phonon annihilation (creation) opera- II,m m m 1 − 2 + tors for the axial normal modes along the Z direction,   (4)

3 These Cartesian indices must not be confused with the x, y and z indices introduced on quasi-spins of the qubit. While the former (upper-case letters) correspond to the components of laser fields’ 5 At this point, such assignments of a given set of normal modes k-vector, the latter (lower-case letters) correspond to the Bloch- to one of the Hamiltonians in Eqs. (3-5) appear arbitrary. The sphere axes in the qubit Hilbert space rationale behind the choices made will become clear in applica- 4 Although the axial modes are generally low in frequency, such a tions of the scheme to nearest-neighbor Hamiltonians considered rotating-frame approximation is still valid as long as lasers’ de- in this work, see Sec.III. tunings from these modes remain small compared to the sideband 6 The nth blue (red) sideband transition for mode m adds (re- Rabi frequencies of the axial motion. moves) n quanta of motion each with frequency ωm. 5

N (i) (i) (i) an effective Hamiltonian iµIII t+i∆ϕIII −iµIII t+i∆ϕIII0 HIII = iΩIII e + e N i=1 1 X   (i) (i) N HB = Bz σz (6) e T T −2 (i) −iωmt † iωmt (i) (i) i=1 ηIII,m cme + cme (α0I + α3σz ) + h.c., X m=1   is generated, but at the cost of the following change: X (i) (i) (5) µ µ + Bz and µ µ Bz to the laser I → I I → I − detuning in the first and second occurrences of µI in (i) 1 (i) (i) Eq. (3), respectively. Similarly, µ must be replaced by where σ = (σx iσy ), and tilde over the Hamil- II ± 2 ± (i) (i) tonians imply the use of the rotated frame described µII µII + Bz and µII µII Bz in the first and → → − above. Here, it is assumed that µI ω0 where second occurrences in Eq. (4), respectively. The laser 0 | |  µI ω0 ∆ωI = ω0+∆ωI . Similarly, µII ω0 where detuning µIII , on the other hand, remains unchanged. ≡ − − 0 | |  µII ω0 ∆ωII = ω0 + ∆ωII . On the other hand, for Note that this scheme requires a frequency control, as ≡ − − the detunings are now generally different at the location the Hamiltonian HIII , it is assumed that µIII ω0 where µ ∆ω = ∆ω0 . Further,| two distinct|  of each ion. III ≡ − III III Raman-beam phasee differences are assigned to each of the With the Hamiltonians in Eqs. (3-6), an evolution op- red (unprimed) and blue (primed) detuned frequencies of erator can be formed by applying a Magnus expansion, 2 (i) (∆kI ) (i) (i) taking into account all contributions up to and including the beam. η = T bm , where bm is the (normal- 2 I,m 2Mωm η , ηBz in the exponent: ized) normal-modeq eigenvector components between ion O
N i and mode m, and M denotes the mass of the ion. Sim- (α) 2 2 (i) (i) (∆kII ) (i) (i) (∆kIII ) (i) U(t, 0) = exp φi (t) σα + ilarly, η = A bm and η = T bm II,m 2Mωm III,m 2Mωm "α=x,y,z i=1 for the axial andq transverse modes, respectively.q For each X X pair of Raman beams L, the same ∆kL-vector is applied (α) (i) (j) 1 χi,j (t) σα σα , (7) at the location of each ion. α1 = 2 and α2 = 0 cor- ⊗  i,j respond to the well-known Molmer-Sorenson scheme, al- X  ready applied in a number of experiments. In order to where eliminate a bias σz interaction arising from HIII , it is essential that α is set to zero. With the scheme pre- N N N 0 φ(x)(t) = α(x) (t) a† + β(x) (t) b† c + h.c., sented in AppendixB, it is shown that one cane achieve i i,m m i,m,n m n m=1 m=1 n=1 this requirement by tuning the Raman-beam frequencies X X X 1 (8) and polarization vectors. We further set α3 = 4 for con- sistency between the effective spin-spin couplings arising N N N (y) (y) (y) 7 † † from HI , HII , and HIII . Now by setting the phases of φi (t) = αi,m(t) bm + βi,m,n(t) amcn + h.c., (i) m=1 m=1 n=1 the blue and red-sideband detuned beams to ∆ϕI = 0, X X X 0 0 0 (9) ∆ϕ (i)e= πe, ∆ϕ(i) e= ∆ϕ (i) = ∆ϕ(i) = ∆ϕ (i) = 0, I II II III III N the Hamiltonians HI , HII , and HIII can be seen to be (z) (z) (z) † (i) (i) (i) φi (t) = γi (t) + αi,m(t) cm + proportional to σx , σy , and σz , respectively. m=1 X Finally, an effectivee longitudinale e magnetic field can be N N (z) † introduced at the location of each ion by another N sets βi,m,n(t) bman + h.c. (10) of beams inducing a Stark shift to be tuned to the desired m=1 n=1 X X value of the magnetic field. Alternatively, a Bz field can be generated with the existing sets of Raman beams, i.e., The definition of the rest of the functions in Eqs. (7-10) by shifting the frequency of red and blue-detuned beams are provided in AppendixC. (i) (i) When Bz = 0, all contributions proportional to by Bz . This can be seen by noting that if the rotating frame that led to Eqs. (3-5) is assumed to rotate with phonon creation and annihilation operators in the ex- ponent in Eq. (7) are bounded in time, provided the Hamiltonian H + 1 N B(i)σ(i) instead of H , in 0 2 i=1 z z 0 that µ = µ = µ . As a result, an effec- addition to the interacting Hamiltonians in Eqs. (3-5), I 6 II 6 III P tive Heisenberg model can be achieved when t T −1 A −1 T −1  µI ωm , µII ωm , µIII ωm , so that the terms| − linear| in| time− in| Eq.| (7) (those− | proportional to (α) (α) χi,j ) dominate the evolution. In such a limit, χi,j 7 There will be no ambiguity in the overall constants in the Hamil- → i J (αα)t (see Eqs. (13-15)), and other contributions will tonian. Rescaling these coefficients by a constant means the Rabi − 2 i,j frequencies must be rescaled accordingly so that the expected be subdominant. For practical (noisy) implementations, strength of the state-dependent force is produced on a given ion, one needs to minimize the spin-phonon entanglement and with given choices of the internal levels for the qubit. arising from the first term in the exponent in Eq. (7) at 6 early times. This is achieved with η(i) Ω(i) µ ωT , tions are suppressed compared to the first-sideband tran- | I,m I |  | I − m| (i) (i) A (i) (i) T sitions. These conditions are easier to satisfy for trans- ηII,mΩII µII ωm , and ηIII,mΩIII µIII ωm . | (i|)  | (−x) | | (y) |  | − | verse modes than the axial modes. This is because the When Bz = 0, α (t), and α (t) in Eqs. (8) and (9) 6 i,m i,m axial modes have lower frequencies, and their correspond- develop an oscillatory time dependence but with a lin- ing Lamb-Dicke parameters are larger. Finally, one notes ear growth in the magnitude of its amplitude. These that coherent operations on a single spin correspond to (i) (i) (i) (i) terms are proportional to Bz σy and Bz σx . Assum- the zeroth-order terms in Eq. (1) in the Lamb-Dicke limit, ing that the magnetic field is comparable in size to the (i) and with ∆ωL = ω0. Hence, the laser frequencies ap- effective spin-spin couplings, such contaminating terms plied must be far detuned from such “carrier transitions” do not severely impact the desired evolution as long as of the ions. (i) (i) (i) (i) (i) Bz η Ω , η Ω . Unfortunately, this con- | |  | I,m I | | II,m II | dition limits the size of (effective) magnetic fields that can be studied in models considered below. Nonetheless, III. OPTIMIZED SPIN-SPIN HAMILTONIANS a range of interesting possibilities can still be explored. IN AN ION TRAP: 1+1D SCHWINGER MODEL Under the conditions described above, the time- evolution operator in Eq. (7) can be approximated as A unique testbed for exploring theoretical and experi- mental proposals for quantum simulations of gauge theo- U(t) e−iHefft, (11) ≈ ries is the 1+1D QED, i.e., the Schwinger model. It is an where Abelian gauge theory, hence avoiding complexities of its non-Abelian counterparts. It is also a low-dimensional (xx) (i) (j) (yy) (i) (j) Heff = Ji,j σx σx + Ji,j σy σy + theory, allowing numerical and experimental studies of ⊗ ⊗ its approximate dynamics with finite resources. Despite i,j h Xj

7 units of ag2/2, where a denote the lattice spacing and g is a long-range spin-spin interaction, representing the 1D is the original fermion-gauge field coupling. The dimen- Coulomb interaction among the charged fermions. sionless parameters x and µ are related to dimensionful Given the experimental setup presented in the previ- parameter g (with mass dimension one) and the original ous section, engineering the Schwinger Hamiltonian for mass m via: x = 1/(ag)2 and µ = 2m/(ag2).8 given values of N (which maps directly to the number of The familiar Jordan-Wigner transformations Φn = ions), x and µ amounts to finding values of lasers’ Rabi (l) (n) † (l) (n) (i) (i) (i) l

Here, 0 is the electric field flux into the first lattice site which can be set to zero without loss of generality. To With a single beatnote frequency on each pair of Ra- make explicit the mapping of this Hamiltonians to that man beams, the Schwinger Hamiltonian on small lattices of the Hamiltonian of the ion-laser system in our pro- can be realized with good accuracy. For this example, posed scheme, Eq. (12), one can note that Eq. (17) can an ion trap consisting of 171Yb+ ions will be considered. be rewritten as The specifications of this system are presented in Ap- pendixA. Consider the case of N = 4, and further set (xx) (yy) (zz) (z) H = H + H + H + H , (18) the values of the parameters of the Schwinger Hamilto- nian to x = 6 and µ = 1. The Hamiltonian H(xx) can be where achieved by first noting that a certain detuning from the − x N 1 CM transverse mode with the same amplitude on each H(xx) = σ(n)σ(n+1), (19) ion produces the coupling matrix shown in the left panel 2 x x n=1 of Fig.2. This matrix can be systematically turned into X N−1 a nearest-neighbor form: the slope of the decline in the x H(yy) = σ(n)σ(n+1), (20) strength of nearest-neighbor couplings from the center of 2 y y n=1 the chain can be determined, and be used to systemati- X N−2 N−1 cally adjust the Rabi frequencies in such a way that an (zz) 1 (m) (n) H = (N n)σ σ , (21) equal strength is achieved on all Ji,j with i j = 1, as 2 − z z | − | m=1 n=m+1 demonstrated in the right panel of Fig.2. The most ac- X X − curate nearest-neighbor Hamiltonian achieved with this µ N 1 N 1 n H(z) = ( 1)nσ(n) (n mod 2) σ(l). procedure presents a 3% contamination on the non- 2 − z − 2 z ∼ n=1 n=1 nearest-neighbor elements, and no contamination on the X X Xl=1 (22) nearest-neighbor elements. As mentioned in Sec.II, the H(yy) effective Hamilto- H(xx) and H(yy) represent nearest-neighbor spin-spin in- nian is chosen to arise from the Raman beams that ad- teractions and share the same coupling strength. H(zz) dress the axial modes of the ions. If the transverse modes were to be addressed, the Raman beams would have to be detuned from the modes by the same amount as those for the H(xx) Hamiltonian, as these appear with the same 8 x and µ here should not to be confused by the spin x axis and the coupling in the Schwinger Hamiltonian. This however lasers’ detunings, respectively. Their meaning should be clear in would cause the dynamics to deviate from the effective the context they appear. Heisenberg model in Eq. (12), given the non-zero commu- 8

(xx) (xx) (xx) Ji,j /2⇡ [kHz] H Ji,j /2⇡ [kHz]

ion index j ion index j ion index i ion index i

��� �� �� [kHz] [kHz] �� ⇡ ⇡ �� 2 �� 2 /

�� ) / i

) �� I ( i I (

⌦ � ⌦ � � � � ⌘ � � � � ⌘ ion index ion index

T T !m/2⇡ !m/2⇡ µI /2⇡ µI /2⇡

� � � � � � � � � � µ/2⇡ [MHz] µ/2⇡ [MHz]

FIG. 2. [Left panel] The effective spin-spin coupling matrix J (xx) in Eq. (13) resulting from pairs of Raman beams addressing 4 individual ions at the Rabi frequency Ω(i), where i = 1, ··· , 4. All beams are detuned from the transverse COM mode, T T ω1 = 2π × 4.135 MHz, by the same frequency, µI − ω1 = −2π × 830 kHz. The Lamb-Dicke parameter, η, multiplying the p 2 T Rabi frequencies in the figure is η = (∆kI ) /4πMν ≈ 0.068. [Right panel] With the same detuning, the Rabi frequencies (xx) can be adjusted to match the magnitude of the Ji,j matrix elements for |j − i| = 1 in the left panel, producing exactly equal magnitude on these elements, in addition to small non-nearest neighbor contributions, as shown in the right panel. Here, the J (xx) matrix is tuned to produce H(xx) of the 4 fermion-site Schwinger model in Eq. (19) with x = 6. Numerical values associated with this figure are provided in Supplemental Material.

tations between HI and HII in Eqs. (3) and (4), generat- in Fig.4 allow the Ji,j coupling to be tuned to the de- ing phonon-dependent terms that grow (or decline) lin- sired values with below-percent accuracy. However, in early with time.e Such contaminationse are circumvented contrast with the case of nearest-neighbor Hamiltonians, by producing the H(yy) Hamiltonian with the Raman the procedure that finds the adjusted Rabi frequencies for beams that couple to the axial modes. Note that the ax- H(zz) is not systematic, making it challenging to general- ial modes have a very different frequency spectrum com- ize such an ad hoc tuning procedure to a higher number pared with the transverse modes. The same procedure as of ions. Finally, an effective H(z) Hamiltonian can be for the H(xx) mapping can be used to find the values of induced using N sets of Raman beams with their Stark the laser beatnote and Rabi frequencies that generate a shift tuned to reproduce H(z) of the Schwinger Hamil- nearest-neighbor interaction with these modes, see Fig.3. tonian in Eq. (22). The values of the effective magnetic As discussed at the end of Sec.II, a critical check is to fields that are required given the chosen parameters of ensure the higher-sideband contributions to the applied the model are depicted in Fig.5. Molmer-Sorenson scheme are not significant given the low normal-mode frequencies in the axial direction, and given It is crucial to verify that the laser parameters found the laser frequencies applied. It can be shown that the in such a mapping do not violate the conditions enumer- largest contribution from these higher-order sidebands is ated in the previous section, and the true dynamics is only a few percent of the contribution from the first side- that dictated by the effective Heisenberg Hamiltonian in band, and will be ignored in the current proposal. Eq. (12). This check can be done by a numerical evalua- tion of all contributions to the exponent of the evolution An effective H(zz) Hamiltonian that matches that of operator in Eq. (7), up to and including (η2, ηB). Here, the Schwinger model can be achieved with a single beat- we assume that the experiment can be initiatedO in a state note frequency, and by addressing the other set of trans- with zero phonon occupation in all modes. The results verse normal modes of the ions. Here, the values shown of this investigation are shown in Fig. 14 of AppendixE 9

(yy) (yy) H Ji,j /2⇡ [kHz] ion index i ion index j

ion index j

(yy) ion index i Ji,j /2⇡ [kHz]

��� [kHz] [kHz] ��� � ⇡

⇡ ��� 2 2 ��� -��� / / ) )

�� i i II ( II ( -��� ⌦ ⌦ � � � �

⌘ � � � � ⌘ ion index ion index

A A !m/2⇡ !m/2⇡ µII/2⇡ µII/2⇡

� � � � � � � � � � � � � � � � µ/2⇡ [MHz] µ/2⇡ [MHz]

FIG. 3. [Left panel] The effective spin-spin coupling matrix J (yy) in Eq. (14) resulting from pairs of Raman beams addressing (i) A 4 individual ions at the Rabi frequency Ω , where i = 1, ··· , 4. All beams are detuned from the axial COM mode, ω4 = A 2π × 0.713 MHz, by the same frequency, µII − ω4 = 2π × 3160 kHz. The Lamb-Dicke parameter, η, multiplying the Rabi p 2 A frequencies in the figure is η = (∆kII ) /4πMν ≈ 0.081. [Right panel] With the same detuning, the Rabi frequencies can (yy) be adjusted to match the magnitude of the Ji,j matrix elements for |j − i| = 1 in the left panel, producing exactly equal magnitude on these elements, in addition to small . 3% non-nearest neighbor contributions, as shown in the right panel. Here, the J (yy) matrix is tuned to produce H(yy) of the 4 fermion-site Schwinger model in Eq. (20) with x = 6. Numerical values associated with this figure are provided in Supplemental Material.

for the first ion, and in Supplemental Material for the ments. Twenty such Hamiltonians are considered, as rest of the ions. As shown, the dominant source of error listed in Supplemental Material, with errors on the non- is related to the nonzero commutations of H in Eq. (6) nearest-neighbor elements in the range 3% 18%. The B ∼ − with HI and HII in Eqs. (3) and (4), introducing effective evolution of the 4 fermion-site Schwinger model is then magnetic fields along the x and y spin axes. These are a considered. The quantity of interest here is the vacuum smalle fractione of the desired field along the z direction, persistence amplitude (VPA), defined as the (square) of but are however dependent upon the phonon occupation the overlap of the state of the system at time t, ψ(t) | i in the system. with the “vacuum” (a state in the physical sector of the theory with no net electron-positron pair), ψ(vac) . This Hamiltonians of the lattice Schwinger model for a quantity is plotted for select times in the smaller| panelsi of larger number of fermion sites can be shown to be acces- Fig.6 for all the twenty inexact Hamiltonians used in the sible through the single-frequency and multi-amplitude evolution. A procedure is described to estimate a mean scheme described, but deviations from the exact Hamil- and uncertainty band from the most accurate Hamiltoni- tonian can be significant. For N = 10 and the nearest- ans employed. Nonetheless, as is seen in the larger panel neighbor Hamiltonian with transverse modes, the best of the figure, during certain times, the estimate of VPA parameters found give rise to errors as high as 20% ∼ deviates significantly from the expected result, and this in the non-nearest-neighbor elements. To investigate the feature is amplified at longer times. effect of inexact Hamiltonians on the dynamics of the This observation promotes adopting a multi-frequency Schwinger model, we have studied a lattice Schwinger and multi-amplitude scheme,9 as proposed previously in model with N = 4, x = 0.6 and µ = 0.1 with exact engi- neered Hamiltonians H(zz) and H(z) but with a nearest- neighbor Hamiltonian H(xx)(= H(yy)) that differs from the exact form via nonzero non-nearest-neighbor ele- 9 We use the term frequency for the beatnote frequency of the Ra- 10

J (zz)/2⇡ [kHz] J (zz)/2⇡ [kHz] i,j H(zz) i,j

ion index j ion index j

ion index i ion index i

�� ��

[kHz] �� [kHz] ��

⇡ �� ⇡ �� 2 2

/ �� / �� ) )

i � i � ( III ( III

⌦ � ⌦ �

⌘ � � � � ⌘ � � � � ion index ion index

T T !m/2⇡ !m/2⇡ µIII/2⇡ µIII/2⇡

� � � � � � � � � � µ/2⇡ [MHz] µ/2⇡ [MHz]

FIG. 4. [Left panel] The effective spin-spin coupling matrix J (zz) in Eq. (15) resulting from pairs of Raman beams addressing 4 individual ions at the Rabi frequency Ω(i), where i = 1, ··· , 4. All beams are detuned from the transverse COM mode, T T ω1 = 2π × 4.135 MHz, by the same frequency, µIII − ω1 = 2π × 100 kHz. The Lamb-Dicke parameter, η, multiplying the p 2 T Rabi frequencies in the figure is η = (∆kIII ) /4πMν ≈ 0.068. [Right panel] With the same detuning, the Rabi frequencies can be adjusted so that the J (zz) matrix produces the long-range couplings in H(zz) of the 4 fermion-site Schwinger model in Eq. (21) with x = 6. Numerical values associated with this figure are provided in Supplemental Material.

(z) (i) H BJzi,j/2⇡ [kHz]

ion index i

(i) (z) FIG. 5. The effective magnetic field on each ion, Bz , that produces the H Hamiltonian of the Schwinger model, Eq. (22), for N = 4 and µ = 1. Numerical values associated with this figure are provided in Supplemental Material.

Ref. [78] in the context of quantum simulation of the Ising tem to that of the Schwinger model can be achieved with model on two-dimensional lattices. With this scheme, unprecedented accuracy, as is shown in the following. mapping of the effective Hamiltonian of the ion-laser sys-

A multi-frequency and multi-amplitude scheme

man beams unless it is identified as otherwise. A multi-frequency scheme, therefore, refers to when multiple beatnote frequencies The extension of the formalism presented in Sec.II to are used, while a multi-amplitude scheme refers to when multiple a multi-frequency scheme is straightforward. For exam- Rabi frequencies are applied. ple, the effective spin-spin coupling engineered by Raman 11

���� 2

���� t =6 t = 10 Inexact Hamiltonians as listed in Supplemental Material

i ○○ ○○○ ���� ○○ ) ○○ ○○ Inexact Hamiltonians as listed in Supplemental Material t ○ ○

( ○ ��� ○ ○ ○ Inexact Hamiltonians as listed in Supplemental Material

���� ○○ ○ | ○ ○ ○ ○ ���� ○ ○○ ○ ���� ○ ○ Evolution with inexact engineered Hamiltonians ○ ○ ○ ○ ○○ (vac) ○ ○Evolution○ with inexact engineered Hamiltonians

��� ○ � h

Evolution with the exact Hamiltonian � ������ ������ ������ � ������ ������Evolution������ with the exact Hamiltonian (XX) (XX)

○ ��� ���� t = 14 ○ ○○ ���� ○ ○ ��� ○ ○ 2 ���� ○○ ○

i ○ ○ ) ○ t ○○○○○○ ( ��� �

| � ������ ������ ������

○○ (vac) ○ ○ ��� ��� ○ ○○○○ ○○ ○ h ���� ○ ○ ○ ��� ��� ○ ○ ���� ○ ○ ��� t = 18 ��� ○ � ������ ������ ������ � � �� �� �� t (XX)

FIG. 6. Time evolution of the state |ψ(vac)i = | ↓↑↓↑i corresponding to the strongly-interacting vacuum of the 4 fermion- site lattice Schwinger model with x = 0.6 and µ = 0.1. Open circles in the upper and left plots are the values of VPA at select times with 20 inexact H(xx)(= H(yy)) Hamiltonians, as listed in Supplemental Material. The 9 data points that satisfy (xx) (xx)2 (xx)2 (xx)2 −4 ∆ ≡ J1,3 + J1,4 + J2,4 ≤ 10 are chosen to define central values (dark-pink lines) and uncertainties (pink bands) on the VPA, and are compared with the exact expectations (blue lines). The plot in the lower-left panel represents the exact time evolution of vacuum (blue curve) compared with the central value (dark-pink curve) and uncertainty (pink band) on the VPA obtained from 9 Hamiltonians that give rise to ∆(xx) ≤ 10−4. Numerical values associated with these plots are provided in Supplemental Material.

pairs I generalizes to terms that would scale as the number of beatnote fre- quencies introduced. One therefore needs to ensure that n eµI N (i) (j) the cumulative effect of such terms remain negligible (xx) (i) (j) bm bm J = Ω Ω RI , (23) i,j I,m0 I,m0 2 T 2 compared with the desired effective Heisenberg Hamil- m=1 µI,m0 ωm mX0=1 X − tonian. Fig.7 demonstrates the success of this scheme in where n is the number of beatnote frequencies, and µI an accurate generation of the long-range part of the where each detuning µ is associated with the Rabi fre- I,m0 Schwinger Hamiltonian, H(zz), for the case of N = 8 quency Ω(i) .10 Similarly, the J (yy) and J (zz) coupling e I,m0 i,j i,j ions. Here, the corresponding optimization problem is matrices can be obtained by replacements µ µ , II II,m0 solved, and the desired effective spin-spin Hamiltonian is (i) (i) (i) → (i) Ω Ω , µIII µIII,m , and Ω Ω , achieved with errors that are comparable with the ma- II → II,m0 → 0 III → III,m0 0 (yy) chine precision. The laser frequencies are fixed such that where a summation over m is assumed. For Ji,j , one T T T T A µI,m = ω + fs(ω ω ), with fs = 0.5, and must replace ω with ω . More generally, the full time 0 m0 m0 m0+1 m m where m0 runs from 1− to n = 7, see the lower-right− evolution operator in Eq. (7) can be constructed by per- III plot of Fig.7. 11 The corresponding Rabi frequencies at forming the changes described in the ion-laser Hamiltoni- ans in Eqs. (3-5). This introduces additional off-resonant e

11 In the convention of this work, the normal mode frequencies are ordered in a set from the highest value to the lowest value. There- fore for the axial mode, ωA denotes the CM mode, while for the 10 We remind that the effective spin-spin Hamiltonian arises from N T a bichromatic pair of Raman beams, one detuned by −µ transverse mode, the CM mode is ω1 . Because of this conven- I,m0 (red detuned) and one by µ (blue detuned) from the carrier tion, the normal-mode eigenvectors b(i) must be ordered accord- I,m0 m transition, see discussions after Eq. (5). ingly for the transverse and axial modes. (i) ⌘⌦m /2⇡ [kHz] (i) ⌘⌦III,m/2⇡ [kHz] 12

⌘⌦(i) /2⇡ [kHz] (zz) (zz) III,m0 H Ji,j /2⇡ [kHz]

ion index mode index

ion index j

ion index i !T /2⇡ m0 µIII,m0 /2⇡ ��� ��� ��� ��� ��� ��� ��� (t) !m /2⇡ [MHz]

FIG. 7. The effective spin-spin coupling matrix J (zz) in Eq. (15) resulting from multiple pairs of Raman beams addressing N = 8 individual ions at the Rabi frequency Ω(i) , where i = 1, ··· , 8 and m0 = 1, ··· , 7. The pairs of beams addressed at III,m0 T T T ion i are detuned from the transverse COM mode by 7 different frequencies, µIII,m = ω + fs(ω − ω ) with fs = −0.5, 0 m0 m0 m0+1 as denoted in the lower-right of the panel. The Lamb-Dicke parameter, η, multiplying the Rabi frequencies in the figure is η = p(∆k)2/4πMνT ≈ 0.068. Here, the J (zz) matrix is tuned to produce H(zz) of the 8 fermion-sites Schwinger model in Eq. (21). Numerical values associated with this figure are provided in Supplemental Material.

the location of each ion are plotted in the upper-right protocols for a fully analog simulation of the Schwinger plot of Fig.7. As is seen, a perfect agreement between model for given parameters with i) a scheme that requires (zz) only individual amplitude and phase control of the laser Ji,j and that in the Schwinger model with x = 6 and µ = 1 is achieved. The reason for choosing a large value beams and engineers an approximate Schwinger Hamil- of the coupling x in the original theory is to minimize the tonian, and ii) a scheme that takes advantage of indi- error to the effective Heisenberg Hamiltonian due to the vidual amplitude, phase and frequency control and en- unbounded contributions arising from the commutations gineers the desired Hamiltonian with great accuracy (up to errors associated with the difference between the full of the Bz Hamiltonian in Eq. (6) and HI and HII in ion-laser evolution and the effective Heisenberg model, Eqs. (3) and (5). Note that the desired effective Bz field in the Schwinger Hamiltonian grows witheN evene in the which are nonetheless assured to remain negligible in the limit µ = 0. Hence, in order to keep the undesired con- schemes proposed). It is clear that the second scheme tribution small compared with the effective Hamiltonian, can be easily applied to any number of ions at the cost of the strength of the nearest-neighbor terms is taken to be introducing a multitude of laser frequencies, the number stronger by setting x = 6. As is shown in AppendixE for of which grows with the number of ions. This can be al- ready achieved with current technologies for up to 30 the first ion, and in Supplemental Material for the rest of ∼ the ions, all the contributions to the exponent in the full ions, and most importantly is scalable, as it involves a time-evolution operator (up to the order considered) are linear growth in the complexity of the classical control small (and mostly bounded) compared with those that hardware of the experiment. constitute the Hamiltonian of the Schwinger model. The laser parameters for a nearly exact engineering of H(xx), H(yy), and H(z) are shown in Figs. 11-13 of AppendixD. In the following, other examples of LGTs whose dy- It must be noted that the optimization problem in all 1 cases is solved under two constraints: i) the sum of Rabi namics can be mapped onto a spin- 2 system will be dis- frequencies at the location of each ion is less than or equal cussed. The goal is to only point out the potential of an to 2 MHz, ii) the contribution to the full evolution from ion-trap quantum simulator in addressing more complex the first-order terms, those proportional to coefficients spin systems by providing examples of relevant gauge (x) (y) (z) theories. Explicit scenarios for given ion-trap architec- α , α , and α in Eqs. (8-10), remains below 0.5 at i,m i,m i,m tures are straightforward to obtain, following optimiza- several random times up to 1 ms. tion strategies presented for the case of the Schwinger To summarize, we have provided detailed experimental model. 13

IV. ANALOG SIMULATIONS OF SYSTEMS IN of the Chern-Simons’ term in the Lagrangian density: HIGHER DIMENSIONS WITH A 1D CHAIN OF IONS = a†(x)iD a(x) a†(x)eiAj (x)a(x + nˆ )+ LCS 0 − j j=1,2 X h With a generic Heisenberg model and an effective mag- θ h.c.] µνλA (x)F (x) (24) netic field engineered in Sec.II, it is clear that a wide − 4 µ νλ range of couplings among spins can be tailored, as was is θ = 1 [90]. Here, time is assumed to be continuous demonstrated for the case of the Schwinger model. In 2π particular, as seen in Sec.III, the H(αα) with α = x, y, z while spatial coordinates are defined on a square lattice, i.e., x = (t, n) where n is a vector whose components are does not have to be necessarily nearest neighbor or of any 12 particular form, as the multi-frequency, multi-amplitude integer multiples of the lattice spacing. µ, ν = 0, 1, 2 with the zeroth direction being the time direction, a is scheme of this work allows an arbitrary Ji,j to be pro- duced. This observation implies that spin systems in a complex spinless fermion field, Aµ is the gauge field, higher spatial dimensions can be engineered as well, as Dµ = ∂µ iAµ is the covariant derivative, Fµν is the field-strength− tensor: F = ∂ A ∂ A , and µνλ is was also noted in Ref. [78]. One only needs to map the µν µ ν − ν µ points on a 2D or 3D lattice to a linear chain of ions along the Levi-Civita symbol. Note that the A0 field does not with their corresponding couplings. Of course, with a have any dynamics and can be set equal to zero with fixed number of ions in a given experiment, this means the choice of a temporal gauge. The physical sector of that the finite-size effects in the dynamics of the system the theory, i.e., states that satisfy the Gauss’s law, can be identified from the condition δS = 0, where S is under study will be larger, as e.g., in the case of square δA0 the action. These states then correspond to those for and cubic lattices the spatial extent of the system will be † 1/2 1/3 which a (x)a(x) θ [A (x + nˆ ) A (x)] = 0. It is N and N , respectively. Nevertheless, this possibil- − ij j i − j ity implies that a linear quantum system can be used as a also clear that the Hamiltonian of the theory vanishes in platform for analog simulations of theories in any dimen- the absence of matter fields, which is a desired feature of sion, bringing the versatility of such an analog platform the topological theory. In the presence of matter fields, closer to its digital counterpart. i.e., the Hamiltonian corresponding to Eq. (24) is

† iAj (n) HCS = a (n)e a(n + nˆ j) + h.c. .(25) n j=1,2 X X h i A. 2+1D Abelian Chern-Simons theory coupled to Note that the time dependence of the fields is now im- fermions plicit considering the Hamiltonian equations of motion. As is shown in Ref. [90], the gauge links can be elimi- As an example of an interesting field theory in 2 + 1D, nated from the Hamiltonian with the use of Gauss’s law, consider the Chern-Simons theory coupled to fermions. at the cost of changing the equal-time commutation re- This theory is of broad impact on a range of problems lation of fermions. This is in fact a great advantage since 1 1 in theoretical physics, from the theory of the integer and when θ = 2π (or in general when 2θ is an odd multi- fractional quantum Hall effects to knot theory and par- ple of π), the new commutation relations are those of 1 ity anomalies in quantum field theory, see Ref. [89] for a hardcore bosons, i.e., the spin- 2 matrices. As a result, review. Since the theory is topological in the continuum, this procedure can be realized as a 2D generalization of the construction of a discretized counterpart of the theory the familiar Jordan-Wigner transformation. Explicitly, turned out to be non-trivial as a lattice has explicit refer- by performing the transformations a eiφa a˜ and † † −iφ † → ≡ ence to a given coordinate system and metric. However, a a e a˜ , where Aj(n) φj(n + nˆ i) φj(n), it has been shown [90, 91] that one can still formulate a one→ arrives at≡ ≡ − U(1) LGT that retains gauge invariance on arbitrary 2D (n) (n+nˆ j ) planar graphs, has no local excitations (hence is topo- HCS = σ+ σ− + h.c. , (26) n j=1,2 logical) and in the long-wavelength limit approaches the X X h i Chern-Simons theory in the continuum. As is discussed (n) in Ref. [91], a lattice formulation of the Chern-Simons where the following identifications are assumed: σ+ = † (n) (n) † theory is invaluable in investigations of fractional Chern a˜ (n), σ− =a ˜(n), and σz = 1 2a (n)a(n). Eq. (26) insulators that occur in given lattice geometries. As a clearly corresponds to an XY spin− model. Note that a pa- result, it is interesting to ask if a quantum-simulation rameter h could be introduced to control the magnitude protocol for this theory can be devised on the simulating of the hopping term in the Hamiltonian. platform of this work. To perform an analog simulation of such a 2D XY A known result [90] in the context of the generalized model within the scheme presented in Sec.II requires Jordan-Wigner transformation in higher dimensions is 1 Fradkin’s proof of equivalence between the spin- 2 XY model on a 2D Bravais lattice and a Chern-Simons theory 12 in 2+1D coupled to fermions, provided that the strength For a general formulation on 2D planar lattices, see Refs. [91, 92]. 14

2D lattice Coupling matrix

(3,0) (3,1) (3,2) (3,3) J i,j [kHz] 2⇡ (2,0) (2,1) (2,2) (2,3)

(1,0) (1,1) (1,2) (1,3) ion index j

(0,0) (0,1) (0,2) (0,3) ion index i

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 (0,0) (0,1) (0,2) (0,3) (1,0) (1,1) (1,2) (1,3) (2,0) (2,1) (2,2) (2,3) (3,0) (3,1) (3,2) (3,3) Ion chain

1 FIG. 8. The upper left panel shows a 4 × 4 lattice of spins (s = 2 ) with nearest-neighbor interactions, corresponding to the σx ⊗ σx (or equivalently σy ⊗ σy) interactions in the Hamiltonian in Eq. (26) with n = (nx, ny), where nx (ny) runs from 0 to 3, and where an open boundary condition is adopted. The nearest-neighbor interactions of a select site are depicted in green links. This 2D configuration can be mapped to a 1D chain of ions, as shown in the lower panel, along with the couplings of the select site in the new configuration. The obtained 1D coupling matrix Ji,j is shown in the upper right panel.

(XX) (YY ) optimizing a (Ji,j ) Ji,j = Ji,j matrix by perform- B. 2+1D pure Z2 lattice gauge theory ing a multi-frequency,≡ multi-amplitude Molmer-Sorenson scheme using the transverse and axial normal modes of motion. For a 4 4 lattice in the target theory, a system of ZN gauge theories are discrete Abelian gauge theo- N = 16 ions can× be used as is shown in Fig.8, along with ries that given their simple underlying symmetry have long served as a testbed for gaining deeper perspectives the required Ji,j matrix. Obtaining the laser frequencies and amplitudes is a straightforward optimization process, on gauge theories. Despite their simple structure, they as detailed in the previous section, and in fact machine can have non-trivial phase diagrams exhibiting e.g., a precision accuracy can be achieved, as demonstrated in confining phase. In fact, since Z3 is the center of the Ref. [78] for similar geometry and coupling profiles. Fi- SU(3) group, the confinement in the Yang-Mills theory nally, we should remark that the full Hamiltonian in such is attributed to the Z3 symmetry. These gauge theo- a 2+1D Abelian LGT must include the energy stored in ries have been the focus of numerous theoretical and ex- electric and magnetic fields, giving rise to the Maxwell- perimental proposals for quantum simulation of gauge Chern-Simons theory [93, 94].13 Aside from the ques- theories, in particular using neutral atoms in optical lat- tion of what is the proper formulation of a discretized tices [32, 47, 96]. An interesting feature of ZN is its Maxwell-Chern-Simons theory, one needs to account for duality with spin models. This connection has been de- the full dynamics of the gauge fields by mapping them to veloped over decades [97], starting from Wegner’s demon- those in an ion-trap quantum-simulation platform, which stration of such a duality for the case of a Z2 LGT [98], is beyond the scope of the present work. and has inspired similar duality constructions for non- Abelian gauge theories such as SU(N) [99]. Further, re- cent work has suggested that the 4D Z2 LGT provides a complete model for all classical spin models and all Abelian discrete LGTs [100, 101]. The example that will be presented here is a 2+1D Z2 LGT that is dual to a 2D Ising model, and is there- fore amenable to the quantum simulation protocol of 13 See also Ref. [95] for discussions regarding a non-Abelian case, this work. The Hamiltonian of the 2+1D Z2 LGT can the Yang-Mills-Chern-Simons theory. be expressed with a pair of conjugate spin operators 15

Original lattice Dual lattice

(3,0) (3,1) (3,2) (3,3)

(2,0) (2,1) (2,2) (2,3) l3 p l4 p (1,0) (1,1) (1,2) (1,3) l2

l1 (0,0) (0,1) (0,2) (0,3)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 (0,0) (0,1) (0,2) (0,3) (1,0) (1,1) (1,2) (1,3) (2,0) (2,1) (2,2) (2,3) (3,0) (3,1) (3,2) (3,3) Ion chain

FIG. 9. The upper-left panel shows a 5 × 5 spatial lattice corresponding to the Z2 Hamiltonian in Eq. (27). An open boundary condition is adopted, and a select plaquette term in the Hamiltonian is shown. The center of the plaquettes defines the sites of a dual lattice, as depicted by the green points, and are separately shown in the right panel. Such a 2D configuration corresponds to the Ising Hamiltonian in Eq. (28), which can now be mapped to a 1D chain of ions, as shown in the lower panel of the figure.

iπE(l) iA(l) σx(l), σz(l) , where σx(l) = e and σz(l) = e . the gauge conveniently such that σz on all links along one {Here, l denotes} a link on the 2D spatial lattice, A(l) is of the spatial directions is set to unity, ii) use the operator the gauge field evaluated on link l with A(l) = 0, π . identity G(n) = 1 in the physical Hilbert space of the the- { } E(l) is the corresponding “electric field” with E(l) = ory to replace σx along the same space direction as in i) 0, 1 . Note that in order to keep the presentation sim- with those along the other direction. These two steps in- { } ple, we have not used bold-faced quantities for the two- spire the replacements σz(l1)σz(l2)σz(l3)σz(l4) σx(p), 0 → dimensional vectors A(l) and E(l), as their directionality and ˜ ˜σx(˜l ) σz(p) (which is allowed as the new l0≤l → on the 2D plane is implicit from the directionality of the σx, σz set has the same commutation relations as the link arguments. The lattice Hamiltonian of such a pure original{ Q} set). In the first replacement rule, p denotes the gauge theory consists of “electric” and “magnetic” terms: plaquette formed by links l1, l2, l3, l4, and in the second H = σ (l) λ σ (l )σ (l )σ (l )σ (l ). rule, it denotes the plaquette whose left bottom corner 2+1D Z2 − x − z 1 z 2 z 3 z 4 is the point at which ˜l starts. The product is over all l p X X links prior to and including link ˜l, and tilde is used to (27) denote the space dimension for which the gauge remains Here, the first (second) sum runs over all links (plaque- unfixed. It is now easy to see that in terms of the new ttes) on the 2D lattice, and open boundary conditions are spin operators, the Hamiltonian in Eq. (27) can be writ- assumed. A plaquette is defined as the product of four ten as gauge links staring from the lower-left corner and moving counterclockwise, see the upper right panel of Fig.9. The 1 H = λ σ (p) σ (p)σ (p0) Hamiltonian in Eq. (27) remains invariant under a local 2D Ising − x − λ z z  p hp,p i gauge transformation which flips the sign of σz on links X X0  n n n n  sharing site n, but does not affect σx on links sharing the λ σ( ) σ( )σ( +ˆ j ), (28) ≡ − x − z z same site. The Gauss’s law corresponding to this sym- n n j=1,2 metry defines the physical sector of the theory, namely X X X states for which the eigenvalue of the Gauss’s law oper- where in the last line, n refers to points on the “dual” ator G(n) = n σx(ln) is unity, where ln denotes all the lattice defined by the center of spatial plaquettes in the four links that meet at point n. original lattice, see Fig.9. p, p0 in the first line denotes To establishQ a duality relation with the 2D Ising model, the nearest-neighbor plaquettes.h i For further detail on the gauge invariance can be taken into account to: i) fix the expected phase diagram of the theories at different 16 coupling regimes, see e.g., Ref. [97]. shifting the frequency of the red- and blue-detuned The duality between Eq. (27) and (28) allows to simu- Raman beams, eliminating the need for introducing late the dynamics of a Z2 LGT in 2+1D using a chain of another N laser beams to induce local Stark shifts ions in 1D whose interactions are tailored to correspond on the ions. to the Ising Hamiltonian, as discussed in the previous ex- ample of this section. The correspondence between the B Engineering an arbitrary Heisenberg Hamiltonian original 2D lattice, the dual lattice, and the chain of ions is enabled in this work by a thorough optimization is depicted in Fig.9. Engineering the nearest-neighbor procedure that minimizes the contributions arising σz σz interactions was detailed in Sec.II, and the ad- from unwanted couplings to phonon excitations, ditional⊗ global transverse magnetic field can be easily in- contributions that drive the dynamics away from troduced by performing single-qubit rotations, with an the effective spin-spin Hamiltonians. This is a cru- angle determined by the coupling λ in the original the- cial requirement for a reliable quantum simulation ory. that is addressed for the first time in this work. The purely spin formulation of the lattice Schwinger model exists, and corresponds to a Heisenberg XXZ model with both short and long-range interactions, V. CONCLUSION AND OUTLOOK and with an effective local magnetic field. The op- timization procedure described above was applied In this paper, we took on the question of how to best to this example with N = 8, and can be scaled leverage the current technologies in ion-trap analog quan- straightforwardly to any number of ions. tum simulators to engineer the Hamiltonian of gauge field theories. Towards this goal, gauge theories that can be B In this work, equal-size nearest-neighbor couplings experimentally realized in such platforms in the near fu- along the spin axesx ˆ andy ˆ are achieved through ture are enumerated, and are shown to be amenable to coupling to transverse and axial modes of the mo- a particular quantum simulation scheme devised in this tion, respectively, eliminating any significant un- work. The highlights of the scheme presented, and its desired coupling between the two resulting inter- promising applications, can be summarized as: acting Hamiltonians in the evolution given the Raman-beam detunings required. This feature B N sets of laser beams are used to address individ- does not demand the use of a strong effective mag- ual ions in a 1D chain. With the addition of three netic field to induce such nearest-neighbor interac- global laser beams, the Hamiltonian of a Heisen- tions [66, 67, 102], with its known limitations [103]. berg model can be engineered. Certain orientations Although it may be challenging to implement such and frequencies of the beams compared with each a scheme in larger chains of ions with low axial (i) (j) (i) (j) other (see Fig1) allow σx σx , σy σy , and normal-mode frequencies, ideas such as that pro- (i) (j) ⊗ ⊗ σz σz spin-spin interactions to be generated posed in Ref. [104] may allow a scalable scheme in with⊗ negligible couplings among different Raman future investigations. processes. Each set of lasers couples to one set of normal modes of motion (two transverse and one B Another feature of the proposed scheme is a high axial), allowing arbitrary spin-spin couplings to be degree of flexibility in tuning the spin-spin inter- engineered. Our scheme is inspired by that pre- action couplings of arbitrary forms along each axis sented in Ref. [63] but does not require an asym- of the qubit independently. This feature, which for metric trap in the transverse directions, as long as example is not offered in single Molmer-Sorenson one is interested in a Heisenberg XYZ and XXZ schemes [105], is shown to be particularly useful models (see the example of the Schwinger Model in for engineering the Hamiltonians of gauge theories Sec.III). considered in this work.

B The experimental scheme of this work offers the B The high level of control allows quantum simula- capability of engineering a range of interesting dy- tion of models in higher dimensions. Two interest- namics with a single beatnote frequency for each set ing examples of lattice gauge theories presented in of the lasers, denoted as µL with L = I,II,III, but this work (see Sec.IV) are Abelian Chern-Simons (i) theory coupled to matter, and a Z2 pure gauge the- with tunable phases and with Rabi frequencies ΩL at the location of each ion. Moreover, introducing ory, both in 2+1D, whose dynamic can be mapped a frequency control to the system, as is common to a planar Ising model with nearest-neighbor in- in the digital ion-trap platforms, allows arbitrary teractions. Such capability opens up the possibility spin-spin Hamiltonians to be engineered with un- of analog quantum simulations of systems beyond precedented accuracy. what has been possible to date.

B The frequency control allows an effective local mag- A few directions can be recognized as natural exten- netic field to be engineered via asymmetrically sions of the ideas presented in this paper. These include: 17

C There are a range of methods that lead to a Appendix A: Experimental specifications of the truncated angular-momentum representation of the trapped ion system considered for examples of this gauge degrees of freedom in LGTs, such as the work quantum link models [32, 34, 40], or the use of a tensor-network construction in Abelian gauge the- In order to provide explicit protocols in the examples ories coupled to matter [40, 106]. With the ma- provided in Sec.III and AppendixB, the ion-trap system nipulation of a larger number of internal levels of that is considered is assumed to share similar features as the ions, the approach advocated in this paper can those realized in Refs. [68–70]. Nonetheless, the general be applied to engineer interactions of spin systems procedure for obtaining these protocols can be identically with s 1 . An experimental realization of a spin ≥ 2 applied to systems containing other species of ions, and Hamiltonian with s = 1 is presented in Ref. [107], exhibiting different laser characteristics. and can be extended to allow quantum simulation Consider N 171Yb+ ions confined in a radio-frequency of select gauge theories in spin-1 representations. Paul trap [79]. The “qubit” in this system has been commonly encoded in a magnetically-insensitive clock state of 171Yb+. However, for the quantum simula- tions of the gauge theories considered in this study, For a wide range of phenomenologically-interesting C magnetically-sensitive hyperfine levels F = 0, mF = 0 lattice gauge theories for which a purely spin rep- | i and F = 1, mF = 1 will be needed, see Fig. 10. resentation does not exist, it is essential to extend | − i 1 1 The former (latter) level corresponds to sz = 2 ( 2 ) the toolkit of ion-trap analog simulation to leverage component of a quasi-spin operator. These are− split the control over phononic degrees of freedom. This in energy by a corresponding frequency ν0 ω0/2π = will require further technological advancement on 2 2 ≡ 12.642819 GHz + 310.8B0 Hz/G , where B0 denotes an the experimental front, as well as new proposals for external magnetic field [108]. Highly efficient state ini- engineering gauge and gauge-matter interactions in tialization and readout are performed using a laser tuned a highly controlled spin-phonon system. 2 to 369.5 nm, which strongly couples the ground S1/2 and 2 excited P1/2 states. For the Paul trap considered in the proposals of this work, νA = 0.713 MHz and νT = 4.1351 MHz, where νA and νT are the axial and transverse frequencies of the ACKNOWLEDGMENTS confining potential, respectively. The axial and trans- verse normal-mode frequencies in such a trap are tabu- lated in TableI for N = 4 and N = 8. Finally, to achieve We are grateful to Jiehang Zhang for his encouragement the values of Lamb-Dicke parameters used in the exam- during the early stages of this interdisciplinary collabo- ples presented in Sec.III and AppendixD, the lasers are ration. We acknowledge valuable discussions with Nor- aligned such that: ξ = 0.6960 and χ = 0.1767, where ξ bert Linke and Yannick Maurice. ZD is supported in and χ are introduced in the caption of Fig.1. As a result, part by the Maryland Center for Fundamental Physics, the angles between the individual beams and the three global beams (I), (II), and (III) are 88.21◦, 20.36◦, and University of Maryland, College Park. ZD and ASh are ◦ supported in part by the U.S. Department of Energy 88.21 , respectively. (DOE), Office of Science, Office of Advanced Scientific Computing Research (ASCR) Quantum Computing Ap- plication Teams program, under fieldwork proposal num- Appendix B: Tuning spin-dependent forces for the ber ERKJ347. ASh is further supported by the National proposed scheme of this work Science Foundation (NSF) through the Bridge to the Doctorate Fellowship. MH and ASe are supported by the NSF’s Physics Frontier Center at the Joint Quan- The Hamiltonian HIII in Eq. (5) is proportional to the (i) (i) tum Institute (JQI), and by the Air Force Office of Sci- operator α0I +α3σz . As was derived in Sec.II, the ef- (zz) entific Research, Multidisciplinary University Research fective spin-spin interactione H arise from [HIII , HIII ] Initiative (MURI). CM and GP are supported by the commutation at (η2) in the Magnus expansion of the O U.S. DOE Basic Energy Sciences (BES) Quantum Com- time-evolution operator. When α0 = 0, thise commuta-e puting in Chemical and Material Sciences Program, by (i) 6 tion creates an effective σz Hamiltonian with a strength the U.S. DOE High-Energy-Physics (HEP) Quantum In- (i) (i) twice that of the effective σ σ Hamiltonian. Such formation Science Enabled Discovery (QuantISED) Pro- z z a bias magnetic field introduces⊗ a significant error to gram, by the Army Research Office (ARO) MURI on the desired evolution. Any attempt to null out such a Modular Quantum Circuits, and by the NSF’s Physics local magnetic field with additional sets of lasers will Frontier Center at the JQI. cause further nonzero commutations with the HI and HII Hamiltonians, that are generally non-negligible given e e 1(COM) 2 3 4 5 6 7 8 1(COM)(COM 2) 3 4 5 6 7 8 (COM) n -58.837 -58.837 -58.837 - - - - - n -58.837 -58.837 -58.837 - - - - -

(1,1) (1,2) (1,3) (2,1) (2,2) (2,3) (3,1) (3,2) (1,1) (1,2) (1,3)5.598 (2,1) 0.269 (2,2) 2.511 (2,3) 8.010 (3,1) 1.700 (3,2) -6.371 12.565 -6.092 5.598 0.269 2.511 8.010 1.700 -6.371 12.565 -6.092 (ion,mode) (3,3) (4,1) (4,2) (4,3) (5,1) (5,2) (5,3) (6,1) (ion,mode) (3,3) (4,1)⌘⌦i,n (4,2)-2.467 (4,3) 42.683 (5,1) 4.632 (5,2) -8.182 (5,3) -13.841 (6,1) -22.936 12.472 -4.909 ⌘⌦i,n -2.467 42.683 4.632 -8.182 -13.841 -22.936 12.472 -4.909 (6,2) (6,3) (7,1) (7,2) (7,3) (8,1) (8,2) (8,3) (6,2) (6,3) (7,1)-13.534 (7,2) -12.167 (7,3) -1.508 (8,1) 4.581 (8,2) -13.710 (8,3) 0 0 0 -13.534 -12.167 -1.508 4.581 -13.710 0 0 0

01.500 1.250 1.000 0.750 0.500 0.249 0 01.500 1.250 11.000.500 0.750 0 0 1.500.250 0 1.249.000 0 0.750 0.500 0.249 0 1.500 0 1.2500 11.000.250 0 1.750.250 0.500 0 0 1.249.000 0 0.750 0.500 0.249 0 1 0 1.250 1.250 0 1.000 0.750 0.500 0.249 0 1 (ZZ) B 1.000 1.000 1.000 0 0.750 0.500 0.249 0 C B 1.000J 1.000 1.000B 0 0.750 0.500 0.249 0 C C J (ZZ) B B 0.750 0.750 0.750 0.750C 0 0.500 0.249 0 C B 0.750 0.750 0.750B 0.750 0 0.500 0.249 0 C C B B 0.500 0.500 0.500 0.500 0.C500 0 0.249 0 C B 0.500 0.500 0.500B 0.500 0.500 0 0.249 0 C C B B 0.249 0.249 0.249 0.249 0.C249 0.249 0 0 C B 0.249 0.249 0.249B 0.249 0.249 0.249 0 0 C C B B 00000000C C B 00000000B C C B @ C A @. A 18 . i, j i, j .j

TABLE I. Transverse normal modes of the motion of 4 (upper table) and 8 (lower table) ions in the Paul trap considered in this work. Frequencies are in kHz.

14 the strength of the bias magnetic field. It is there- tional mode’s frequency ωm, that is ∆ω = ωb ωr = ωm, fore important to investigate solutions that eliminate the see Fig. 10. In order to find appropriate polarizations− and term proportional to α0 in the native Hamiltonian in detuning that allow a pure σz Hamiltonian, three quanti- Eq. (5). One such solution relies on tuning the polariza- ties must be calculated in this scheme: i) the Stark shift tions and detuning of the Raman beams used to produce induced by red and blue lasers in the Raman pair, ii) the HIII Hamiltonian such that the spin-dependent force the spontaneous emission rate from excited states, and acting on the state is negative to that on the state : finally iii) the spin-dependent force on the qubit. (iii) |↑i |↓i F↑ =e F↓. This then sets α0 = 0, which is the choice must be studied to deduce the conditions under which − used in our proposal in Sec.II. To demonstrate this so- F↑ = F↓, while at the same time (i) must be ensured − lution, we consider the example of 171Yb+, however, the to vanish, and (ii) must be minimized. same approach can be taken to find schemes that work Let us denote the polarization of each beam by ˆr = for other ion traps as well. r−σˆ− + r0πˆ + r+σˆ+ and ˆb = b−σˆ− + b0πˆ + b+σˆ+, where 2 2 2 2 2 2 r− + r0 + r+ = b− + b0 + b+ = 1. In cal- As mentioned in AppendixA, the qubit is encoded culating| | | these| | quantities,| | | matrix| | elements| | in the form 0 0 0 in the magnetically-sensitive F = 0, mF = 0 α F m d ˆ αF mF need to be evaluated, where d is |↑i ≡ | 2 i h F | · | i and F = 1, mF = 1 hyperfine S1/2 states of the electric dipole operator, and α represents all other |↓i ≡ | − i 171Yb+. Consider a set of Raman beams with frequencies quantum numbers of the state besides the total spin F 2 ωr and ωb, detuned from P1/2 manifold by ∆. In order (nuclear spin added to electron’s total angular momen- to produce a spin-dependent force as discussed in Sec.II, tum) and its component along the quantization axis, mF . the beams have to be detuned from each other by the mo- Such a matrix element can be evaluated using [109]

0 0 5 0 0 0 0 J 0+I−mF0 0 J F I F 1 F 0 0 α F mF d.ˆ αF mF = ( 1) (2F + 1)(2F + 1)5 0 α J d αJ . (B1) h | | i − FJ 1 mF q m h || || i    − F  p

Here, q = 1 for theσ ˆ−-polarized light, q = 0 for the i) Stark shift: In the limit where ∆ γ, the Stark shift −  πˆ-polarized light, and q = 1 for theσ ˆ+-polarized light. for mS = , is given by [110] I and J denote the total nuclear spin and the electron’s | i |↑i |↓i 2 total angular momentum, respectively. “()” corresponds 1 mS d ˆjEj i δStark(mS) = | h | · | i | , (B3) Wigner’s 3j symbol while “ ” corresponds to Wigner’s 4 ∆i {} 0 0 j=r,b i 6j symbols. The reduced matrix element α J d αJ is X X related to the spontaneous emission rate γhbetween|| || statesi where ∆ is the detuning from the states that are vir- with J and J 0 quantum numbers for an atom coupled to i tually occupied, and E is the electric-field amplitude. free space: j Using Eq. (B1), the net Stark shift is found to be 0 0 2 0 α J d αJ = c0(2J + 1)γ, (B2) |h || || i| c0γωF δStark( ) δStark( ) = 2 where c0 is a number that depends on the transitions. ↑ − ↓ 12 ∆(∆ ωF ) 2 − For simplicity, in the following we assume that the P1/2 2 2 2 2 2 b− + r− b+ r+ . (B4) and the P3/2 states have the same c0 and γ. × | | | | − | | − | |   2 2 2 2 As is evident, by choosing b− + r− = b+ + r+ , the net shift can be set to zero.| | | | | | | |

14 Such a bias magnetic field term is discussed in Ref. [65]. ii) Spontaneous emission: The spontaneous emission rate 19

2 P3/2 ωF 2 Δ P1/2

ωb(r)

2 S1/2

FIG. 10. The level diagram of 171Yb+ relevant to the scheme presented in this Appendix.

can be evaluated using [110] and ∆ = (√2 1)ω , one find that − F ∗ ∗ ∗ γ(b r + b−r + b r ) Ω( ) = − 0 0 − + + , (B8) ↓ 12ωF ∗ √ ∗ √ ∗ d ˆ 2 γ( 2b0r0 + (2 + 3 2)b+r+ 3(2 + 2)b−r−) 1 PmS γi mS jEj i Ω( ) = − − . RSE = | h | · | i | , 4 ∆2 ↑ 24ωF i ↓ ↑ i X jX=r,b mSX= , (B9) (B5) In order to satisfy the condition Ω( ) = Ω( ) or in turn 15 ↓ − ↑ F↑ = F↓, a choice for the polarization vectors is − 3 3 where PmS is the probability of being in the mS ground ˆb = ( 1, 2 + , 1), (B10) state. Under the constraint that sets Eq. (B4) to zero, 2 √2 − s √2 − one finds that 3 3 ˆr = (1, 2 + , 1). (B11) 2 √2 s √2 −

2 2 2 Of course, these analytical solutions rely on the approx- c0γ (2 + r0 + b0 ) 1 2 imations that were made throughout these calculations, RSE = | | | | 2 + 2 . 12 (1 + r 2)(1 + b 2) ∆ (∆ ωF ) such as equal spontaneous emission rate from all the ex- | 0| | 0|  −  cited states considered. When precise values of the phys- p (B6) ical parameters in the system are input, the optimal val- ues for the parameters can still be evaluated numerically using the formalism outlined. See also Ref. [111] for a similar approach in achieving the condition F↑ = F↓. − As is seen, with the choice ∆ = (√2 1)ωF one is close to a local minimum of the spontaneous− emission rate.

Appendix C: Details of the laser-ion evolution operator iii) Spin-dependent force: Finally, the spin-dependent force can be found by considering the resonant two- In this Appendix, the explicit forms of the functions ap- photon Raman Rabi rate [110] peared in Eqs. (7-10) of the main text will be provided. The following frequency parameters are used:

T T T T ∆m µI + ωm, δm µI ωm, (C1) i(ϕb−ϕr ) ≡ ≡ − e mS d ˆrEr i i d ˆbEb mS ∆A µ + ωA , δA µ ωA , (C2) Ω(mS) = h | · | i h | · | i, m ≡ II m m ≡ II − m 4 ∆i T T T T i ∆ µIII + ω , δ µIII ω , (C3) X (B7) m ≡ m m ≡ − m where ϕ and ϕ are the phases of the red- and blue- while the rest of the parameters/functions are already r b e e detuned beams, respectively. With ∆ϕ ϕ ϕ = 0 defined in Sec.II. ≡ b − r 20

(i) (i) t (i) t t2 (x) ηI,mΩI T − T iB A − A α (t) = dt ei∆mt1 e iδmt1 + dt dt ei∆mt1 + e iδmt1 t t , (C4) i,m 2 1 − 2 2 1 − 1 ↔ 2 Z0 Z0 Z0  (i) (i)   h  i t (i) t t2 iη Ω A A iB T T (y) II,m II i∆mt1 −iδmt1 i∆mt1 −iδmt1 αi,m(t) = dt1 e + e dt2 dt1 e e t1 t2 , (C5) 2 0 − 2 0 0 − − ↔ Z   Z Z h  i (i) (i) t η Ω T T (z) III,m III i∆e mt1 −iδemt1 αi,m(t) = dt1 e e . (C6) 2 0 − Z  

(i) (i) (i) (i) t t2 η η Ω Ω A A T T (x) II,m III,n II III i∆mt2 −iδmt2 i∆e n t1 −iδen t1 βi,m,n(t) = dt2 dt1 e + e e e t1 t2 , (C7) 2 0 0 − − ↔ Z Z h    i (i) (i) (i) (i) t t2 iη η Ω Ω T T T T (y) I,m III,n I III i∆mt2 −iδmt2 i∆e n t1 −iδen t1 βi,m,n(t) = dt2 dt1 e e e e t1 t2 , (C8) − 4 0 0 − − − ↔ Z Z h    i (i) (i) (i) (i) t t2 η η Ω Ω T T A A (z) I,m II,n I II i∆mt2 −iδmt2 i∆n t1 −iδn t1 βi,m,n(t) = dt2 dt1 e e e + e t1 t2 . (C9) 4 0 0 − − ↔ Z Z h    i

iB(i) t γ(z)(t) = dt . (C10) i 4 1 Z0

N (i) (j) (i) (j) t t2 (x) ηI,mηI,mΩI ΩI T − T T − T χ (t) = dt dt ei∆mt2 e iδmt2 ei∆mt1 e iδmt1 , (C11) i,j 8 2 1 − − m=1 0 0 X Z Z h   i N (i) (j) (i) (j) t t2 (y) ηII,mηII,mΩII ΩII A − A A − A χ (t) = dt dt ei∆mt2 + e iδmt2 ei∆mt1 + e iδmt1 , (C12) i,j − 8 2 1 m=1 0 0 X Z Z h   i N (i) (j) (i) (j) t t2 (z) ηIII,mηIII,mΩIII ΩIII T − T T − T χ (t) = dt dt ei∆e mt2 e iδemt2 ei∆e mt1 e iδemt1 . (C13) i,j 8 2 1 − − m=1 0 0 X Z Z h   i

Appendix D: Engineered Hamiltonian of the Appendix E: Numerical evaluation of lasers-ions Schwinger model with N = 8 ions evolution

In order to confirm that the evolution of laser-ion sys- The multi-frequency, multi-amplitude scheme pre- tems in the scheme proposed in this work follows that sented at the end of Sec.III describes the engineering of a Heisenberg spin model with a magnetic field, the of the long-range Hamiltonian of the Schwinger model in exponent of the full evolution operator up to (η2, ηB) the 8 fermion-site theory, see Fig.7. The same optimiza- (see Eq. (7)) can be numerically evaluated forO each set tion procedure can be adopted to engineer the nearest- of laser beatnote and Rabi frequencies found. Here, we neighbor Hamiltonians in the same theory using sets of assume that the ions are in their motional ground state, laser beams that address transverse (for H(xx)) and ax- which can be achieved in current ion-trap experiments. ial (for H(yy)) normal modes of motion. The associated The results of this evaluation are plotted, respectively, results, as well as the required effective magnetic field in Figs. 14 and 15 for the case of the Schwinger-model that produces H(z), are depicted in Figs. 11-13 of this parameters with N = 4 and N = 8 that were studied in appendix. Associated numerical values are presented in Sec.III. These figures correspond to the evolution of the Supplemental Material. first ion in the chain and the results for the rest of the ions are included in Supplemental Material. To interpret these plots, note that the quantities that are plotted are contributions to the exponent of the evolution operator as a function of time t in millisecond (ms), and that:

15 Note that the spin-dependent force is related to the Rabi fre- -( a) plots in different colors the real and imag- quency via F = ∆k Ω(m ). mS S inary parts of all contributions arising from (i) ⌘⌦I,m/2⇡ [kHz]

(i) ⌘⌦m /2⇡ [kHz] 21

⌘⌦(i) /2⇡ [kHz] (xx) (xx) I,m0 H Ji,j /2⇡ [kHz]

ion index mode index

ion index j

ion index i !T /2⇡ m0 µI,m0 /2⇡ ��� ��� ��� ��� ��� ��� ��� (t) !m /2⇡ [MHz]

FIG. 11. The effective spin-spin coupling matrix J (xx) in Eq. (13) resulting from multiple pairs of Raman beams addressing !T /2⇡ N = 8 individual ions at the Rabi frequency Ω(i) , where i = 1, ··· , 8 and m0 = 1, ··· , 7. Them pairs of beams addressed at I,m0 µI,m /2⇡ T 0 T T ion i are detuned from the transverse COM mode by 7 different frequencies, µI,m = ω + fs(ω − ω ) with fs = 0.5, 0 m0 m0 m0+1 as denoted in the lower-right of the panel. The Lamb-Dicke parameter, η, multiplying the Rabi frequencies in the figure is p 2 T (xx) (xx) (i) η = (∆kI ) /4πMν ≈ 0.068. Here, the J matrix is tuned to produce H of the 8⌘⌦ fermion-sitem /2⇡ [kHz] Schwinger model in Eq. (19) with x = 6. Numerical values associated with this figure are provided in Supplemental Material.(i) ⌘⌦II,m/2⇡ [kHz]

⌘⌦(i) /2⇡ [kHz] (yy) (yy) II,m0 H Ji,j /2⇡ [kHz]

ion index mode index

ion index j

!A /2⇡ ion index i m0 µII,m0 /2⇡ � � � � � (t) !m /2⇡ [MHz]

FIG. 12. The effective spin-spin coupling matrix J (yy) in Eq. (14) resulting from multiple pairs of Raman beams addressing N = 8 individual ions at the Rabi frequency Ω(i) , where i = 1, ··· , 8 and m0 = 1, ··· , 7. The pairs of beams addressed at II,m0 A A A ion i are detuned from the axial COM mode by 7 different frequencies, µI,N−m +1 = ω + fs(ω − ω ) with 0 N−m0+1 N−m0 N−m0+1 fs = −0.5, as denoted in the lower-right of the panel. The Lamb-Dicke parameter, η, multiplying the Rabi frequencies in the p 2 A (yy) (yy) figure is η = (∆kII ) /4πMν ≈ 0.081. Here, the J matrix is tuned to produce H of the 8 fermion-site Schwinger model in Eq. (20) with x = 6. Numerical values associated with this figure are provided in Supplemental Material. 22

(z) (i) H BJzi,j/2⇡ [kHz]

ion index i

(z) FIG. 13. The effective magnetic field, Bz, that produces the H Hamiltonian of the Schwinger model, Eq. (22), for N = 8 and µ = 1. Numerical values associated with this figure are provided in Supplemental Material.

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

FIG. 14. Contributions to the exponent of the full laser-ion evolution operator up to and including O(η2, ηB) for laser parameters found in the single-frequency, multi-amplitude scheme in Sec.III to engineer the 4 fermion-site Schwinger Hamiltonian with x = 6 and µ = 1. The quantities plotted are enumerated in this Appendix and are dimensionless. The horizontal axis is time in ms. The plots shown correspond to the evolution of the first ion in the chain. The results for the rest of the ions can be found in Supplemental Material.

i t dt H0 (t ) with L = I,II,III acting on a state with phonon number n = 0, and ignoring − 0 1 L 1 ph R 23

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

FIG. 15. Contributions to the exponent of the full laser-ion evolution operator up to and including O(η2, ηB) for laser parameters found in the multi-frequency, multi-amplitude scheme in Sec.III to engineer the 8 fermion-site Schwinger Hamiltonian with x = 6 and µ = 1. The quantities plotted are enumerated in this Appendix and are dimensionless. The horizontal axis is time in ms. The plots shown correspond to the evolution of the first ion in the chain. The results for the rest of the ions can be found in Supplemental Material.

the (1) numerical factor arising from spin opera- arising from 1 t dt t2 dt ([H (t ), H (t )] + O − 2 0 2 0 1 I 2 II 1 tors acting on a general spin state. These are re- [H (t ), H (t )]), 1 t dt t2 dt ([H (t ), H II 2 I 1 R 2 R0 2 0 1 I 2 III ferred to as first-order terms, (η), elsewhere. − t t2 (t )] + [H (t ), H (t )]), ande 1e dt O 1 III 2 I R1 R 2 0 2 0 -( b), (c), and (d) plot in different colors e e −e e dt1([HII (t2), HIII (t1)] + [HIII (t2), HII (tR1)]),R re- the real and imaginary parts of all contribu- spectively,e acting one a state with nph = 0, and 1 t t2 tions arising from dt2 dt1[HI (t2), HI (t1)], ignoringe the e (1) numericale factore arising from − 2 0 0 O 1 t dt t2 dt [H (t ), H (t )], and 1 t dt spin operators acting on a general spin state. 2 0 2 0 1 II R2 IIR 1 2 0 2 −t2 − The small contributions observed show that the dt [H (t ), H (t )], respectively,e actinge on 0 R 1 IIIR 2 III 1 R choice of lasers’ detunings in our scheme leads to a state with n e= 0, ande ignoring the (1) nu- ph negligible commutations among the two sets of the Rmerical factor arising from spin operatorsO acting e e lasers. on a general spin state. As is seen, effective H(xx), H(yy), and H(zz) Hamiltonians originate from the -( h) plots in different colors the real and imaginary part of these contributions, signified by imaginary parts of all contributions aris- an almost exact linear dependence in time. 1 t t2 0 ing from 2 0 dt2 0 dt1([HB(t2),HL(t1)] + 0 − -( e), (f), and (g) plot in different colors the [HL(t2),HB(t1)]) for L = I,II,III acting on a real and imaginary parts of all contributions state with n =R 0, andR ignoring the (1) numeri- ph O 24

cal factor arising from spin operators acting on a i t dt H (t ) acting on a state with phonon − 0 1 B 1 general spin state. While these contributions are number nph = 0, and ignoring the (1) numeri- assured to remain a small fraction of the effective calR factor arising from spin operatorsO acting on a magnetic field desired, they are not bounded in general spin state. The real part of this contribu- time and couple to motional degrees of freedom. tion corresponds to the desired H(z) Hamiltonian. As a result, these contributions constitute the largest error to the desired effective-Hamiltonian Note that in the multi-frequency, multi-amplitude description that is engineered. scheme applied to the case of N = 8, the Hamiltonians in Eqs. (3-5) must be generalized as described in Sec.III (see discussions after Eq. (23)). The relation between the -( i) plots in different colors the real and imag- contributions enumerated and those given in Eqs. (7-10) inary parts of the contributions arising from and (C4-C13) is evident.

[1] M. Tanabashi et al. (Particle Data Group), “Review of [16] Hannes Bernien, Sylvain Schwartz, Alexander Keesling, Particle Physics,” Phys. Rev. D98, 030001 (2018). Harry Levine, Ahmed Omran, Hannes Pichler, Soon- [2] S. Aoki et al. (Flavour Lattice Averaging Group), won Choi, Alexander S Zibrov, Manuel Endres, Markus “FLAG Review 2019,” (2019), arXiv:1902.08191 [hep- Greiner, et al., “Probing many-body dynamics on a 51- lat]. atom quantum simulator,” Nature 551, 579 (2017). [3] M.A. Nielsen and I.L. Chuang, Quantum Computation [17] Sylvain de L´es´eleuc,Vincent Lienhard, Pascal Scholl, and Quantum Information, Cambridge Series on Infor- Daniel Barredo, Sebastian Weber, Nicolai Lang, mation and the Natural Sciences (Cambridge University Hans Peter B¨uchler, Thierry Lahaye, and Antoine Press, 2000). Browaeys, “Experimental realization of a symme- [4] Seth Lloyd, “Universal quantum simulators,” Science , try protected topological phase of interacting bosons 1073–1078 (1996). with rydberg atoms,” arXiv preprint arXiv:1810.13286 [5] Iulia M Georgescu, Sahel Ashhab, and Franco Nori, (2018). “Quantum simulation,” Reviews of Modern Physics 86, [18] Rainer Blatt and Christian F Roos, “Quantum simula- 153 (2014). tions with trapped ions,” Nature Physics 8, 277 (2012). [6] R. P. Feynman, “Simulating Physics with Computers,” [19] David J Wineland, C Monroe, Wayne M Itano, Dietrich International Journal of Theoretical Physics 21, 467– Leibfried, Brian E King, and Dawn M Meekhof, “Ex- 488 (1982). perimental issues in coherent quantum-state manipula- [7] John Preskill, “Quantum computing in the NISQ era tion of trapped atomic ions,” Journal of Research of the and beyond,” Quantum 2, 79 (2018). National Institute of Standards and Technology 103, [8] Dieter Jaksch, Christoph Bruder, Juan Ignacio Cirac, 259 (1998). Crispin W Gardiner, and Peter Zoller, “Cold bosonic [20] J Ignacio Cirac, Paolo Maraner, and Jiannis K Pachos, atoms in optical lattices,” Physical Review Letters 81, “Cold atom simulation of interacting relativistic quan- 3108 (1998). tum field theories,” Physical review letters 105, 190403 [9] Markus Greiner, Olaf Mandel, Tilman Esslinger, (2010). Theodor W H¨ansch, and Immanuel Bloch, “Quantum [21] O. Boada, A. Celi, J. I. Latorre, and M. Lewen- phase transition from a superfluid to a mott insulator stein, “Dirac Equation For Cold Atoms In Artificial in a gas of ultracold atoms,” nature 415, 39 (2002). Curved Spacetimes,” New J. Phys. 13, 035002 (2011), [10] Hendrik Weimer, Markus M¨uller,Igor Lesanovsky, Pe- arXiv:1010.1716 [cond-mat.quant-gas]. ter Zoller, and Hans Peter B¨uchler, “A rydberg quan- [22] HP B¨uchler, M Hermele, SD Huber, Matthew PA tum simulator,” Nature Physics 6, 382 (2010). Fisher, and P Zoller, “Atomic quantum simulator for [11] Maciej Lewenstein, Anna Sanpera, and Veronica lattice gauge theories and ring exchange models,” Phys- Ahufinger, Ultracold Atoms in Optical Lattices: Sim- ical review letters 95, 040402 (2005). ulating quantum many-body systems (Oxford University [23] Erez Zohar and Benni Reznik, “Confinement and lat- Press, 2012). tice QED electric flux-tubes simulated with ultra- [12] Immanuel Bloch, Jean Dalibard, and Sylvain Nascim- cold atoms,” Phys. Rev. Lett. 107, 275301 (2011), bene, “Quantum simulations with ultracold quantum arXiv:1108.1562 [quant-ph]. gases,” Nature Physics 8, 267 (2012). [24] Gergely Szirmai, Edina Szirmai, Alejandro Zamora, [13] J Ignacio Cirac and Peter Zoller, “Goals and opportu- and Maciej Lewenstein, “Gauge fields emerging from nities in quantum simulation,” Nature Physics 8, 264 time-reversal symmetry breaking for spin-5/2 fermions (2012). in a honeycomb lattice,” Physical Review A 84, 011611 [14] Christian Gross and Immanuel Bloch, “Quantum simu- (2011). lations with ultracold atoms in optical lattices,” Science [25] Haiyuan Zou, Yuzhi Liu, Chen-Yen Lai, J. Unmuth- 357, 995–1001 (2017). Yockey, A. Bazavov, Z. Y. Xie, T. Xiang, S. Chan- [15] Javier Arg¨uello-Luengo, Alejandro Gonz´alez-Tudela, drasekharan, S. W. Tsai, and Y. Meurice, “Progress Tao Shi, Peter Zoller, and J Ignacio Cirac, “Ana- towards quantum simulating the classical O(2) model,” log quantum chemistry simulation,” arXiv preprint Phys. Rev. A90, 063603 (2014), arXiv:1403.5238 [hep- arXiv:1807.09228 (2018). lat]. 25

[26] L. Tagliacozzo, A. Celi, P. Orland, and M. Lewen- a lattice by cold Bose gases,” Phys. Rev. D95, 094507 stein, “Simulations of non-Abelian gauge theories with (2017), arXiv:1605.00333 [cond-mat.quant-gas]. optical lattices,” Nature Commun. 4, 2615 (2013), [40] E. Rico, M. Dalmonte, P. Zoller, D. Banerjee, M. Bgli, arXiv:1211.2704 [cond-mat.quant-gas]. P. Stebler, and U. J. Wiese, “SO(3) ”Nuclear Physics” [27] Erez Zohar, J. Ignacio Cirac, and Benni Reznik, “Cold- with ultracold Gases,” Annals Phys. 393, 466–483 Atom Quantum Simulator for SU(2) Yang-Mills Lattice (2018), arXiv:1802.00022 [cond-mat.quant-gas]. Gauge Theory,” Phys. Rev. Lett. 110, 125304 (2013), [41] Jin Zhang, J. Unmuth-Yockey, J. Zeiher, A. Bazavov, arXiv:1211.2241 [quant-ph]. S. W. Tsai, and Y. Meurice, “Quantum simulation of [28] D. Banerjee, M. Dalmonte, M. Muller, E. Rico, P. Ste- the universal features of the Polyakov loop,” Phys. Rev. bler, U. J. Wiese, and P. Zoller, “Atomic Quan- Lett. 121, 223201 (2018), arXiv:1803.11166 [hep-lat]. tum Simulation of Dynamical Gauge Fields coupled to [42] Jean Dalibard, Fabrice Gerbier, Gediminas Juzeli¯unas, Fermionic Matter: From String Breaking to Evolution and Patrik Ohberg,¨ “Colloquium: Artificial gauge po- after a Quench,” Phys. Rev. Lett. 109, 175302 (2012), tentials for neutral atoms,” Reviews of Modern Physics arXiv:1205.6366 [cond-mat.quant-gas]. 83, 1523 (2011). [29] D. Banerjee, M. Bgli, M. Dalmonte, E. Rico, P. Ste- [43] Marco Mancini, Guido Pagano, Giacomo Cappellini, bler, U. J. Wiese, and P. Zoller, “Atomic Quantum Lorenzo Livi, Marie Rider, Jacopo Catani, Carlo Sias, Simulation of U(N) and SU(N) Non-Abelian Lattice Peter Zoller, Massimo Inguscio, Marcello Dalmonte, Gauge Theories,” Phys. Rev. Lett. 110, 125303 (2013), et al., “Observation of chiral edge states with neutral arXiv:1211.2242 [cond-mat.quant-gas]. fermions in synthetic hall ribbons,” Science 349, 1510– [30] Erez Zohar, J. Ignacio Cirac, and Benni Reznik, “Simu- 1513 (2015). lating (2+1)-Dimensional Lattice QED with Dynamical [44] BK Stuhl, H-I Lu, LM Aycock, D Genkina, and Matter Using Ultracold Atoms,” Phys. Rev. Lett. 110, IB Spielman, “Visualizing edge states with an atomic 055302 (2013), arXiv:1208.4299 [quant-ph]. bose gas in the quantum hall regime,” Science 349, [31] Erez Zohar, J. Ignacio Cirac, and Benni Reznik, 1514–1518 (2015). “Quantum simulations of gauge theories with ultra- [45] M. Aidelsburger, S. Nascimbene, and N. Goldman, cold atoms: local gauge invariance from angular mo- “Artifcial gauge felds in materials and engineered sys- mentum conservation,” Phys. Rev. A88, 023617 (2013), tems,” Comptes Rendus Physique 19, 394–432 (2018), arXiv:1303.5040 [quant-ph]. arXiv:1710.00851 [cond-mat.mes-hall]. [32] Uwe-Jens Wiese, “Ultracold Quantum Gases and [46] Logan W Clark, Brandon M Anderson, Lei Feng, Anita Lattice Systems: Quantum Simulation of Lattice Gaj, Kathryn Levin, and Cheng Chin, “Observation Gauge Theories,” Annalen Phys. 525, 777–796 (2013), of density-dependent gauge fields in a bose-einstein arXiv:1305.1602 [quant-ph]. condensate based on micromotion control in a shaken [33] E. Rico, T. Pichler, M. Dalmonte, P. Zoller, and two-dimensional lattice,” Physical review letters 121, S. Montangero, “Tensor networks for Lattice Gauge 030402 (2018). Theories and Atomic Quantum Simulation,” Phys. [47] Christian Schweizer, Fabian Grusdt, Moritz Berngru- Rev. Lett. 112, 201601 (2014), arXiv:1312.3127 [cond- ber, Luca Barbiero, Eugene Demler, Nathan Gold- mat.quant-gas]. man, Immanuel Bloch, and Monika Aidelsburger, [34] Uwe-Jens Wiese, “Towards Quantum Simulating “Floquet approach to Z2 lattice gauge theories with QCD,” Proceedings, 24th International Conference on ultracold atoms in optical lattices,” arXiv preprint Ultra-Relativistic Nucleus-Nucleus Collisions (Quark arXiv:1901.07103 (2019). Matter 2014): Darmstadt, Germany, May 19-24, 2014, [48] Juan I Cirac and Peter Zoller, “Quantum computations Nucl. Phys. A931, 246–256 (2014), arXiv:1409.7414 with cold trapped ions,” Physical review letters 74, 4091 [hep-th]. (1995). [35] Erez Zohar, J. Ignacio Cirac, and Benni Reznik, “Quan- [49] Klaus Mølmer and Anders Sørensen, “Multiparticle en- tum Simulations of Lattice Gauge Theories using Ultra- tanglement of hot trapped ions,” Physical Review Let- cold Atoms in Optical Lattices,” Rept. Prog. Phys. 79, ters 82, 1835 (1999). 014401 (2016), arXiv:1503.02312 [quant-ph]. [50] E Solano, RL de Matos Filho, and N Zagury, “Deter- [36] M. Dalmonte and S. Montangero, “Lattice gauge theory ministic bell states and measurement of the motional simulations in the quantum information era,” Contemp. state of two trapped ions,” Physical Review A 59, R2539 Phys. 57, 388–412 (2016), arXiv:1602.03776 [cond- (1999). mat.quant-gas]. [51] GJ Milburn, S Schneider, and DFV James, “Ion trap [37] Simone Notarnicola, Elisa Ercolessi, Paolo Facchi, quantum computing with warm ions,” Fortschritte der Giuseppe Marmo, Saverio Pascazio, and Francesco V. Physik: Progress of Physics 48, 801–810 (2000). Pepe, “Discrete Abelian Gauge Theories for Quantum [52] Rainer Blatt and David Wineland, “Entangled states of Simulations of QED,” J. Phys. A48, 30FT01 (2015), trapped atomic ions,” Nature 453, 1008 (2008). arXiv:1503.04340 [quant-ph]. [53] Ben P Lanyon, Cornelius Hempel, Daniel Nigg, Markus [38] V. Kasper, F. Hebenstreit, F. Jendrzejewski, M. K. M¨uller,Rene Gerritsma, F Z¨ahringer,Philipp Schindler, Oberthaler, and J. Berges, “Implementing quantum Julio T Barreiro, Markus Rambach, Gerhard Kirch- electrodynamics with ultracold atomic systems,” New mair, et al., “Universal digital quantum simulation with J. Phys. 19, 023030 (2017), arXiv:1608.03480 [cond- trapped ions,” Science 334, 57–61 (2011). mat.quant-gas]. [54] Christopher Monroe and Jungsang Kim, “Scaling the [39] Yoshihito Kuno, Shinya Sakane, Kenichi Kasamatsu, ion trap quantum processor,” Science 339, 1164–1169 Ikuo Ichinose, and Tetsuo Matsui, “Quantum simula- (2013). tion of (1 + 1)-dimensional U(1) gauge-Higgs model on 26

[55] C Monroe, R Raussendorf, A Ruthven, KR Brown, crystal,” Nature 543, 217 (2017). P Maunz, L-M Duan, and J Kim, “Large-scale modu- [69] PW Hess, P Becker, HB Kaplan, A Kyprianidis, lar quantum-computer architecture with atomic mem- AC Lee, B Neyenhuis, G Pagano, P Richerme, C Senko, ory and photonic interconnects,” Physical Review A 89, J Smith, et al., “Non-thermalization in trapped atomic 022317 (2014). ion spin chains,” Philosophical Transactions of the [56] E. A. Martinez et al., “Real-time dynamics of lattice Royal Society A: Mathematical, Physical and Engineer- gauge theories with a few-qubit quantum computer,” ing Sciences 375, 20170107 (2017). Nature 534, 516–519 (2016), arXiv:1605.04570 [quant- [70] Brian Neyenhuis, Jiehang Zhang, Paul W Hess, Jacob ph]. Smith, Aaron C Lee, Phil Richerme, Zhe-Xuan Gong, [57] Shantanu Debnath, Norbert M Linke, Caroline Fig- Alexey V Gorshkov, and Christopher Monroe, “Ob- gatt, Kevin A Landsman, Kevin Wright, and Christo- servation of prethermalization in long-range interacting pher Monroe, “Demonstration of a small programmable spin chains,” Science advances 3, e1700672 (2017). quantum computer with atomic qubits,” Nature 536, [71] Fangli Liu, Rex Lundgren, Paraj Titum, Guido Pagano, 63 (2016). Jiehang Zhang, Christopher Monroe, and Alexey V. [58] Norbert M Linke, Dmitri Maslov, Martin Roetteler, Gorshkov, “Confined quasiparticle dynamics in long- Shantanu Debnath, Caroline Figgatt, Kevin A Lands- range interacting quantum spin chains,” Phys. Rev. man, Kenneth Wright, and Christopher Monroe, “Ex- Lett. 122, 150601 (2019). perimental comparison of two quantum computing ar- [72] Lucas Lamata, Juan Le´on, Tobias Sch¨atz, and chitectures,” Proceedings of the National Academy of E Solano, “Dirac equation and quantum relativistic ef- Sciences 114, 3305–3310 (2017). fects in a single trapped ion,” Physical review letters [59] C. Figgatt, A. Ostrander, N. M. Linke, K. A. Landsman, 98, 253005 (2007). D. Zhu, D. Maslov, and C. Monroe, “Parallel entangling [73] Rene Gerritsma, Gerhard Kirchmair, Florian Za- operations on a universal ion-trap quantum computer,” ehringer, E Solano, R Blatt, and CF Roos, “Quan- Nature (2019), 10.1038/s41586-019-1427-5. tum simulation of the dirac equation,” Nature 463, 68 [60] Kevin A Landsman, Caroline Figgatt, Thomas Schus- (2010). ter, Norbert M Linke, Beni Yoshida, Norm Y Yao, [74] Jorge Casanova, Juan Jos´eGarc´ıa-Ripoll, Rene Ger- and Christopher Monroe, “Verified quantum informa- ritsma, Christian F Roos, and Enrique Solano, “Klein tion scrambling,” Nature 567, 61 (2019). tunneling and dirac potentials in trapped ions,” Physi- [61] Yunseong Nam, Jwo-Sy Chen, Neal C Pisenti, Ken- cal Review A 82, 020101 (2010). neth Wright, Conor Delaney, Dmitri Maslov, Kenneth R [75] J. Casanova, L. Lamata, I. L. Egusquiza, R. Ger- Brown, Stewart Allen, Jason M Amini, Joel Apisdorf, ritsma, C. F. Roos, J. J. Garcia-Ripoll, and E. Solano, et al., “Ground-state energy estimation of the water “Quantum Simulation of Quantum Field Theories in molecule on a trapped ion quantum computer,” arXiv Trapped Ions,” Phys. Rev. Lett. 107, 260501 (2011), preprint arXiv:1902.10171 (2019). arXiv:1107.5233 [quant-ph]. [62] K Wright, KM Beck, S Debnath, JM Amini, Y Nam, [76] Philipp Hauke, David Marcos, Marcello Dalmonte, N Grzesiak, J-S Chen, NC Pisenti, M Chmielewski, and Peter Zoller, “Quantum simulation of a lattice C Collins, et al., “Benchmarking an 11-qubit quantum Schwinger model in a chain of trapped ions,” Phys. Rev. computer,” arXiv preprint arXiv:1903.08181 (2019). X3, 041018 (2013), arXiv:1306.2162 [cond-mat.quant- [63] Diego Porras and J Ignacio Cirac, “Effective quantum gas]. spin systems with trapped ions,” Physical review letters [77] Masuo Suzuki, “Generalized trotter’s formula and sys- 92, 207901 (2004). tematic approximants of exponential operators and in- [64] R Islam, EE Edwards, K Kim, S Korenblit, C Noh, ner derivations with applications to many-body prob- H Carmichael, G-D Lin, L-M Duan, C-C Joseph Wang, lems,” Communications in Mathematical Physics 51, JK Freericks, et al., “Onset of a quantum phase tran- 183–190 (1976). sition with a trapped ion quantum simulator,” Nature [78] Simcha Korenblit, Dvir Kafri, Wess C Campbell, Ra- communications 2, 377 (2011). jibul Islam, Emily E Edwards, Zhe-Xuan Gong, Guin- [65] Ch Schneider, Diego Porras, and Tobias Schaetz, “Ex- Dar Lin, Lu-Ming Duan, Jungsang Kim, Kihwan Kim, perimental quantum simulations of many-body physics et al., “Quantum simulation of spin models on an arbi- with trapped ions,” Reports on Progress in Physics 75, trary lattice with trapped ions,” New Journal of Physics 024401 (2012). 14, 095024 (2012). [66] Petar Jurcevic, Ben P Lanyon, Philipp Hauke, Cor- [79] Wolfgang Paul, “Electromagnetic traps for charged and nelius Hempel, Peter Zoller, Rainer Blatt, and Chris- neutral particles,” Reviews of modern physics 62, 531 tian F Roos, “Quasiparticle engineering and entangle- (1990). ment propagation in a quantum many-body system,” [80] Shi-Liang Zhu, Chris Monroe, and L-M Duan, Nature 511, 202 (2014). “Trapped ion quantum computation with transverse [67] Philip Richerme, Zhe-Xuan Gong, Aaron Lee, Crys- phonon modes,” Physical review letters 97, 050505 tal Senko, Jacob Smith, Michael Foss-Feig, Spyridon (2006). Michalakis, Alexey V Gorshkov, and Christopher Mon- [81] Shi-Liang Zhu, Christopher Monroe, and L-M Duan, roe, “Non-local propagation of correlations in quantum “Arbitrary-speed quantum gates within large ion crys- systems with long-range interactions,” Nature 511, 198 tals through minimum control of laser beams,” EPL (2014). (Europhysics Letters) 73, 485 (2006). [68] J Zhang, PW Hess, A Kyprianidis, P Becker, A Lee, [82] Crispin Gardiner and Peter Zoller, The Quan- J Smith, G Pagano, I-D Potirniche, Andrew C Potter, tum World of Ultra-Cold Atoms and Light A Vishwanath, et al., “Observation of a discrete time Book II: The Physics of Quantum-Optical De- 27

vices (IMPERIAL COLLEGE PRESS, 2015) [98] F. J. Wegner, “Duality in Generalized Ising Models and https://www.worldscientific.com/doi/pdf/10.1142/p983. Phase Transitions Without Local Order Parameters,”J. [83] G-D Lin, S-L Zhu, Rajibul Islam, Kihwan Kim, M-S Math. Phys. 12, 2259–2272 (1971). Chang, Simcha Korenblit, Christopher Monroe, and [99] Manu Mathur and TP Sreeraj, “Lattice gauge theories L-M Duan, “Large-scale quantum computation in an and spin models,” Physical Review D 94, 085029 (2016). anharmonic linear ion trap,” EPL (Europhysics Letters) [100] G De las Cuevas, Wolfgang Duer, Hans J Briegel, and 86, 60004 (2009). Miguel Angel Martin-Delgado, “Unifying all classical [84] G Pagano, PW Hess, HB Kaplan, WL Tan, P Richerme, spin models in a lattice gauge theory,” Physical review P Becker, A Kyprianidis, J Zhang, E Birckelbaw, letters 102, 230502 (2009). MR Hernandez, et al., “Cryogenic trapped-ion system [101] G De las Cuevas, W D¨ur,HJ Briegel, and MA Martin- for large scale quantum simulation,” Quantum Science Delgado, “Mapping all classical spin models to a lat- and Technology 4, 014004 (2018). tice gauge theory,” New Journal of Physics 12, 043014 [85] John Kogut and Leonard Susskind, “Hamiltonian for- (2010). mulation of wilson’s lattice gauge theories,” Physical [102] Michael L Wall, Arghavan Safavi-Naini, and Ana Maria Review D 11, 395 (1975). Rey, “Boson-mediated quantum spin simulators in [86] Tom Banks, Leonard Susskind, and John Kogut, transverse fields: X y model and spin-boson entangle- “Strong-coupling calculations of lattice gauge theo- ment,” Physical Review A 95, 013602 (2017). ries:(1+ 1)-dimensional exercises,” Physical Review D [103] Thomas G Kiely and JK Freericks, “Relationship be- 13, 1043 (1976). tween the transverse-field ising model and the x y model [87] C. J. Hamer, Wei-hong Zheng, and J. Oitmaa, “Series via the rotating-wave approximation,” Physical Review expansions for the massive Schwinger model in Hamil- A 97, 023611 (2018). tonian lattice theory,” Phys. Rev. D56, 55–67 (1997), [104] Kenton R Brown, Christian Ospelkaus, Yves Colombe, arXiv:hep-lat/9701015 [hep-lat]. Andrew C Wilson, D Leibfried, and David J Wineland, [88] Christine Muschik, Markus Heyl, Esteban Martinez, “Coupled quantized mechanical oscillators,” Nature Thomas Monz, Philipp Schindler, Berit Vogell, Mar- 471, 196 (2011). cello Dalmonte, Philipp Hauke, Rainer Blatt, and Pe- [105] Alejandro Berm´udez, L Tagliacozzo, Germ´anSierra, ter Zoller, “U (1) wilson lattice gauge theories in dig- and P Richerme, “Long-range heisenberg models in ital quantum simulators,” New Journal of Physics 19, quasiperiodically driven crystals of trapped ions,” Phys- 103020 (2017). ical Review B 95, 024431 (2017). [89] Eduardo Fradkin, Field Theories of Condensed Matter [106] Alexei Bazavov, Chen-Yen Lai, Shan-Wen Tsai, and Physics, 2nd ed. (Cambridge University Press, 2013). Yannick Meurice, “The effective U(1)-Higgs theory at [90] Eduardo Fradkin, “Jordan-wigner transformation for strong couplings on optical lattices,” Proceedings, 32nd quantum-spin systems in two dimensions and fractional International Symposium on Lattice Field Theory (Lat- statistics,” Physical review letters 63, 322 (1989). tice 2014): Brookhaven, NY, USA, June 23-28, 2014, [91] Kai Sun, Krishna Kumar, and Eduardo Fradkin, “Dis- PoS LATTICE2014, 391 (2014), arXiv:1411.4989 cretized abelian chern-simons gauge theory on arbitrary [hep-lat]. graphs,” Physical Review B 92, 115148 (2015). [107] C Senko, P Richerme, J Smith, A Lee, I Cohen, A Ret- [92] Jianlong Fu, “Properties and application of so (3) ma- zker, and C Monroe, “Realization of a quantum integer- jorana representation of spin: Equivalence with jordan- spin chain with controllable interactions,” Physical Re- wigner transformation and exact z 2 gauge theories for view X 5, 021026 (2015). spin models,” Physical Review B 98, 214432 (2018). [108] Patrick Gill, “When should we change the definition of [93] Roman Jackiw and Erick J Weinberg, “Self-dual chern- the second?” Philosophical Transactions of the Royal simons vortices,” Physical Review Letters 64, 2234 Society A: Mathematical, Physical and Engineering Sci- (1990). ences 369, 4109–4130 (2011). [94] M Cristina Diamantini, P Sodano, and CA Trugen- [109] Jonathan Albert Mizrahi, Ultrafast control of spin and berger, “Topological excitations in compact maxwell- motion in trapped ions, Ph.D. thesis (2013). chern-simons theory,” Physical review letters 71, 1969 [110] David J Wineland, M Barrett, J Britton, J Chiaverini, (1993). B DeMarco, Wayne M Itano, B Jelenkovi´c,Christopher [95] Stephan Caspar, David Mesterhazy, Therkel Z Olesen, Langer, D Leibfried, V Meyer, et al., “Quantum in- Nadiia Dmytrivna Vlasii, and U-J Wiese, “Doubled formation processing with trapped ions,” Philosophical lattice chern–simons–yang–mills theories with discrete Transactions of the Royal Society of London. Series A: gauge group,” Annals of physics 374, 255–290 (2016). Mathematical, Physical and Engineering Sciences 361, [96] Luca Barbiero, Christian Schweizer, Monika Aidels- 1349–1361 (2003). burger, Eugene Demler, Nathan Goldman, and Fabian [111] Joseph W Britton, Brian C Sawyer, Adam C Keith, Grusdt, “Coupling ultracold matter to dynamical gauge C-C Joseph Wang, James K Freericks, Hermann Uys, fields in optical lattices: From flux-attachment to Z2 Michael J Biercuk, and John J Bollinger, “Engi- lattice gauge theories,” (2018), arXiv:1810.02777 [cond- neered two-dimensional ising interactions in a trapped- mat.quant-gas]. ion quantum simulator with hundreds of spins,” Nature [97] John B Kogut, “An introduction to lattice gauge theory 484, 489 (2012). and spin systems,” Reviews of Modern Physics 51, 659 (1979).