betaFeedback

WorksPhysics and Astronomy

Exploring Light-Cone Distribution Amplitudes from Quantum Computing

Abstract

Light-cone distribution amplitudes (LCDAs) are essential nonperturbative quantities for theoretical predictions of exclusive high-energy processes in quantum chromodynamics (QCD). We demonstrate the prospect of calculating LCDAs on a quantum computer by applying a recently proposed quantum algorithm, with staggered fermions, to the simulation of the LCDA in the (1+1)-dimensional Nambu-Jona-Lasinio (NJL) model on classical hardware. The agreement between the result from the classical simulation of the quantum algorithm and that from exact diagonalization justifies the proposed quantum algorithm. We find that the resulting LCDA in the NJL model exhibits features shared with the LCDAs obtained from QCD.

Introduction

Light-cone distribution amplitudes (LCDAs) are quantities that describe the nonperturbative aspects of bound states in high-energy exclusive processes in quantum chromodynamics (QCD). They are complementary to the parton distribution functions (PDFs), which are nonperturbative quantities for inclusive QCD processes with incoming hadrons, such as deep-inelastic scattering of the proton at the Hadron-Electron Ring Accelerator (HERA) and future Electron-Ion Colliders [1, 2, 3, 4]. In an exclusive high-energy QCD process, up to corrections suppressed by inverse powers of a large energy scale, the scattering amplitude can be factorized into the convolution of a perturbative Wilson coefficient and a nonperturbative amplitude, the LCDA [5, 6]. A seminal example is the electromagnetic form factor F(Q2)F(Q^{2}) for the process γ∗γ→qqˉ→π0\gamma^{*}\gamma\to q\bar{q}\to\pi^{0} at large momentum transfer QQ, for which the factorization reads

F(Q2)=fπ∫01dx TH(x,Q2;μ)ϕπ(x;μ)+O(ΛQCD2/Q2),(1)F(Q^{2})=f_{\pi}\int_{0}^{1}dx\,T_{H}(x,Q^{2};\mu)\phi_{\pi}(x;\mu)+\mathcal{O}(\Lambda_{\mathrm{QCD}}^{2}/Q^{2}), \tag*{(1)}

where Q2Q^{2} is the 4-momentum squared carried by the virtual photon γ∗\gamma^{*}, ΛQCD\Lambda_{\mathrm{QCD}} is the energy scale below which QCD becomes nonperturbative, and μ\mu is the factorization scale, which separates the short-distance physics from the long-distance wave function ϕπ\phi_{\pi}. Here, TH(x,Q2;μ)T_{H}(x,Q^{2};\mu) is the hard kernel, which describes the production of a quark-antiquark pair by short-distance dynamics. TH(x,Q2;μ)T_{H}(x,Q^{2};\mu) is perturbatively calculable as an expansion in the strong coupling constant αs\alpha_{s}. The wave function ϕπ(x,μ)\phi_{\pi}(x,\mu) and the factor fπf_{\pi} are the LCDA and the decay constant of the neutral pion, respectively. They together encode the hadronization of a quark-antiquark pair into a pion, and are nonperturbative as they are sensitive to the long-distance dynamics of QCD. The LCDA of a meson can be viewed as the probability amplitude to find the valence qqˉq\bar{q} Fock state in which the quark qq and the antiquark qˉ\bar{q} carry respectively the momentum fraction xx and 1−x1-x of the highly boosted meson. The decay constant is defined as the overall normalization of the LCDA. The LCDAs and decay constants for baryons can be likewise defined. The LCDAs, being the essential ingredients for reliable predictions for exclusive QCD processes, have been studied intensively in various directions. Early studies include their perturbative evolution with the scale μ\mu and their asymptotic behaviors [5, 6, 7]. Estimations of the LCDAs using sum rules, illustrative models, [8, 9, 10, 11, 12, 13, 14, 15, 16, 17], light-front quantization [18, 19, 20, 21], as well as the refactorization in the nonrelativistic expansion [22, 23] have also been discussed under various circumstances in the literature. Evaluations of the LCDAs in lattice QCD were initially performed by taking the moments [24, 25, 26, 27, 28, 29], and later with direct calculations in the momentum fraction xx within the large momentum effective theory (LaMET) framework [30, 31, 32, 33]. There are two major obstacles to acquiring knowledge about the LCDAs, the first being the elusiveness of the relevant experimental data that hinders QCD global analyses, the second being the nature of real-time evolution on the light cone involved in the definition of LCDAs, which is not amenable to direct evaluations using Euclidean lattice QCD [34].

Stimulated by the promising prospect of quantum computing [35], there has been a rapidly growing wave of research on applications of quantum computing in elementary particle physics [36, 37, 38]. Early theoretical attempts had showed that quantum computation costs polynomial time in simulations of real time dynamics in quantum field theory [39, 40, 41, 42]. Validity of quantum computing in various problems in particle physics have been studied by proposals of quantum algorithms as well as simulations with real quantum computers or classical hardware. These studies include evaluations of nonperturbative quantities [43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53], simulations of real-time processes [54, 55, 56, 57, 58, 59, 60, 61, 62], as well as evaluations of thermodynamical quantities at finite chemical potential [63, 64].

Recently, a quantum algorithm was proposed for both the preparation of a hadron state and the evaluation of real-time light-like correlators in Ref. [51]. The algorithm was demonstrated feasible by evaluating directly the parton distribution function in the (1+1)-dimensional Nambu-Jona-Lasinio (NJL) model with staggered fermions on classical hardware. The results obtained from the quantum algorithm were checked against exact diagonalization to find full agreements, which justifies the validity of the algorithm and suggests the possibility of evaluating the hadron parton distributions in QCD by quantum computation.

In this study, we extend the previous studies to apply the quantum algorithm to study the LCDA in the (1+1)D 1-flavor NJL model, using staggered fermions. In Sec. II, we provide the operator definition of the LCDA in the (1+1)D NJL model. Then we present the quantum algorithm for both the hadronic state preparation and the direct computation of the quark-antiquark correlator in Sec. III. By implementing the proposed algorithm on classical hardware, we found good consistency between the results obtained from the quantum algorithm and exact diagonalization. The final results for the LCDA are presented in Sec. IV. We give a summary in Sec. V.

Light-cone distribution amplitude and the NJL model

The LCDA of a meson hh is defined as

ϕh(x)=1fh∫dz e−i(x−1)n⋅Pz×⟨Ω∣ψˉ(zn)γ+ψ(0)∣h(P)⟩(2)\phi_h(x) = \frac{1}{f_h} \int dz\,e^{-i(x-1)n\cdot Pz} \times\langle\Omega|\bar{\psi}(zn)\gamma^+\psi(0)|h(P)\rangle \tag*{(2)}

where PP is the momentum of the meson and nn is a light-like vector defined by n=(1,−n^)n=(1,-\hat{n}), with n^\hat{n} being a spatial unit vector along the direction of motion of the meson. The plus-component of the gamma matrix γμ\gamma^\mu in light-cone coordinates is denoted by γ+\gamma^+, i.e. γ+=n⋅γ\gamma^+=n\cdot\gamma. The prefactor fhf_h is the decay constant, defined such that ∫01dx ϕh(x)=1\int_0^1 dx\,\phi_h(x)=1. The matrix element in the second line in Eq. (2) describes the transition amplitude from the vacuum ∣Ω⟩|\Omega\rangle to the hadron state ∣h(P)⟩|h(P)\rangle via insertion of a quark-antiquark-pair operator. The W(zn,0)W(zn,0) is the Wilson line (gauge link) on the light cone,

W(zn,0)=Pexp⁡(ig∫0zdz′ Aa+(z′n)ta),(3)W(zn,0)=\mathcal{P}\exp\left(ig\int_0^z dz'\,A_a^+(z'n)t_a\right), \tag*{(3)}

where tat_a are the SU(3) fundamental generators, P\mathcal{P} denotes path-ordering, and Aa+=n⋅AaA_a^+=n\cdot A_a is the plus-component of the SU(3) gauge potential. Generally, to simulate the LCDA on a quantum computer, one has to: 1. prepare the vacuum state ∣Ω⟩|\Omega\rangle and the hadronic state ∣h(P)⟩|h(P)\rangle on the quantum computer; 2. simulate the Wilson line on the quantum computer; 3. evaluate the matrix element ⟨Ω∣O∣h⟩\langle\Omega|{\cal O}|h\rangle. We will elaborate on the first and third steps later in Section III. We note that the presence of the gauge field in a gauge theory will dramatically increase the demand for quantum resources. However, as noted in Ref. [51], simulating the Wilson line in the second step does not introduce substantial complexity even though the Wilson line is non-local, which is contrary to the claim in Ref. [45].

Since simulating QCD on a quantum computer remains a formidable task to date [65], we resort to a simple model, the Nambu-Jona-Lasinio (NJL) model [66, 67] in (1+1) dimensions, also known as the Gross-Neveu model [68], in order to demonstrate simulations of the LCDA on a quantum computer. The Lagrangian of the (1+1)D NJL model is given by

L=ψˉ(iγμ∂μ−mq)ψ+g(ψˉψ)2(4)\mathcal{L}=\bar{\psi}(i\gamma^\mu\partial_\mu-m_q)\psi+g(\bar{\psi}\psi)^2 \tag*{(4)}

where gg is the strong coupling constant and mqm_q is the quark mass. The LCDA ϕh(x)\phi_h(x) of a meson hh in the NJL model is defined by

ϕh(x)=1fh∫dz e−i(x−1)n⋅Pz×⟨Ω∣ψˉ(zn)γ+ψ(0)∣h(P)⟩.(5)\phi_h(x)=\frac{1}{f_h}\int dz\,e^{-i(x-1)n\cdot Pz} \times\langle\Omega|\bar{\psi}(zn)\gamma^+\psi(0)|h(P)\rangle. \tag*{(5)}

We note that the LCDA defined as in eq. (5) is independent of the frame of reference. For the sake of practical computation, we will evaluate the LCDA in the rest frame of the meson, in which case we have

ϕh(x)=1fh∫dz e−i(x−1)mhz×⟨Ω∣eiHtψˉ(0,−z)e−iHtγ+ψ(0,0)∣h⟩(6)\phi_h(x)=\frac{1}{f_h}\int dz\,e^{-i(x-1)m_hz}\times\langle\Omega|e^{iHt}\bar{\psi}(0,-z)e^{-iHt}\gamma^+\psi(0,0)|h\rangle \tag*{(6)}

where we have written the quark field ψˉ(zn)\bar{\psi}(zn) as ψˉ(zn)=eiHzψˉ(0,−z)e−iHz\bar{\psi}(zn)=e^{iHz}\bar{\psi}(0,-z)e^{-iHz} with HH the Hamiltonian of the NJL model, and we will set t=zt=z in eq. (6) to put the correlator on the light cone. Here mhm_h is the mass of the meson hh.

To facilitate quantum simulations, we discretize the space into N/2N/2 lattice sites and place the fermion field on the lattice following:

ψ(0,z)=(ψ1(0,z)ψ2(0,z))=(φ2nφ2n+1),(7)\psi(0,z)=\begin{pmatrix}\psi_1(0,z)\\\psi_2(0,z)\end{pmatrix}=\begin{pmatrix}\varphi_{2n}\\\varphi_{2n+1}\end{pmatrix}, \tag*{(7)}

where 0≤n≤N2−10\le n\le\frac{N}{2}-1. Notice that, throughout this paper, the subscript nn denotes the qubit index. Note that we have distributed the upper and lower component of the Dirac spinor to the even and odd lattice sites respectively. After performing the Jordan-Wigner transformation [69],

φn≡Ξn3σn+,(8)\varphi_n\equiv\Xi_n^3\sigma_n^+, \tag*{(8)}

the fields operator φn\varphi_n can be represented by quantum gates on a quantum circuit. Here we have introduced the raising and lowering operators σn±=12(σn1±iσn2)\sigma_n^\pm=\frac{1}{2}(\sigma_n^1\pm i\sigma_n^2), and the string operator Ξn3≡∏n′<nσn′3\Xi_n^3\equiv\prod_{n'<n}\sigma_{n'}^3, with σnj\sigma_n^j denoting the jj-th component of the Pauli matrix on the qubit nn. Throughout this study, we impose the periodic boundary condition. The LCDA then reads

ϕh(x)=1fh∑z14πe−i(x−1)mhzϕ~h(z),(9)\phi_h(x)=\frac{1}{f_h}\sum_z\frac{1}{4\pi}e^{-i(x-1)m_hz}\widetilde{\phi}_h(z), \tag*{(9)}

where

ϕ~h(z)=∑i,j=01⟨Ω∣eiHzφ−2z+i†e−iHzφj∣h⟩.(10)\widetilde{\phi}_h(z)=\sum_{i,j=0}^{1}\langle\Omega|e^{iHz}\varphi_{-2z+i}^\dagger e^{-iHz}\varphi_j|h\rangle. \tag*{(10)}

Quantum algorithm for LCDAs

We implement the quantum algorithm proposed in Ref. [51] to simulate the real-time light-cone correlator. The quantum algorithm is described by the quantum circuit shown in Figure 1, which consists of two parts: preparation of the hadronic state and evaluation of the correlator. Note that the vacuum state can be prepared as a particular case of a hadronic state.

Figure 1. Quantum circuit for calculation of the LCDA. Diagram (a) gives the circuit for hadronic state preparation, while (b) is for evaluation of the correlator.

Preparation of the hadronic state

The preparation of the hadronic state is achieved by the quantum-number-resolving variational quantum eigensolver (VQE), summarized as follows. To find the first kk excited states ∣h⟩|h\rangle with quantum numbers ll, we contruct the trial states ∣ψli(θ)⟩|\psi_{li}(\theta)\rangle, i=1,2,…,ki=1,2,\ldots,k, (i=1i=1 corresponds to the lowest-lying state) by

∣ψli(θ)⟩=U(θ)∣ψli⟩ref,(11)|\psi_{li}(\theta)\rangle=U(\theta)|\psi_{li}\rangle_{\mathrm{ref}}, \tag*{(11)}

where ∣ψli⟩ref|\psi_{li}\rangle_{\mathrm{ref}} are some input reference states with the same quantum numbers ll as the hadron states ∣h⟩|h\rangle, and U(θ)U(\theta) a symmetry-preserving unitary operator with parameters θ\theta. Since U(θ)U(\theta) preserves the quantum numbers, all of the trial states will have the same quantum numbers as ∣h⟩|h\rangle. Then we can find out the hadron states among the trial states by minimizing the cost function

El(θ)=∑i=1kwli⟨ψli(θ)∣H∣ψli(θ)⟩.(12)E_l(\theta)=\sum_{i=1}^{k}w_{li}\langle\psi_{li}(\theta)|H|\psi_{li}(\theta)\rangle. \tag*{(12)}

Here we require wl1>wl2>…>wlkw_{l1}>w_{l2}>\ldots>w_{lk}. The ii-th excited state ∣h⟩|h\rangle is then prepared as ∣h⟩=U(θ∗)∣ψli⟩ref|h\rangle=U(\theta^*)|\psi_{li}\rangle_{\mathrm{ref}} with θ∗\theta^* the optimized values of θ\theta. In the following, we describe in detail how the input reference states are prepared, and how the operator U(θ)U(\theta) is constructed with the quantum alternating operator ansatz (QAOA).

Preparation of input reference states

We complete the quantum algorithm framework for preparing hadronic states with detailed constructions of the input states. The input states have the same quantum numbers as the hadron and in general, they should be superposed states of the computational basis. We find that the input states are closely related to the so-called Dicke states, whose efficient preparation with explicit quantum circuits is available in Ref. [70]. To study generic hadronic states on a quantum computer, we outline the basic construction of the quantum circuit for the Dicke states.

In the NJL model, to prepare the lightest ∣qqˉ⟩|q\bar{q}\rangle state that has the same quantum numbers as the vacuum, the NN-qubit zero-momentum input state of the QAOA can be chosen as:

∣ψΩ,1⟩ref=∣010101…01⟩,|\psi_{\Omega,1}\rangle_{\mathrm{ref}}=|010101\ldots01\rangle,
∣ψΩ,2⟩ref=1N/2(∣1001…01⟩+∣0110…01⟩+⋯+∣0101…10⟩),(13)|\psi_{\Omega,2}\rangle_{\mathrm{ref}}=\frac{1}{\sqrt{N/2}}\left(|1001\ldots01\rangle+|0110\ldots01\rangle+\cdots+|0101\ldots10\rangle\right) , \tag*{(13)}

where ∣x1x2…xn⟩|x_1x_2\ldots x_n\rangle with x1,x2,…,xn=0,1x_1,x_2,\ldots,x_n=0,1 are the computational basis states for an nn-qubit system [71], both ∣ψΩ,1⟩ref|\psi_{\Omega,1}\rangle_{\mathrm{ref}} and ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} share the same quantum numbers with the ∣qqˉ⟩|q\bar{q}\rangle state. The state ∣ψΩ,1⟩ref|\psi_{\Omega,1}\rangle_{\mathrm{ref}} is a product state that can be easily prepared from the state ∣0000…00⟩|0000\ldots00\rangle. The preparation of the superposed state ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} is more involved, which we will focus on. We first denote ∣0ˉ⟩≡∣01⟩|\bar{0}\rangle\equiv|01\rangle and ∣1ˉ⟩≡∣10⟩|\bar{1}\rangle\equiv|10\rangle to write ∣ψΩ,2⟩|\psi_{\Omega,2}\rangle as:

∣ψΩ,2⟩ref=1CN/21(∣1ˉ0ˉ…0ˉ⟩+∣0ˉ1ˉ…0ˉ⟩+⋯+∣0ˉ0ˉ…1ˉ⟩).(14)|\psi_{\Omega,2}\rangle_{\mathrm{ref}}=\sqrt{\frac{1}{C_{N/2}^{1}}}\left(|\bar{1}\bar{0}\ldots\bar{0}\rangle+|\bar{0}\bar{1}\ldots\bar{0}\rangle+\cdots+|\bar{0}\bar{0}\ldots\bar{1}\rangle\right) . \tag*{(14)}

It can be seen that ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} is closely related to the Dicke state ∣D1N/2⟩|D_{1}^{N/2}\rangle [70], which can be prepared by a series of split-and-cyclic-shift (SCSn,1\mathrm{SCS}_{n,1}) gates as:

∣D1N/2⟩=∏n=2N/2SCSn,1∣0⟩⊗N2−1∣1⟩.(15)|D_{1}^{N/2}\rangle=\prod_{n=2}^{N/2}\mathrm{SCS}_{n,1}|0\rangle^{\otimes\frac{N}{2}-1}|1\rangle. \tag*{(15)}

The SCSn,1\mathrm{SCS}_{n,1} gates can be written as elementary controlled NOT (CNOT) and controlled RyR_y gates,

SCSn,1=[CNOT(n−1,n)×CRy(n,n−1,2cos⁡−11n)×CNOT(n−1,n)],(16)\mathrm{SCS}_{n,1}=\left[\mathrm{CNOT}(n-1,n)\times\mathrm{CR}_y\left(n,n-1,2\cos^{-1}\sqrt{\frac{1}{n}}\right)\times\mathrm{CNOT}(n-1,n)\right] , \tag*{(16)}

where CNOT(i,j)\mathrm{CNOT}(i,j) is the controlled NOT gate, with the qubit ii being the control qubit and the NOT gate acting on the qubit jj. CRy(i,j,θ)\mathrm{CR}_y(i,j,\theta) is the controlled RyR_y gate, where ii is the control qubit, and Ry(θ)R_y(\theta) acts on the qubit jj as a rotation about the yy-axis by an angle θ\theta. To prepare

∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}}, we first prepare the Dicke state ∣D1N/2⟩|D_{1}^{N/2}\rangle. Then to the right of each qubit we attach a new qubit initialized as ∣0⟩|0\rangle and perform a controlled-NOT gate operation so that ∣00⟩→∣01⟩|00\rangle\to|01\rangle and ∣10⟩→∣10⟩|10\rangle\to|10\rangle. In this way, ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} can be prepared from the Dicke state with an additional layer of two-qubit gates. The quantum circuit to prepare ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} from the Dicke state ∣D1N/2⟩|D_{1}^{N/2}\rangle is shown in Figure 2.

Figure 2. Quantum circuit for preparing ∣ψΩ,2⟩ref|\psi_{\Omega,2}\rangle_{\mathrm{ref}} from the Dicke state ∣D1N/2⟩|D_{1}^{N/2}\rangle. The input state for the even qubits is the Dicke state and the input state for the odd qubits is the state ∣0⟩⊗N/2|0\rangle^{\otimes N/2}.

As another example, the input state of ∣qq⟩|q q\rangle can be prepared in the following way. Since ∣qq⟩|q q\rangle is a superposition of configurations of two fermions in the odd sites while the even sites are empty, the input state is then a Dicke state ∣D1N/2⟩|D_{1}^{N/2}\rangle on the odd sites with all even sites at the state ∣1⟩|1\rangle. Input states for other hadrons can be constructed in a similar fashion.

Constructing U() with QAOA

The operator U(θ)U(\theta) can be constructed by the quantum alternating operator ansatz (QAOA). The Hamiltonian is split as H=H1+H2+⋯+HMH=H_1+H_2+\cdots+H_M, where M≥2M\geq2, with every HiH_i inheriting the symmetries of HH and [Hi,Hi+1]≠0[H_i,H_{i+1}]\ne0. Then U(θ)U(\theta) is given by

U(θ)=∏i=1p∏j=1Mexp⁡(iθijHj).(17)U(\theta)=\prod_{i=1}^{p}\prod_{j=1}^{M}\exp(i\theta_{ij}H_j). \tag*{(17)}

Because every HiH_i inherits the symmetries of HH, the time evolution exp⁡(iθijHj)\exp(i\theta_{ij}H_j) preserves the quantum numbers of the input reference state. The operator exp⁡(iθijHj)\exp(i\theta_{ij}H_j) can be regarded as a rotation in the high-dimensional Hilbert space about the axis HiH_i. The larger the MM, the more choices of rotational axes we have. Furthermore, since [Hi,Hi+1][H_i,H_{i+1}] does not vanish, successive rotations about HiH_i and Hi+1H_{i+1} are equivalent to a rotation about a new axis. Therefore, the larger the values of pp and MM, the better are the true hadronic states approximated by the trial states after the optimization. It should be noted that the required size of pp depends on the value of the bare coupling gg. If gg is small, say g∼0.1g \sim0.1, pp can be chosen to be NN; while if g∼0.5g \sim0.5, pp can be chosen to be N/2N/2. Optimization is achieved by minimizing the cost function. In our case of the (1+1)(1+1)D 1-flavor NJL model with staggered fermions, after the Jordan-Wigner transformation, the original Hamiltonian HH is split as H=H1+H2+H3+H4H = H_1 + H_2 + H_3 + H_4, with

H1=∑n=evenN/2−114(σn1σn+12−σn2σn+11),H_1 = \sum_{\substack{n=\mathrm{even}}}^{N/2-1} \frac{1}{4}\left(\sigma_n^1\sigma_{n+1}^2-\sigma_n^2\sigma_{n+1}^1\right),
H2=∑n=evenN/2−1g2σn3σn+13,H_2 = \sum_{\substack{n=\mathrm{even}}}^{N/2-1} \frac{g}{2}\sigma_n^3\sigma_{n+1}^3,
H3=H1(n=even→n=odd)+14ΞN−13(σN−12σ01−σN−11σ02),H_3 = H_1(n=\mathrm{even}\to n=\mathrm{odd})+\frac{1}{4}\Xi_{N-1}^3\left(\sigma_{N-1}^2\sigma_0^1-\sigma_{N-1}^1\sigma_0^2\right),
H4=∑n=0N/2−1mq2(−1)n(I−σn3)−g2(I−σn3).(18)H_4 = \sum_{n=0}^{N/2-1}\frac{m_q}{2}(-1)^n(I-\sigma_n^3)-\frac{g}{2}(I-\sigma_n^3). \tag*{(18)}

We will consider the lowest-lying scalar meson state, i.e. the lightest hadron state ∣h⟩|h\rangle with the same quantum numbers as the vacuum.3 The input reference states are as in eq. (13).

Evaluation of the correlator

For evaluation of the correlator, as depicted in Figure 1(b), with the help of an ancillary qubit we measure the correlation function

Smn(t)=⟨Ω∣eiHtΞm3σmie−iHtΞn3σnj∣h⟩,(19)S_{mn}(t)=\langle\Omega|e^{iHt}\Xi_m^3\sigma_m^i e^{-iHt}\Xi_n^3\sigma_n^j|h\rangle, \tag*{(19)}

of which ϕ~h(z)\tilde{\phi}_h(z) in eq. (10) can be written as a sum (see Ref. [51] for details). In Figure 1(b), the input state of the quantum circuit is taken as ∣ψΩ,1⟩ref=∣0101,…,01⟩|\psi_{\Omega,1}\rangle_{\mathrm{ref}}=|0101,\ldots,01\rangle. The quantum gate DD is implemented for preparation of the Dicke state for the reference hadron state: D∣ψΩ,1⟩ref=∣ψh⟩refD|\psi_{\Omega,1}\rangle_{\mathrm{ref}}=|\psi_h\rangle_{\mathrm{ref}}. The unitary gate U(θΩ∗)U(\theta_\Omega^*) and U(θh∗)U(\theta_h^*) can help us prepare the vacuum and the hadronic state: U(θΩ∗)∣ψΩ,1⟩ref=∣Ω⟩U(\theta_\Omega^*)|\psi_{\Omega,1}\rangle_{\mathrm{ref}}=|\Omega\rangle, U(θh∗)∣ψh⟩ref=∣h⟩U(\theta_h^*)|\psi_h\rangle_{\mathrm{ref}}=|h\rangle. Specifically, if we want to prepare the kk-th excited hadronic state with the same quantum numbers ll as the vacuum state, we have ∣ψh⟩ref=∣ψΩ,k⟩ref|\psi_h\rangle_{\mathrm{ref}}=|\psi_{\Omega,k}\rangle_{\mathrm{ref}} and θh∗=θΩ∗\theta_h^*=\theta_\Omega^*. Acting the controlled gates DD, U(θΩ∗)U(\theta_\Omega^*), and U(θh∗)U(\theta_h^*) on the circuit will facilitate evaluation of the dynamical correlation function ⟨Ω∣O∣h⟩\langle\Omega|O|h\rangle, in which the bra state is different from the ket state. In short, when we act all the gates before

the controlled Ξn3σnj\Xi_n^3\sigma_n^j gate in the circuit, the state looks like 12(∣0⟩∣Ω⟩+∣1⟩∣h⟩)\frac{1}{\sqrt{2}}(|0\rangle|\Omega\rangle+|1\rangle|h\rangle). After acting the controlled Pauli operators and the time evolution on the quantum circuit, we trace out the system and obtain the density matrix ρA\rho_A of the ancillary qubit. The (ρA)12(\rho_A)_{12} will have the form ⟨Ω∣O∣h⟩\langle\Omega|O|h\rangle since the states ∣0⟩|0\rangle and ∣1⟩|1\rangle of the ancillary qubit are entangled with ∣Ω⟩|\Omega\rangle and ∣h⟩|h\rangle respectively.

It should be noted that, in the case of QCD, inclusion of the Wilson line will complicate the quantum circuit for the evaluation of the light-cone correlator. With the Wilson line, the light-cone correlator can still be expressed as a sum of Smn(t)=⟨h∣U1(t)OmU2(t)On∣h⟩S_{mn}(t)=\langle h|U_1(t)O_mU_2(t)O_n|h\rangle, where U1U_1 and U2U_2 are unitary operators, and OmO_m and OnO_n are Hermitian operators. However, unlike the case without the Wilson line, now we have U1(t)U2(t)≠1U_1(t)U_2(t)\ne1. When OmO_m and OnO_n are the Pauli operators, which are unitary, Smn(t)S_{mn}(t) can be viewed as an overlap between the states U1(t)OmU2(t)On∣h⟩U_1(t)O_mU_2(t)O_n|h\rangle and ∣h⟩|h\rangle, which can be evaluated with the standard Hadamard test, the quantum circuit of which is shown in Figure 3.

Figure 3. Quantum circuit for the Hadamard test. The input state for the ancillary qubit is ∣0⟩|0\rangle and the input state for the system is the hadronic state ∣h⟩|h\rangle.

Results

The classical simulation of the quantum algorithm is performed on a desktop workstation with 16 cores, using open source packages QuSpin [72] and projectQ [73]. We perform the simulation of the LCDA for the lowest-lying scalar meson in the (1+1)(1+1)D 1-flavor NJL model with N=14N=14 qubits and different values of the bare coupling constant gg and hadron mass mhm_h. In practice, we first choose the values of the two free dimensionless parameters mqam_q a and gg, then from which we can determine the hadron state ∣h⟩|h\rangle, its mass mhm_h in units of a−1a^{-1}, as well as its LCDA. The values of mqam_q a and gg are chosen in such a way that the condition 2πNa<mh<πa\frac{2\pi}{Na}<m_h<\frac{\pi}{a} is satisfied, so that the lattice size is bigger than the hadron size and the lattice spacing is smaller than the hadron size. We choose the phase of the hadronic state ∣h⟩|h\rangle such that ϕh(x)\phi_h(x) is a real function.

We show in Figure 4 and Figure 5 the results for the real part and imaginary part of the LCDA in position space ϕ~h(z)\tilde{\phi}_h(z), respectively, with fixed value of mh=1.5a−1m_h=1.5a^{-1} and different values of gg. We also show as dotted lines the inverse Fourier transform of the asymptotic form of ϕh(x)\phi_h(x) in QCD, namely ϕasymp(x)=6x(1−x)\phi_{\mathrm{asymp}}(x) = 6x(1-x) [5]. One notices that, as expected, the result from the quantum algorithm converges to the asymptotic LCDA as g→0g \to0, which mimics the asymptotic behavior of LCDAs in QCD.

Figure 4. Real part of ϕ~h(z)\tilde{\phi}_h(z) in the (1+1)D 1-flavor NJL model with N=14N = 14, mh=1.5a−1m_h = 1.5a^{-1}.

Figure 5. Imaginary part of ϕ~h(z)\tilde{\phi}_h(z) in the (1+1)D 1-flavor NJL model with N=14N = 14, mh=1.5a−1m_h = 1.5a^{-1}.

Now we can evaluate the LCDA ϕh(x)\phi_h(x) by taking the Fourier transform of ϕ~h(z)\tilde{\phi}_h(z) as described in eq. (9). The final results are shown in Figure 6, where the curves represent the results from exact diagonalization (ED), which are obtained from numerically diagonalizing the discretized NJL Hamiltonian, and the discrete open markers denote the results from quantum computing (QC) using classical hardware simulation. The excellent agreement between the results from the quantum algorithm and those from exact diagonalization justifies the designed quantum algorithm. Similar to Figure 4 and Figure 5, we also show the asymptotic form of ϕh(x)\phi_h(x) in QCD for comparison. As expected, the peak of the LCDA gets narrower and converges to the asymptotic LCDA as g→0g \to0. Notice that there is non-vanishing but suppressed contributions in the nonphysical region (x>1x > 1 or x<0x < 0). Such unphysical oscillations are caused by finite volume effects in the naively truncated Fourier transform, and are also commonly seen in lattice QCD calculations [29]. We also check the dependence of the LCDA on the hadron mass mhm_h in Figure 7. For this purpose,

Figure 6. LCDA in the (1+1)D 1-flavor NJL model with N=14N = 14, mh=1.5a−1m_h = 1.5a^{-1}.

Figure 7. Dependence of the LCDA on the hadron mass mhm_h with fixed bare coupling g=0.1g = 0.1 in the (1+1)D 1-flavor NJL model.

we take mh=1.3a−1,1.5a−1,1.7a−1m_h = 1.3a^{-1}, 1.5a^{-1}, 1.7a^{-1}, and fix the bare coupling constant g=0.1g = 0.1, with N=14N = 14. One can see that the peak of ϕh(x)\phi_h(x) gets narrower when the hadron mass increases. This is the expected behavior when the valence quark and antiquark become nonrelativistic, in which case the quark masses dominate the momenta of the quark and antiquark, while the relative momentum between the two becomes small. The behavior also agrees with the results from lattice QCD [29]. We leave the extrapolation to the continuum limit as a follow-up work in the future.

Summary

In this study, we presented the first direct simulation of the light-cone distribution amplitude (LCDA) with a quantum algorithm on classical hardware. Using a quantum algorithm we proposed recently for the evaluation of the parton distribution functions, here we performed the simulation for the LCDA in the (1+1)(1+1)-dimensional Nambu-Jona-Lasinio (NJL) model on classical hardware. With 14 qubits, our results from the quantum algorithm agree with those from exact diagonalization of the discretized NJL model. Our results of the LCDA showed the expected dependence on the coupling constant and the hadron mass.

The result presented in this study manifests the feasibility of using quantum computing to solve intrinsic difficulties in realizing real-time dynamics with classical computing facilities. Meanwhile, it demonstrates that the recently proposed quantum computing framework for preparing hadronic states and measuring dynamical correlation functions is generally applicable. The extension of the algorithm to other applications in high energy particle and nuclear physics can be expected.

Acknowledgments

This work is supported by the National Natural Science Foundation of China under Grant No. 12022512, No. 12035007 (H.X.), Grant No. 12175016 (X.L.), Grant No.12005065 (D.B.), Grant No. 12074180 (S.L.), and by the Guangdong Major Project of Basic and Applied Basic Research No. 2020B0301030008, the Key-Area Research and Development Program of GuangDong Province (Grant No. 2019B030330001), the Guangdong Basic and Applied Basic Research Fund No.2021A1515010317 (D.B.), the Key Project of Science and Technology of Guangzhou (Grant No. 2019050001), the National Special Support Program for High-level Talents (X.L.).

References

  1. [1]H. Abramowicz et al., Eur. Phys. J. C 75, no.12, 580 (2015)
  2. [2]R. Abdel Khalek, A. Accardi, J. Adam, D. Adamiak, W. Akers, M. Albaladejo, A. Al-batainah, M. G. Alexeev, F. Ameli and P. Antonioli, et al. Nucl. Phys. A 1026, 122447 (2022)
  3. [3]Cao, X. et al., Nucl. Tech., 43, 020001 (2020).
  4. [4]D. P. Anderle, V. Bertone, X. Cao, L. Chang, N. Chang, G. Chen, X. Chen, Z. Chen, Z. Cui and L. Dai et al., Front. Phys. 16, 64701 (2021).
  5. [5]G. P. Lepage and S. J. Brodsky, Phys. Rev. D 22, 2157 (1980).
  6. [6]A. V. Efremov and A. V. Radyushkin, Phys. Lett. B 94, 245 (1980).
  7. [7]V. L. Chernyak and A. R. Zhitnitsky, Phys. Rep. 112, 173 (1984).
  8. [8]V. L. Chernyak and A. R. Zhitnitsky, Nucl. Phys. B 201, 492 (1982), [Erratum: Nucl.Phys.B 214, 547 (1983)].
  9. [9]V. L. Chernyak and I. R. Zhitnitsky, Nucl. Phys. B 246, 52 (1984).
  10. [10]I. D. King and C. T. Sachrajda, Nucl. Phys. B 279, 785 (1987).
  11. [11]V. L. Chernyak, A. A. Ogloblin, and I. R. Zhitnitsky, Yad. Fiz. 48, 1410 (1988).
  12. [12]V. L. Chernyak, A. A. Ogloblin, and I. R. Zhitnitsky, Yad. Fiz. 48, 1398 (1988).
  13. [13]A. V. Radyushkin, Nucl. Phys. A 532, 141 (1991).
  14. [14]E. Ruiz Arriola and W. Broniowski, Phys. Rev. D 74, 034008 (2006), arXiv:hep-ph/0605318.
  15. [15]E. Ruiz Arriola and W. Broniowski, Phys. Rev. D 66, 094016 (2002), arXiv:hep-ph/0207266.
  16. [16]L. Chang et al., Phys. Rev. Lett. 110, 132001 (2013), arXiv:1301.0324.DOI
  17. [17]S. S. Agaev, V. M. Braun, N. Offen, and F. A. Porkert, Phys. Rev. D 86, 077504 (2012), arXiv:1206.3968.DOI
  18. [18]S. J. Brodsky and G. F. de Teramond, Phys. Rev. Lett. 96, 201601 (2006)
  19. [19]J. P. Vary, H. Honkanen, J. Li, P. Maris, S. J. Brodsky, A. Harindranath, G. F. de Teramond, P. Sternberg, E. G. Ng and C. Yang, Phys. Rev. C 81, 035205 (2010)
  20. [20]S. J. Brodsky, G. F. de Teramond, H. G. Dosch and J. Er-
  21. [21]lich, Phys. Rep. 584, 1-105 (2015) S. Xu et al. [BLFQ], Phys. Rev. D 104, no.9, 094036 (2021)
  22. [22]J. P. Ma and Z. G. Si, Phys. Lett. B 647, 419 (2007), arXiv:hep-ph/0608221.DOI
  23. [23]Y. Jia and D. Yang, Nucl. Phys. B 814, 217 (2009), arXiv:080812.1965.
  24. [24]M. Gockeler et al., Phys. Rev. Lett. 101, 112002 (2008), arXiv:0804.1877.DOI
  25. [25]V. M. Braun et al., Phys. Rev. D 79, 034504 (2009), arXiv:0811.2712.DOI
  26. [26]V. M. Braun et al., Phys. Rev. D 89, 094511 (2014), arXiv:1403.4189.DOI
  27. [27]G. S. Bali et al., J. High Energ. Phys. 02, 070 (2016), arXiv:1512.02050.DOI
  28. [28]G. S. Bali et al., J. High Energ. Phys. 08, 065 (2019), arXiv:1903.08038.arxiv.org/abs/1903.08038
  29. [29]R. Zhang, C. Honkala, H.-W. Lin, and J.-W. Chen, Phys. Rev. D 102, 094519 (2020), arXiv:2005.13955.DOI
  30. [30]J.-W. Zhang, J.-W. Chen, X. Ji, L. Jin, and H.-W. Lin, Phys. Rev. D 95, 094514 (2017), arXiv:1702.00008.DOI
  31. [31]J.-H. Zhang et al., Nucl. Phys. B 939, 429 (2019), arXiv:1712.10025.arxiv.org/abs/1712.10025
  32. [32]J. Hua et al., Phys. Rev. Lett. 127, 062002 (2021), arXiv:2011.09788.DOI
  33. [33]J. Hua et al., arXiv:2201.09173.DOI
  34. [34]A. Alexandru, G. Basar, P. F. Bedaque, S. Vartak, and N. C. Warrington, Phys. Rev. Lett. 117, 081602 (2016), arXiv:1605.08040.DOI
  35. [35]F. Arute et al., Nature 574, 505 (2019), arXiv:1910.11333.DOI
  36. [36]M. R. Dietrich et al., Opportunities for Nuclear Physics and Quantum Information Science, arXiv:1903.05453.arxiv.org/abs/1903.05453
  37. [37]D.-B. Zhang, H. Xing, H. Yan, E. Wang, and S.-L. Zhu, Chin. Phys. B 30, 020306 (2021), arXiv:2011.01431.DOI
  38. [38]C. W. Bauer et al., arXiv:2204.03381.DOI
  39. [39]S. P. Jordan, K. S. M. Lee, and J. Preskill, Science 336, 1130 (2012), arXiv:1111.3633.DOI
  40. [40]S. P. Jordan, K. S. M. Lee, and J. Preskill, Quant. Inf. Comput. 14, 1014 (2014), arXiv:1112.4833.arxiv.org/abs/1112.4833
  41. [41]S. P. Jordan, K. S. M. Lee, and J. Preskill, arXiv:1404.7115.arxiv.org/abs/1404.7115
  42. [42]N. Klco and M. J. Savage, Phys. Rev. A 99, 052335 (2019), arXiv:1808.10378.DOI
  43. [43]E. F. Dumitrescu et al., Phys. Rev. Lett. 120, 210501 (2018), arXiv:1801.03897.DOI
  44. [44]H.-H. Lu et al., Phys. Rev. A 100, 012320 (2019), arXiv:1810.03959.DOI
  45. [45]H. Lamm, S. Lawrence, and Y. Yamauchi, Phys. Rev. Res. 2, 013272 (2020), arXiv:1908.10439.DOI
  46. [46]N. Mueller, A. Tarasov, and R. Venugopalan, Phys. Rev. D 102, 016007 (2020), arXiv:1908.07051.DOI
  47. [47]A. Roggero, A. C. Y. Li, J. Carlson, R. Gupta, and G. N. Perdue, Phys. Rev. D 101, 074038 (2020), arXiv:1911.06368.DOI
  48. [48]M. G. Echevarria, I. L. Egusquiza, E. Rico, and G. Schnell, arXiv:2011.01275.DOI
  49. [49]C. W. Bauer, M. Freytsis, and B. Nachman, arXiv:2102.05044.DOI
  50. [50]Y. Y. Atas et al., Nature Commun. 12, 6499 (2021), arXiv:2102.08920.DOI
  51. [51]T. Li et al., Phys. Rev. D 105, L111502 (2022), arXiv:2106.03865.DOI
  52. [52]M. Kreshchuk, S. Jia, W. M. Kirby, G. Goldstein, J. P. Vary and P. J. Love, Entropy 23, no.5, 597 (2021).
  53. [53]D. Gallimore and J. Liao, arXiv:2202.03333.DOI
  54. [54]E. A. Martinez et al., Nature 534, 516 (2016), arXiv:1605.04570.DOI
  55. [55]Z. Hu, R. Xia, and S. Kais, Sci. Rep. 10, 3301 (2020), arXiv:1904.00910.DOI
  56. [56]C. W. Bauer, W. A. de Jong, B. Nachman, and D. Provasoli, Phys. Rev. Lett. 126, 062001 (2021), arXiv:1904.03196.DOI
  57. [57]W. A. De Jong et al., Phys. Rev. D 104, 051501 (2021), arXiv:2010.03571.DOI
  58. [58]Z.-Y. Zhou et al., arXiv:2107.13563.DOI
  59. [59]W. A. de Jong et al., arXiv:2106.08394.DOI
  60. [60]S. Williams, S. Malik, M. Spannowsky, and K. Bepari, arXiv:2109.13975.DOI
  61. [61]Y. Y. Atas et al., arXiv:2207.03473.arxiv.org/abs/2207.03473
  62. [62]X. Yao, arXiv:2205.07902.arxiv.org/abs/2205.07902
  63. [63]A. M. Czajka, Z.-B. Kang, H. Ma, and F. Zhao, arXiv:2112.03944.DOI
  64. [64]X.-D. Xie et al., arXiv:2205.12767.DOI
  65. [65]M. C. Bañuls et al., Eur. Phys. J. D 74, 165 (2020), arXiv:1911.00003.DOI
  66. [66]Y. Nambu and G. Jona-Lasinio, Phys. Rev. 122, 345 (1961).
  67. [67]Y. Nambu and G. Jona-Lasinio, Phys. Rev. 124, 246 (1961).
  68. [68]D. J. Gross and A. Neveu, Phys. Rev. D 10, 3235 (1974).
  69. [69]S. Backens, A. Shnirman, and Y. Makhlin, Sci. Rep. 9 (2019).
  70. [70]A. Bärtschi and S. Eidenbenz, Deterministic preparation of dicke states, in L.A. Gasiënice, J. Jansson, and C. Levopopoulos, eds., Fundamentals of Computation Theory (Springer, Cham,2019), pp. 126–139.DOI
  71. [71]M. A. Nielsen, and I.L. Chuang, Quantum Computation and Quantum Information (Cambridge University Press, Cambridge, 2010).
  72. [72]P. Weinberg and M. Bukov, SciPost Phys. 2, 003 (2017).
  73. [73]D. S. Steiger, T. Häner, and M. Troyer, Quantum 2, 49 (2018).

Paper details

vlnc:W_j2q6wdhf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents