Quantum simulations of one dimensional quantum systems

• Quantum simulation can require rapidly growing classical time and memory, motivating quantum approaches for continuous-variable systems whose effective energy scale may grow with Hilbert-space dimension.Source: §I · pp. 1–2

QHO evolution

• The QHO evolution algorithm has complexity O(exp(γ√log(N/ϵ))) and uses refined Trotter-Suzuki analysis exploiting Lie-algebra structure.Source: §I · p. 2

QHO eigenstates

• The paper prepares discrete QHO eigenstates with complexity polynomial in log(N)/ϵ, including states with Gaussian-like amplitudes.Source: §I · p. 2

Quartic potential

• Numerical simulations suggest quartic-potential evolution with complexity O~(|t|^(1+η)N^(1/3+4η/3)/ϵ^η) for arbitrarily small constant η.Source: §I · p. 2

General systems

• For general one-dimensional systems, an interaction-picture series method gives complexity O~(|t|N) with polylogarithmic dependence on 1/ϵ under fairly general assumptions.Source: §I · p. 2

Application

• The results are also discussed in connection with the fractional Fourier transform, a signal-analysis generalization of the Fourier transform formulated through QHO evolution.Source: §I · p. 2

• The QHO has Hermite-function eigenstates and periodic evolution U(t)=exp(−iHt), allowing the evolution time to be treated as O(1).Source: §II · pp. 2–3
• Raising and lowering operators generate QHO eigenstates from the ground state, motivating quantum procedures for preparing approximate eigenstates.Source: §II · p. 3
• QHO eigenstates are Fourier-transform eigenstates with eigenvalues (−i)^n, a property used in the simulation analysis.Source: §II · p. 3
• The Hamiltonian operators generate a three-dimensional sp(2) Lie algebra, making nested-commutator norms smaller than worst-case norm products in error analysis.Source: §II · p. 3
• The algorithms target scattering amplitudes, expectation values, and correlation functions derived from evolved states or observables.Source: §II · p. 3
• Classical approximations can require polynomial complexity in the relevant cutoff because higher Hermite functions oscillate more strongly, although some QHO quantities are analytically accessible.Source: §II · p. 3

• The discrete QHO uses an even Hilbert-space dimension N and defines discrete position and momentum operators together with a discrete Fourier transform.Source: §III · p. 3
• The centered discrete Fourier transform differs from the standard discrete Fourier transform by a cyclic permutation of indices.Source: §III · pp. 3–4
• The cyclic permutation and centered Fourier transform admit efficient quantum-circuit implementations using polylogarithmic numbers of two-qubit gates.Source: §III · p. 4
• The analysis generally assumes large N, although some results remain applicable when N is fairly small.Source: §III · p. 4

• The low-energy subspace is spanned by states with n≤cN, where continuous Hermite functions are well approximated by piecewise-constant functions.Source: A. Spectral properties · p. 4
• The discrete construction replaces continuous integrals with finite sums, but large-n Hermite oscillations may not be captured and increase approximation error.Source: A. Spectral properties · p. 4
• Discrete Hermite states approximate discrete-QHO eigenstates, and their eigenvalues and eigenvectors can achieve errors exp(−Ω(N)) for n≤cN.Source: A. Spectral properties · p. 4
• The spectral approximation follows from bounding the squared residual of the discrete Hamiltonian acting on discrete Hermite states and using gaps between low-energy eigenvalues.Source: A. Spectral properties · p. 4
• Numerical results suggest these discrete states remain excellent eigenstate approximations across a fraction c≥3/4 of the spectrum.Source: A. Spectral properties · p. 4
• Numerical eigenvalue comparisons indicate that the difference from n+1/2 decays exponentially with N, with the exponent depending on n.Source: A. Spectral properties · p. 4

Discretization accuracy

• The discrete Hamiltonian eigenvalues approach the continuous QHO values n + 1/2 exponentially with N, although the exponential constant decreases as n increases.The numerical study examines n = N/2 and n = 3N/4 and reports different decay constants.Source: Figure 2 · p. 5, Figure 3 · p. 5

Scattering-amplitude approximation

• The discrete QHO evolution Ud(t) is used to approximate amplitudes ⟨ϕ′|U(t)|ϕ⟩ through corresponding discretized states and matrix elements.The construction uses low-energy expansions in discrete Hermite states and includes a position-eigenstate case.Source: §IV · p. 5

Discretization accuracy

• Theorem 1 bounds discretization errors by exponentially small terms in the Hilbert-space dimension, yielding O(ϵ) accuracy after choosing N logarithmic in 1/ϵ.The exponential approximation concerns both the discrete eigenvalues and the relevant eigenstate representations.Source: Figure 3 · p. 5, §IV · p. 6

Scattering-amplitude approximation

• N = O(log(1/ϵ) + N′) suffices to approximate continuous-variable QHO scattering amplitudes for evolution times |t| ≥ 1 with t = O(1).The bound applies when the states are supported on quantum numbers up to N′ or when one state is a discretized position eigenstate.Source: §IV · pp. 5–6

Algorithmic consequence

• If the discretized input and output states can be efficiently prepared, scattering-amplitude complexity is dominated by implementing the discrete evolution operator.This separates state-preparation cost from the cost of simulating Ud(t).Source: §IV · p. 6

Implementation

• Each exponential of (xd)^2 or (pd)^2 can be implemented with polylogarithmic two-qubit-gate cost in N and the target precision.The momentum-squared evolution is implemented using the centered Fourier transform and efficient cyclic-shift or diagonal operations.Source: A. Time evolutions: Trotter-Suzuki… · p. 6

Lie-algebra structure

• The improvement comes from bounding projected nested commutators instead of using the full Hamiltonian norm O(N).The projected commutator norm is O(N), whereas a direct analysis would give O(N^2).Source: A. Time evolutions: Trotter-Suzuki… · p. 7
• The operators (xd)^2, (pd)^2, and {xd, pd} nearly form a three-dimensional Lie algebra on the low-energy subspace.This structure controls the relevant commutators in the high-order product-formula analysis.Source: A. Time evolutions: Trotter-Suzuki… · p. 7

Complexity bound

• The refined Trotter-Suzuki analysis yields M = O(|t| exp(γ√log(N′|t|/ϵ))) exponentials for simulating Ud(t).The bound follows from choosing the product-formula order and step size using the low-energy structure.Source: A. Time evolutions: Trotter-Suzuki… · p. 7

Numerical behavior

• For fixed step size and order, simulations observe nearly linear Trotter error growth in n rather than the worst-case O(n^(2p+1)/2) scaling.Additional simulations report the same linear behavior for higher product-formula orders.Source: A. Time evolutions: Trotter-Suzuki… · p. 7

Algorithm and complexity

• The algorithm has three stages: prepare the initial state, simulate the discrete evolution, and perform a projective measurement.The propagator is encoded as an overlap and estimated by repeated executions of an ancilla-based circuit.Source: B. Subexponential-time quantum… · pp. 7–8

Numerical validation

• The observed Trotter error is almost linear in n for N = 400, p = 4, and s = 1, with similar scaling at higher p.This behavior is identified as a main reason for the quantum speedup.Source: Figure 4 · p. 8
• With parameters selected according to Theorem 2, simulations show the approximation error is bounded by ϵ for t = π/2 and n = N/2.The figure varies the dimension while using the theorem-determined product-formula parameters.Source: Figure 5 · p. 8

Algorithm and complexity

• MQ = O(exp(γ′√log(N′))) two-qubit gates suffice for the scattering-amplitude algorithm at constant precision and evolution time.The algorithm combines the evolution simulation with state preparation and measurement procedures.Source: B. Subexponential-time quantum… · p. 8

Quantum speedup

• The algorithm is subexponential in log(N′), whereas classical propagator computation is expected to require poly(N′) operations in the worst case.The paper characterizes this gap as a superpolynomial quantum speedup.Source: B. Subexponential-time quantum… · p. 9

Approximation scope

• The section presents techniques for computing expectation values on various QHO eigenstates by replacing the initial state |ϕ⟩ with corresponding |ψn⟩, or |ϕd⟩ with |φd n⟩.Source: §V · p. 9

Ground-state preparation

• Polynomial in log(N/ϵ) two-qubit gates suffice to prepare an approximation of the discrete QHO ground state.The construction uses a Gaussian-like initial state, free-particle evolution, and discrete quadratic evolutions.Source: §V · p. 9

Numerical validation

• For the ground-state construction, the preparation error decays exponentially with the Gaussian-width parameter δ in numerical simulations.The simulations use N = 800 and show no significant changes for larger dimensions.Source: §V · p. 9

Excited-state preparation

• The Jaynes-Cummings evolution transfers population from eigenstate n with ancilla |0⟩ toward eigenstate n+1 with ancilla |1⟩, enabling recursive preparation.An ancilla σx operation resets the ancilla between preparation steps.Source: §V · pp. 9–10
• O((n^2/ϵ)polylog(N/ϵ)) two-qubit gates prepare the n-th approximate eigenstate for n ≤ N′ ≤ cN.The circuit uses a discrete Jaynes-Cummings construction and high-order Trotter-Suzuki simulation.Source: §V · pp. 9–10

Method and scope

• The quartic-potential analysis is numerical only because no exact solution is available in this case.The authors therefore state the polynomial speedup as a simulation-supported suggestion.Source: §VI · p. 10

Discretization

• The discrete quartic system approximates a large low-energy sector, with eigenvalue ratios suggesting N = O(n log n) suffices to resolve the n-th eigenvalue.The ratio scaling appears closer to 1/log N than to 1/N^χ.Source: §VI · pp. 10–11

Method and scope

• The quartic Hamiltonian lacks the finite-dimensional Lie algebra structure of the QHO, so tight high-order Trotter-Suzuki bounds are obtained through numerical observations.Its nested-commutator structure nevertheless appears to produce smaller effective norms than worst-case norm products.Source: §VI · p. 11

Trotter-Suzuki error

• The observed Trotter-Suzuki error scales as O(|s|^(2p+1)n^((2p+4)/3)) for orders p = 1, 2, 3.The scaling was supported by numerical simulations in a convergence region.Source: §VI · p. 11, Figure 9 · p. 11

Complexity bound

• As the Trotter order grows, simulations suggest evolution complexity O(|t|^(1+η)N^(1/3+4η/3)/ϵ^η) up to polylogarithmic factors.This yields a polynomial quantum speedup in N relative to the stated classical worst-case scaling.Source: §VI · p. 12

Assumptions

• The theorem assumes the initial state has support only through eigenstate N′ and that the potential norm on those eigenstates is polynomial in N′.It also assumes the discrete system approximates the continuous scattering amplitudes.Source: §VII · p. 12

Upper bound

• Under the stated assumptions, a quantum circuit simulates the discretized evolution with complexity Õ(|t|N) for |t| = Ω(1).The discretization dimension N is assumed polynomial in the relevant initial-state energy cutoff N′ at constant precision.Source: §VII · p. 12

Interaction-picture method

• The interaction-picture representation avoids a large direct cost from the non-sparse Hamiltonian and high potential norm.The potential evolution can be simulated separately with polylogarithmic complexity in N|t|/ϵ.Source: §VII · p. 12
• The discretized position operator is 1-sparse, while the interaction Hamiltonian can be decomposed into O(N) unitary operators.The number of qubits is q = O(log N) under the theorem’s assumptions.Source: §VII · p. 12

Fractional Fourier transform

• QHO evolution realizes a fractional Fourier transform corresponding to an arbitrary phase-space rotation.The transform has applications in signal analysis, particularly noise filtering without a well-defined frequency spectrum.Source: §VIII · pp. 12–13
• The simulated QHO evolution provides an approximate discrete fractional Fourier transform for possible classical or quantum signal-analysis algorithms.Efficient signal encoding and decoding may require additional assumptions such as sparsity.Source: §VIII · p. 13

Related methods

• The paper’s Trotter-Suzuki error analysis can be used to bound errors in split-step Fourier methods for nonlinear Schrödinger equations.Both approaches apply short evolutions under separate operators after discretizing continuous-variable coordinates.Source: §VIII · p. 13

Related work

• Unlike prior continuous-variable algorithms described as polynomial in system energy, the QHO evolution algorithm has subexponential dependence on log N′.The QHO is also identified as a basis for quantizing the electromagnetic field and quantum field theories.Source: §VIII · p. 13

Quantum harmonic oscillator

• The QHO propagator algorithm achieves superpolynomial quantum speedup over the classical O(N) method, with subexponential complexity in the relevant energy scale.Its complexity is O(exp(γ√log(N/ϵ))) for precision ϵ, and MQ/N^η → 0 for every η > 0.Source: §IX · p. 13
• The QHO improvement relies on high-order Trotter-Suzuki error bounds exploiting the three-dimensional sp(2) Lie algebra.The discrete QHO reproduces the low-energy spectrum with errors decaying exponentially in N.Source: §IX · p. 13

Eigenstate preparation

• The authors provide eigenstate-preparation algorithms with complexity polynomial in log(N)/ϵ, including states with Gaussian-like amplitudes.The preparation simulates a discrete Jaynes-Cummings evolution using high-order Trotter-Suzuki approximation.Source: §IX · p. 13

More complex one-dimensional systems

• For quartic potentials, numerical evidence suggests complexity ˜O(N^(1/3+o(1))) at constant t and ϵ, yielding a polynomial quantum speedup.For more general one-dimensional systems, the paper proves a ˜O(N) bound under fairly general assumptions.Source: §IX · p. 13

Scope

• The authors conjecture that some results may extend to other continuous-variable quantum systems.This extension remains a conjecture rather than a proved result.Source: §IX · p. 13

Appendix A: Properties of the discrete QHO

• For n, m ≤cN, discrete Hermite states are almost orthonormal because truncation and nonzero-period translations contribute errors exponentially small in N.The analysis controls Hermite-function tails and shows discrete inner products approximate δn,m with exponentially small precision loss.Source: Appendix A · p. 16
• For n ≤cN, discrete Hermite states retain the centered Fourier-transform eigenvalue structure up to exponentially small errors.The proof uses Hermite-function Fourier eigenvalues and exponentially suppressed contributions from shifted Dirac-comb terms.Source: Appendix A · p. 17

Appendix B: High-order Trotter-Suzuki formula for the discrete QHO

• The high-order Trotter-Suzuki formula achieves order 2p + 1 in s, with approximation error improving for lower-energy states and depending only linearly on n.For sufficiently small |s|, the construction exploits the QHO Lie algebra; its coefficients scale as O(|s|2p), and it uses O(5^p) exponentials.Source: Appendix B · pp. 18–19
• For constant |t| and ϵ, the exponential count is subpolynomial but grows faster than every power of log n, satisfying lim M/n^α = 0 and lim (log n)^α/M = 0 for all α > 0.This establishes the intermediate asymptotic scaling of the optimized high-order formula in the discrete QHO setting.Source: Appendix B · p. 20
• The discrete approximation achieves O(ϵ) error for low-energy states, with n ≤ c′N and exponentially small discretization errors made negligible by sufficiently large N.The required regime includes N ≥ exp(δ log(N′|t|/ϵ)) and ensures the continuous-to-discrete approximation errors remain within the target precision.Source: Appendix B · pp. 20–21
• Each high-order short-time approximant uses fewer than 5p exponentials of (x_d)^2 and (p_d)^2, yielding total cost O(|t|5^(2p)).The exponentials are approximated by truncated Taylor series, whose error is controlled using ||(x_d)^2|| = ||(p_d)^2|| = O(N).Source: Appendix B · pp. 20–21
• M = O(|t| exp(γ√log(N′|t|/ϵ))) exponentials suffice to simulate the discrete QHO for all eigenstates n ≤ N′ under the stated dimension conditions.The construction uses choices of s and p satisfying 5^p|s| = O(1), with N constrained exponentially in log(N′|t|/ϵ).Source: Appendix B · p. 21

Appendix C: Error bounds for the preparation of quantum states with Gaussian amplitudes

• The remaining approximation terms include a global phase depending only on t and an eigenvalue-related error of O(exp(−βN)).These terms are combined with the other errors through subadditivity.Source: Appendix C · pp. 22–24
• The Gaussian-state construction approximates the continuous-variable evolution using finite sums on a discretized grid, with total error bounded by the sum of component errors.The derivation connects continuous-variable integrals and evolutions to discrete states and operators.Source: Appendix C · pp. 22–23
• The dominant finite-grid approximation errors decay exponentially, including ε2 = O(exp(−λ^(2)N^2/δ)) and ε7 = O(exp(−λ^(7)N^2/δ)).The cutoff and Fourier-transform arguments control truncation and discretization effects.Source: Appendix C · pp. 23–24
• The discrete Gaussian-like state approximates the target continuous state with an error of order O(N^(5/4) exp(−λ^(3)N^2)).The construction uses Dirac-comb convolutions and discrete Fourier-transform properties.Source: Appendix C · p. 23

Appendix D: The discrete Jaynes-Cummings Hamiltonian

• The discrete Jaynes-Cummings Hamiltonian has Hilbert-space dimension 2N, and its relevant normalized states approximate eigenstates with exponentially small error.The analysis applies for n ≤ N′ ≤ cN under the stated asymptotic assumptions.Source: Appendix D · p. 24
• A first-order asymmetric Trotter-Suzuki approximation already provides the desired N scaling, while higher-order formulas can improve precision dependence.The approximation is applied to the discrete Jaynes-Cummings evolution.Source: Appendix D · p. 24