Spatial Discretization Error for Quantum Chemistry Simulation

AI summary

Overview

Research area: Quantum simulation of quantum chemistry, specifically the numerical analysis of spatial discretization error for continuum many-body Coulomb Hamiltonians.

Technical level: Advanced. The paper is a rigorous mathematical analysis using Sobolev spaces, many-body Hardy inequalities, Galerkin projection, and quantum-algorithm complexity bounds.

Scope: A Sobolev-Galerkin framework that bounds finite-cell and finite-plane-wave errors for many-body Coulomb dynamics and ground-state energies, and converts those bounds into end-to-end logical gate complexities for quantum simulation.

What This Paper Is About

Quantum chemistry algorithms are usually formulated in a finite basis, and their costs are stated in terms of the number of basis functions, treated as an independent input. For the underlying continuum Schrödinger problem, that number is not independent: it must be chosen to achieve a target spatial accuracy, which depends on particle number, simulation time, target precision, and physical scales. This paper derives explicit spatial discretization error estimates for the original, unregularized many-body Coulomb Hamiltonian and uses them to determine the basis size and the resulting end-to-end quantum gate costs.

The treatment is analytical: the paper reports rigorous bounds and complexity tables rather than numerical benchmarks.

Key Contributions

1. A general Sobolev-Galerkin error framework for unbounded Hamiltonians. The framework separately accounts for the finite computational domain and the finite number of basis functions, rather than treating the basis size as an independent parameter.

2. Periodization and Galerkin error estimates for many-body Coulomb systems. The paper establishes bounds for both real-time wave-function dynamics and ground-state energies, with explicit dependence on time, particle number, nuclear charges, cell size, momentum cutoff, and initial-state norms.

3. Identification of a dynamical projection error distinct from the state projection error. The paper shows that the finite-basis projection error of the exact wave function does not capture the full Galerkin error. An inner-product argument controls the additional dynamical contribution and yields a (4/3)-order convergence bound from (H^{2}) initial regularity, whereas applying the Hardy estimate directly in the forcing norm gives only a first-order bound. The paper also proves a sharp (M^{-1}) ground-state energy rate at fixed cell size and shows that a uniform (M^{-1}) real-time (L^{2}) rate is impossible for general (H^{2}) initial data.

4. End-to-end complexity bounds with all spatial errors explicitly controlled. To the authors' knowledge, these are the first rigorous end-to-end complexity bounds for quantum simulation of molecular and electronic dynamics with spatial errors controlled from the continuum problem, without regularizing the Coulomb interaction.

Main Findings

• The finite-basis error has two distinct parts. The exact decomposition splits the Galerkin error into the state finite-basis projection error and the dynamical projection error (w(t)=P_K\tilde\psi(t)-\tilde\psi_K(t)). For smooth bounded potentials these can share a rate; for Coulomb interactions their rates differ, so estimating only the projection error changes the predicted accuracy.

• Periodization error bound. For normalized initial states in (H^{2}(\mathbb{R}^{3\eta})) with (|x|\psi_0\in L^{2}), the phase-corrected (L^{2}) periodization error is bounded by [ \mathcal{O}!\left(\frac{T\eta^{3/2}}{\rho}+\frac{T(\eta\rho^{2}+\eta^{2})}{\ell^{3}}\right), ] where (\rho) is the localization radius and (\ell) is the cell side length. The two terms correspond to the wave-function tail and periodic-image contributions.

• Wave-function Galerkin convergence. For plane waves with one-particle basis size (M), cutoff (\Lambda_m), and (H^{2}) initial data, the (L^{2}) Galerkin error bound is [ \mathcal{O}!\left(\frac{T^{2/3}\eta^{4}}{\Lambda_m^{4/3}}\right)

  \mathcal{O}!\left(\frac{T^{2/3}\eta^{4}\ell^{4/3}}{M^{4/9}}\right). ] The finite-basis projection error alone has a (\Lambda_m^{-2}) bound; the slower (4/3)-order term comes from the dynamical projection error.

• Sharp ground-state energy rate. At fixed cell size, the Galerkin ground-state energy error converges as (\mathcal{O}(M^{-1})). The paper states that this rate is sharp in general, matching one-body asymptotics, earlier expectations, and numerical evidence for the homogeneous electron gas.

• Whole-space energy bound. Combining Galerkin and periodization estimates gives [ |E_{\ell,K}-E_{\infty}| \leq C\left(\frac{\eta^{2}}{\ell^{3}}+\frac{\eta^{27/4}}{\Lambda_m^{3}}\right) \leq C\left(\frac{\eta^{2}}{\ell^{3}}+\frac{\eta^{27/4}\ell^{3}}{M}\right). ] The fixed-cell (M^{-1}) rate does not imply (M=\mathcal{O}(\varepsilon^{-1})) for the whole-space problem; under the stated assumptions the sufficient choices are (\ell=\mathcal{O}(\varepsilon^{-1/3})), (\Lambda_m=\mathcal{O}(\varepsilon^{-1/3})), and therefore (M=\mathcal{O}(\varepsilon^{-2})).

• Dynamics cannot uniformly match the energy rate. The paper proves that a uniform (M^{-1}) rate, equivalently a uniform (\Lambda_m^{-3}) error bound at fixed cell size, is impossible for the real-time (L^{2}) wave-function error with general (H^{2}) initial data at a fixed positive time, including in the fermionic subspace.

• Spatial parameter choices differ between tasks. For charge-neutral fixed-nuclei systems, Table 1 reports:

  • Real-time (L^{2}) state error: (\ell=\mathcal{O}(T\eta^{3/2}\varepsilon^{-1})), (\Lambda_m=\mathcal{O}(T^{1/2}\eta^{3}\varepsilon^{-3/4})), (M=\mathcal{O}(T^{9/2}\eta^{27/2}\varepsilon^{-21/4})).
  • Ground-state energy: (\ell=\mathcal{O}(\eta^{2/3}\varepsilon^{-1/3})), (\Lambda_m=\mathcal{O}(\eta^{9/4}\varepsilon^{-1/3})), (M=\mathcal{O}(\eta^{35/4}\varepsilon^{-2})).

• End-to-end logical gate costs. Table 2 reports leading polynomial dependence for charge-neutral fixed-nuclei systems:

  • Qubitization: real-time (\widetilde{\mathcal{O}}(T^{2}\eta^{8}\varepsilon^{-3/2})); ground-state energy (\widetilde{\mathcal{O}}(\eta^{13/2}\varepsilon^{-5/3})).
  • Interaction picture: real-time (\widetilde{\mathcal{O}}(T^{3/2}\eta^{6}\varepsilon^{-3/4})); ground-state energy (\widetilde{\mathcal{O}}(\eta^{21/4}\varepsilon^{-4/3})).
  • (k)-th order Trotterization: real-time (\widetilde{\mathcal{O}}_k(T^{2+3/(2k)}\eta^{7+5/k}\varepsilon^{-3/2-7/(4k)})); ground-state energy (\widetilde{\mathcal{O}}_k(\eta^{11/2+17/(4k)}\varepsilon^{-5/3-4/(3k)})).

• Precision accounting. A tolerance (\varepsilon) is assigned separately to periodization, Galerkin, and finite-dimensional algorithmic errors. The resulting guarantees are a whole-space (L^{2}) state error of at most (3\varepsilon) for dynamics and an absolute energy error of at most (3\varepsilon) with constant success probability. Input-state preparation and final measurement costs are excluded.

Methodology in Plain English

The authors replace the whole-space problem with two successive finite approximations and bound the error introduced at each step. First, they place the system in a periodic cubic cell of side length (\ell), which introduces periodic replicas and a finite-domain truncation error. Second, they keep finitely many plane waves, indexed by a momentum cutoff (\Lambda_m), with (M) one-particle basis functions and a (K=M^{\eta})-dimensional many-body space.

For the periodization step, they use the mean-zero periodic Green function for the Coulomb kernel, together with localization and first-moment propagation bounds, to control both the part of the wave function outside the cell and the error from periodic images. For the Galerkin step, they split the error into the state projection error and the dynamical projection error. Rather than estimating the dynamical error only through its forcing term, they keep the differential inequality for its squared norm and use an inner-product argument that retains dependence on the error itself. Many-body Hardy-type inequalities and Sobolev norm estimates, with explicit polynomial dependence on system size, close the resulting nonlinear inequality.

For ground-state energies, they use the variational principle to separate Galerkin and periodization energy errors. A cancellation removes the term linear in the omitted component, and the eigenvalue equation relates the omitted modes to the high-frequency part of the potential acting on the ground state. A Coulomb tail estimate supplies the remaining control needed for the sharp cutoff dependence.

Finally, they choose (\ell), (\Lambda_m), and (M) as functions of the target precision, particle number, time, and physical scales, and insert these choices into first-quantized plane-wave algorithms based on qubitization, the interaction picture, and high-order Trotterization. Nested-commutator estimates for the Galerkin-compressed Hamiltonian determine the Trotter step count.

Why This Matters

Impact on research. Algorithmic costs for quantum chemistry are normally quoted in terms of a finite basis size. This paper connects those costs back to the continuum problem, showing that the basis size must be chosen from the target whole-space accuracy and that the choice differs between real-time dynamics and ground-state energy estimation. It also shows that the dynamical projection error, often omitted, can set the convergence rate for Coulomb systems. The framework provides a template for rigorous spatial error analysis of other unbounded Hamiltonians.

Real-world applications. The paper lists the following application areas for many-body Coulomb simulation:

• Electronic structure and chemical dynamics.
• Materials science.
• Reaction mechanisms and battery materials.
• Stopping-power calculations for inertial fusion.

Industry relevance. The results are directly relevant to resource estimation

Authors’ abstract

Nonrelativistic quantum chemistry is based on the Schrödinger equation in continuous, unbounded space. Computational approaches on classical and quantum computers often truncate the spatial domain and introduce a finite-dimensional basis, reducing the continuum problem to a finite-dimensional one. There is therefore a fundamental question: to what extent does this approximation capture the original continuum problem? In this work, we develop a general Sobolev-Galerkin framework for estimating spatial discretization errors in quantum simulation with unbounded Hamiltonians. We apply this framework to many-body Coulomb systems and establish both periodization and Galerkin error estimates for wave-function dynamics and ground-state energies. For dynamics, the finite-basis projection error of the exact wave function does not capture the full basis-discretization error: an additional dynamical projection error enters our bound at a different rate, and we introduce an inner-product argument to control this contribution. Interestingly, the spatial convergence rates for wave-function dynamics and ground-state energies differ quantitatively. For ground-state energies, we establish a sharp $M^{-1}$ convergence rate at fixed cell size, where $M$ is the number of one-particle basis functions. We also show that a uniform $M^{-1}$ convergence rate is impossible for the corresponding real-time wave-function error with general $H^2$ initial data. The analysis relies on many-body Sobolev estimates with explicit polynomial dependence on system size. Combining these estimates with state-of-the-art quantum algorithms, we obtain, to our knowledge, the first rigorous end-to-end complexity bounds for quantum simulation of molecular and electronic dynamics with all spatial errors explicitly controlled. The number of basis functions is determined directly from the continuum problem, without regularization.