Low Depth Quantum Simulation of Electronic Structure

Ryan Babbush, Nathan Wiebe, Jarrod R. McClean, James McClain, Hartmut Neven, Garnet Kin‐Lic ChanarXiv 2017 · DOI · arXiv

Abstract

Quantum simulation of the electronic structure problem is one of the most researched applications of quantum computing. The majority of quantum algorithms for this problem encode the wavefunction using $N$ Gaussian orbitals, leading to Hamiltonians with ${\cal O}(N^4)$ second-quantized terms. We avoid this overhead and extend methods to the condensed phase by utilizing a dual form of the plane wave basis which diagonalizes the potential operator, leading to a Hamiltonian representation with ${\cal O}(N^2)$ second-quantized terms. Using this representation we can implement single Trotter steps of the Hamiltonians with linear gate depth on a planar lattice. Properties of the basis allow us to deploy Trotter and Taylor series based simulations with respective circuit depths of ${\cal O}(N^{7/2})$ and $\widetilde{\cal O}(N^{8/3})$ for fixed charge densities - both are large asymptotic improvements over all prior results. Variational algorithms also require significantly fewer measurements to find the mean energy in this basis, ameliorating a primary challenge of that approach. We conclude with a proposal to simulate the uniform electron gas (jellium) using a low depth variational ansatz realizable on near-term quantum devices. From these results we identify simulations of low density jellium as a promising first setting to explore quantum supremacy in electronic structure.

Introduction

Simulating the behavior of electrons in molecules and materials—the electronic structure problem—stands as a primary objective for the field of quantum computing. Accurate simulations could revolutionize material science and drug discovery. However, a significant hurdle remains: the computational complexity of existing algorithms. Most current approaches utilize Gaussian orbitals as a basis set. While effective for describing localized atomic systems, Gaussians result in a Hamiltonian with $\mathcal{O}(N^4)$ interaction terms, where $N$ represents the number of orbitals.

This $\mathcal{O}(N^4)$ scaling creates a massive bottleneck. As researchers attempt to simulate larger or more complex systems, the number of required quantum gates grows exponentially, resulting in "deep" circuits that exceed the coherence times of near-term quantum hardware. This paper addresses this challenge by introducing a "plane wave dual basis" that simplifies the Hamiltonian representation, drastically reducing the required circuit depth.

Methods

The authors propose a shift from Gaussian orbitals to a plane wave basis and its "dual" form. This transition is motivated by the need for a sparser Hamiltonian representation.

Plane Wave Dual Basis

By employing a basis that is essentially a discrete Fourier transform of plane waves, the authors demonstrate that the potential energy operator becomes diagonal. This crucial step reduces the number of terms in the Hamiltonian from $\mathcal{O}(N^4)$ to $\mathcal{O}(N^2)$. This reduction is applicable to both periodic systems and, with appropriate adjustments, isolated molecules.

Fermionic Fast Fourier Transform (FFFT)

To efficiently handle the kinetic energy operator—which is diagonal in the original plane wave basis but not in the dual basis—the authors developed the Fermionic Fast Fourier Transform (FFFT). This quantum circuit allows the system to "rotate" between the basis where potential energy is easily calculated and the basis where kinetic energy is diagonal. This dual-basis approach keeps the overall circuit depth low.

Hardware Mapping

A key practical contribution is the demonstration of mapping these fermionic operators onto a 2D planar lattice of qubits. This architecture is common in many superconducting quantum chips. The authors show that this mapping can be achieved with only linear gate depth, $\mathcal{O}(N)$, without the need for additional ancilla qubits.

Results and Discussion

The application of the plane wave dual basis to major simulation strategies—Trotter-Suzuki decomposition, Taylor series-based evolution, and Variational Quantum Eigensolvers (VQE)—yields significant asymptotic improvements.

Comparison of asymptotic complexity for various quantum simulation algorithms of electronic structure.

As shown in Figure 1, the proposed methods outperform all prior results in terms of total circuit depth. For fixed charge densities, the Trotter-Suzuki approach achieves a depth of $\mathcal{O}(N^{7/2})$, a quadratic improvement. The Taylor series-based method achieves $\tilde{\mathcal{O}}(N^{8/3})$, making it the most efficient time-evolution algorithm currently available in the literature.

Beyond time-evolution, the dual basis also enhances the efficiency of VQE. In variational algorithms, the number of measurements required to estimate total energy is a major practical hurdle. In the dual basis, this scaling is reduced to $\mathcal{O}(N^2/\mu^2)$ for materials in the thermodynamic limit, where $\mu$ is the relative error.

The Path to Quantum Supremacy: Jellium

The paper identifies the Uniform Electron Gas (jellium) model as a prime candidate for demonstrating quantum supremacy. Jellium is a fundamental model in condensed matter physics where electrons move in a uniform background of positive charge. At low densities, jellium becomes extremely difficult for classical Quantum Monte Carlo methods due to the "fermion sign problem." This makes it a "killer app" for quantum computers, as they can simulate these strongly correlated phases (like Wigner crystals) where classical methods fail.

Conclusion and Future Work

While the plane wave basis may require more basis functions ($N$) than Gaussians for small, specific molecules, its superior scaling makes it the preferred choice as system size increases. Future research will focus on extending these methods to more realistic, non-uniform materials by incorporating pseudopotentials. Additionally, empirical testing of the proposed VQE ansatz for jellium on real-world hardware will be essential to determine the practical number of Trotter layers required to surpass classical benchmarks.

Summary updated

Abstract

Quantum simulation of the electronic structure problem is one of the most researched applications of quantum computing. The majority of quantum algorithms for this problem encode the wavefunction using $N$ Gaussian orbitals, leading to Hamiltonians with ${\cal O}(N^4)$ second-quantized terms. We avoid this overhead and extend methods to the condensed phase by utilizing a dual form of the plane wave basis which diagonalizes the potential operator, leading to a Hamiltonian representation with ${\cal O}(N^2)$ second-quantized terms. Using this representation we can implement single Trotter steps of the Hamiltonians with linear gate depth on a planar lattice. Properties of the basis allow us to deploy Trotter and Taylor series based simulations with respective circuit depths of ${\cal O}(N^{7/2})$ and $\widetilde{\cal O}(N^{8/3})$ for fixed charge densities - both are large asymptotic improvements over all prior results. Variational algorithms also require significantly fewer measurements to find the mean energy in this basis, ameliorating a primary challenge of that approach. We conclude with a proposal to simulate the uniform electron gas (jellium) using a low depth variational ansatz realizable on near-term quantum devices. From these results we identify simulations of low density jellium as a promising first setting to explore quantum supremacy in electronic structure.
BibTeX
@misc{babbush2017low,
  author = {Ryan Babbush and Nathan Wiebe and Jarrod R. McClean and James McClain and Hartmut Neven and Garnet Kin‐Lic Chan},
  title = {{Low Depth Quantum Simulation of Electronic Structure}},
  year = {2017},
  eprint = {1706.00023},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1706.00023},
}
Figure