Compact wavefunctions from compressed imaginary time evolution

Jarrod R. McClean, Alán Aspuru-GuzikarXiv 2014 · DOI · arXiv

Abstract

Simulation of quantum systems promises to deliver physical and chemical predictions for the frontiers of technology. Unfortunately, the exact representation of these systems is plagued by the exponential growth of dimension with the number of particles, or colloquially, the curse of dimensionality. The success of approximation methods has hinged on the relative simplicity of physical systems with respect to the exponentially complex worst case. Exploiting this relative simplicity has required detailed knowledge of the physical system under study. In this work, we introduce a general and efficient black box method for many-body quantum systems that utilizes technology from compressed sensing to find the most compact wavefunction possible without detailed knowledge of the system. It is a Multicomponent Adaptive Greedy Iterative Compression (MAGIC) scheme. No knowledge is assumed in the structure of the problem other than correct particle statistics. This method can be applied to many quantum systems such as spins, qubits, oscillators, or electronic systems. As an application, we use this technique to compute ground state electronic wavefunctions of hydrogen fluoride and recover 98% of the basis set correlation energy or equivalently 99.996% of the total energy with 50 configurations out of a possible 10^7. Building from this compactness, we introduce the idea of nuclear union configuration interaction for improving the description of reaction coordinates and use it to study the dissociation of hydrogen fluoride and the helium dimer.

Introduction

The "curse of dimensionality" remains the central challenge in quantum chemistry and physics. As the number of particles in a system increases, the Hilbert space required to describe the wavefunction exactly grows exponentially. Traditional approximation methods, such as Hartree-Fock or Density Matrix Renormalization Group, often require significant system-specific tuning to remain tractable. This research introduces a "black box" approach designed to find the sparsest possible representation of a wavefunction without prior physical intuition of the specific system.

The MAGIC Algorithm

The core of this approach is the Multicomponent Adaptive Greedy Iterative Compression (MAGIC) scheme. MAGIC bridges the gap between physical evolution and information-theoretic compression by combining two distinct techniques:

  1. Imaginary Time Evolution: A standard physical technique where a trial wavefunction evolves under the operator $e^{-H\tau}$, naturally filtering out high-energy states to converge toward the ground state.
  2. Compressed Sensing (Orthogonal Matching Pursuit): To prevent the wavefunction from expanding into an unmanageable number of terms, MAGIC employs Orthogonal Matching Pursuit (OMP). This greedy algorithm iteratively selects the single most significant new configuration—one that maximizes overlap with the ideal evolved state—and adds it to the representation.
The MAGIC iterative loop: Expansion through imaginary time evolution, Compression via Orthogonal Matching Pursuit, and final Variational Relaxation.

When applied to electronic systems using non-orthogonal Slater determinants, the method is termed NOMAGIC. To ensure smooth energy curves across reaction coordinates (e.g., bond stretching), the authors also developed Nuclear Union Configuration Interaction (NUCI), which pools compact configurations from multiple points along a path.

Benchmarking and Results

The NOMAGIC method was evaluated on challenging molecular benchmarks: the Helium Dimer ($He_2$), characterized by weak dispersive forces, and Hydrogen Fluoride ($HF$), a classic case for studying homolytic bond breaking.

The results demonstrate unprecedented efficiency. For Hydrogen Fluoride, NOMAGIC achieved accuracy comparable to a standard CISDT calculation—which requires 36,021 determinants—using only 24 determinants. This represents a 1,500-fold reduction in complexity. As shown in the performance comparison, NOMAGIC reaches near-zero error with orders of magnitude fewer determinants than standard Configuration Interaction (CI) methods.

Error in energy vs. the number of determinants (log scale). NOMAGIC (red/gold) achieves high accuracy with significantly fewer determinants than standard CI methods (gray/green).

With just 50 configurations, the method recovered 98% of the correlation energy in a space where the total possible configurations exceeded $10^7$. Furthermore, NOMAGIC correctly captured the dissociation curves for both $He_2$ and $HF$. In the case of $HF$, a mere 32 determinants produced a curve indistinguishable from the exact Full Configuration Interaction (FCI) result, whereas standard Restricted Hartree-Fock (RHF) fails to describe the dissociation qualitatively.

Conclusion

This work highlights that physical ground states are inherently sparse in the appropriate basis. By leveraging non-orthogonality and greedy optimization, MAGIC provides a path toward highly efficient quantum simulations. While future work will address scaling for larger systems and the exploitation of molecular symmetries, the current results establish a powerful new tool for finding compact, high-fidelity wavefunctions in a black-box manner.

Summary updated

Abstract

Simulation of quantum systems promises to deliver physical and chemical predictions for the frontiers of technology. Unfortunately, the exact representation of these systems is plagued by the exponential growth of dimension with the number of particles, or colloquially, the curse of dimensionality. The success of approximation methods has hinged on the relative simplicity of physical systems with respect to the exponentially complex worst case. Exploiting this relative simplicity has required detailed knowledge of the physical system under study. In this work, we introduce a general and efficient black box method for many-body quantum systems that utilizes technology from compressed sensing to find the most compact wavefunction possible without detailed knowledge of the system. It is a Multicomponent Adaptive Greedy Iterative Compression (MAGIC) scheme. No knowledge is assumed in the structure of the problem other than correct particle statistics. This method can be applied to many quantum systems such as spins, qubits, oscillators, or electronic systems. As an application, we use this technique to compute ground state electronic wavefunctions of hydrogen fluoride and recover 98% of the basis set correlation energy or equivalently 99.996% of the total energy with $50$ configurations out of a possible $10^7$. Building from this compactness, we introduce the idea of nuclear union configuration interaction for improving the description of reaction coordinates and use it to study the dissociation of hydrogen fluoride and the helium dimer.
BibTeX
@misc{mcclean2014compact,
  author = {Jarrod R. McClean and Alán Aspuru-Guzik},
  title = {{Compact wavefunctions from compressed imaginary time evolution}},
  year = {2014},
  eprint = {1409.7358},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1409.7358},
}
Figure