Polynomial-time quantum algorithm for the simulation of chemical dynamics

Ivan Kassal, Stephen P. Jordan, Peter J. Love, Masoud Mohseni, Alán Aspuru‐Guzik

Proceedings of the National Academy of Sciences · 2008 · 394 citations · 44 references

DOIFull text

Open access

Concepts

TL;DR

Exact quantum simulation on classical computers scales exponentially with system size, restricting practical applications to very small molecules. Our algorithm employs a split‑operator method that simulates all electron‑nuclear and interelectronic interactions in quadratic time, while efficiently preparing chemically relevant states and extracting observables such as transition probabilities and thermal rates. The approach enables polynomial‑time, exact simulation of chemical reactions, outperforming the Born–Oppenheimer approximation for systems larger than four atoms and surpassing classical performance with as few as 100 qubits.

Abstract

The computational cost of exact methods for quantum simulation using classical computers grows exponentially with system size. As a consequence, these techniques can be applied only to small systems. By contrast, we demonstrate that quantum computers could exactly simulate chemical reactions in polynomial time. Our algorithm uses the split-operator approach and explicitly simulates all electron-nuclear and interelectronic interactions in quadratic time. Surprisingly, this treatment is not only more accurate than the Born–Oppenheimer approximation but faster and more efficient as well, for all reactions with more than about four atoms. This is the case even though the entire electronic wave function is propagated on a grid with appropriately short time steps. Although the preparation and measurement of arbitrary states on a quantum computer is inefficient, here we demonstrate how to prepare states of chemical interest efficiently. We also show how to efficiently obtain chemically relevant observables, such as state-to-state transition probabilities and thermal reaction rates. Quantum computers using these techniques could outperform current classical computers with 100 qubits.

References

44