What the foundations of quantum computer science teach us about chemistry

Jarrod R. McClean, Nicholas C. Rubin, Joonho Lee, Matthew P. Harrigan, Thomas E. O’Brien, Ryan Babbush, William J. Huggins, Hsin-Yuan HuangThe Journal of Chemical Physics 2021 · DOI

TL;DR

This paper argues that quantum computing's impact on chemistry goes beyond faster simulations. By framing chemical simulations as "digital experiments," we can leverage computer science's fundamental limits (like undecidability) to understand why traditional, data-driven chemical models are so successful and how quantum computers can further enhance these "learning models."

What Problem Does This Paper Address?

The paper addresses the question of how quantum computing will revolutionize chemistry. While the promise of exponential speedups for direct chemical simulations is well-known, the authors argue that the true impact lies deeper. They explore how fundamental results from quantum computer science, even before full-scale quantum computers exist, offer profound insights into the nature of chemical simulation, the limitations of predictive models, and the power of data-driven approaches in chemistry. This perspective aims to bridge the gap between theoretical computer science and practical chemical research.

Methods

The authors frame direct chemical simulation as a "digital experiment." They leverage theoretical computer science concepts, particularly the "no-fast-forwarding theorem" and the notion of "undecidable problems," to analyze the fundamental limits of simulating physical systems, including chemical ones. They then contrast this with the success of "learning models" in chemistry, which incorporate data from experiments or simulations. The paper also discusses the potential of quantum sensors and transduction for feeding data into quantum computers, enhancing these learning models.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Chemical Simulation as a Digital Experiment: This framing highlights both the power of quantum computers to extend our reach and their inherent limitations, mirroring the nature of physical experiments.
  • Undecidability in Chemistry: The paper controversially suggests that some chemical problems, like determining if a system will thermalize, are fundamentally undecidable by any algorithm. This has implications for the predictive power of thermodynamic models.
  • The Power of Data-Driven Models: The success of traditional chemical theories (like transition state theory) is attributed to their reliance on data, not just brute-force computation. Quantum computers can amplify this by providing richer data.
  • Quantum Computers as Data Providers for Learning Models: Instead of solely replacing classical simulations, quantum computers may primarily serve to generate data that empowers classical or hybrid learning models.
  • Quantum Transduction: The ability to efficiently transfer quantum states from sensors into quantum computers offers a significant advantage, potentially enabling the characterization of transient or rare systems that are intractable classically.

Key Results

The paper doesn't present new experimental or computational results in the traditional sense. Instead, it synthesizes existing theoretical results from computer science and quantum information theory and applies them to the field of chemistry. The "results" are conceptual:

  • Demonstration that quantum computers cannot "fast-forward" time beyond a certain limit, meaning they can't simulate systems infinitely faster than nature.
  • Identification of chemical problems that are provably undecidable, meaning no algorithm can solve them for all cases.
  • Argument that data-augmented models (both classical and quantum) are fundamentally more powerful than rote simulation alone.

Limitations and Future Work

The authors acknowledge that current quantum devices have error rates too high to be competitive with classical methods for many tasks. They also note that the concept of undecidability in chemistry is a controversial stance. Future work is suggested in several areas:

  • Developing more sophisticated quantum algorithms for chemistry, particularly those that leverage data.
  • Exploring the potential of quantum sensors and transduction for data acquisition.
  • Investigating the application of quantum error correction techniques to chemical systems, especially for understanding quasi-degenerate electronic states in catalysts.
  • Further exploring the interplay between quantum information theory and chemical theory, particularly in areas like quantum control and the development of novel chemical models.

Summary updated

Abstract

With the rapid development of quantum technology, one of the leading applications that has been identified is the simulation of chemistry. Interestingly, even before full scale quantum computers are available, quantum computer science has exhibited a remarkable string of results that directly impact what is possible in a chemical simulation with any computer. Some of these results even impact our understanding of chemistry in the real world. In this Perspective, we take the position that direct chemical simulation is best understood as a digital experiment. While on the one hand, this clarifies the power of quantum computers to extend our reach, it also shows us the limitations of taking such an approach too directly. Leveraging results that quantum computers cannot outpace the physical world, we build to the controversial stance that some chemical problems are best viewed as problems for which no algorithm can deliver their solution, in general, known in computer science as undecidable problems. This has implications for the predictive power of thermodynamic models and topics such as the ergodic hypothesis. However, we argue that this Perspective is not defeatist but rather helps shed light on the success of existing chemical models such as transition state theory, molecular orbital theory, and thermodynamics as models that benefit from data. We contextualize recent results, showing that data-augmented models are a more powerful rote simulation. These results help us appreciate the success of traditional chemical theory and anticipate new models learned from experimental data. Not only can quantum computers provide data for such models, but they can also extend the class and power of models that utilize data in fundamental ways. These discussions culminate in speculation on new ways for quantum computing and chemistry to interact and our perspective on the eventual roles of quantum computers in the future of chemistry.
BibTeX
@article{mcclean2021what,
  author = {Jarrod R. McClean and Nicholas C. Rubin and Joonho Lee and Matthew P. Harrigan and Thomas E. O’Brien and Ryan Babbush and William J. Huggins and Hsin-Yuan Huang},
  title = {{What the foundations of quantum computer science teach us about chemistry}},
  journal = {The Journal of Chemical Physics},
  year = {2021},
  doi = {10.1063/5.0060367},
}
Figure