Quantum computers draw their power from their ability to use information that is in a superposition of states. Problems are solved sequentially in classical computing. Quantum computing lets you build representations of your data that simultaneously hold many values at once. So when you solve your problem, you’re actually solving many problems all at the same time.

This innate parallelism is very powerful for constrained optimization problems — problems where you can count the possible solutions, but the number of them grows extremely rapidly with the size of the problem. Often, this sort of problem has a reasonably straightforward solution, but it is computationally intractable because we just can’t process all the solutions fast enough. Quantum computers will be an enormous help here.

Quantum computers can also help with machine learning problems in a similar way. Many machine learning problems make connections between inputs and “learn” how strong those connections should be by looking at pairs of known inputs and outputs. But the sort of connections we can make and compute on classical computers is limited — this is one reason you’re smarter than your computer. The neurons in your brain are highly connected. Quantum computers will let us exploit a richer set of connections and therefore build more sophisticated and powerful models.

Optimization algorithms occur often in the analysis of HEP data, such as fitting waveforms from particle detectors to templates to determine particle energies and finding the best set of candidate paths that are closest to a particle’s true trajectory. Fermilab is starting with determining quantum algorithms for solving very basic problems (such as the traveling salesman), and then scaling them up to eventually be applicable to the kinds of problems that need to be solved in HEP.

The current noisy quantum computers are too limited to solve problems directly from HEP, but by starting with basic problems, we will learn how the scale as quantum computers become more powerful and less noisy.

Lockheed Martin is a partner on this initiative.