Quantum computing promises to help us understand chemistry in its purest form – ultimately leading to a better understanding of everything from drug development to superconductors. But before we can do any of that, researchers in computational quantum chemistry have to create the basic building blocks for understanding a chemical system: they must prepare the initial state of a system, apply various effects to the system through time, then measure the resulting output.
The first problem, called “state preparation” is a tricky one – researchers have been leaning heavily on “variational” techniques to do this, but those techniques come with huge optimization costs in addition to serious scaling issues for larger systems. An older technique, called “adiabatic state preparation” promises significant speedups on quantum computers vs classical computers, but has been mostly abandoned by researchers because the typical method used for time evolution is costly and introduces too much noise. This method, called “Trotterized adiabatic time evolution”, involves splitting up time into discrete steps, which requires many, many gates, and ultimately needs error rates well out of reach for any near-term quantum computer.
Recently, researchers at Quantinuum found a way around that roadblock – they eliminated the noisy time evolution in favor of a clever averaging approach. Rather than taking a bunch of discrete time steps they simulate different interactions such that on average you get exactly the right time evolution. A nice aspect of this approach is that it has guaranteed “convergence” – ultimately this means that, unlike other approaches, it works all the time. This new approach has also been shown to be possible on near-term quantum computers: it does not require too many gates or computational time, and it scales well with the system size.
This algorithm is designed with Quantinuum’s world-leading hardware in mind, as it requires all-to-all connectivity. Combined with our industry-leading gate fidelities, this new approach is opening the door to many fascinating applications in chemistry, physics, and beyond.