Researchers have used a quantum computer to model the electronic energy of a diazene isomerization (Science 2020, DOI: 10.1126/science.abb9811). The scientists and other experts say the demonstration will be a building block for more complex simulations.
Classical computers encode information using bits, which can toggle between two states: 0 or 1. That means they can only encode approximations of electrons, which exist in quantum superpositions. Quantum bits (qubits) also exist in a superposition of states, meaning that one qubit can exactly represent one electron. Exact solutions on classical computers are impossible, but quantum computers may be capable of them.
Researchers have previously modeled electronic states of simple atomic and molecular systems. Google scientists have now successfully performed Hartree-Fock simulations of diazene isomerizations using 10- and 12-qubit systems. Both were able to accurately predict transition-state energies as the molecule changed conformations.
Current quantum computers are prone to errors, so the group paired Google’s quantum computer with a classical one. Each time the circuit of qubits calculated an energy, the classical computer analyzed the results and suggested new parameters. The cycle repeated until the quantum computer settled on a minimum value, which corresponded to the ground-state energy of the molecular system. The team also applied two different checks on the calculation results in an attempt to catch and correct errors.
Quantum-computing expert James D. Whitfield of Dartmouth College and Sahil Gulania, a PhD student at the University of Southern California, wrote in an email that “the Google work is a phenomenal demonstration of their hardware” but does not show any benefit for Hartree-Fock calculations compared with classical computers. Google quantum chemist Ryan Babbush agrees. He says this work is helping computer scientists learn how to handle quantum-computing errors, which will help them perform more complex calculations in the future.