Pulse-optimised circuit elements for scalable and noise-resilient quantum chemistry

Abstract

Useful chemistry calculations on near-term quantum processors are hindered by current algorithmic runtimes. We develop a methodology to significantly reduce these runtimes. Typically, variational quantum eigensolver (VQE) algorithms are implemented as sequences of primitive gates. Our methodology instead relies on gradient-ascent pulse engineering to construct hardware-tailored pulses for the direct implementation of VQEs. As problem sizes increase, it quickly becomes intractable to optimise a pulse that implements an entire VQE ansatz circuit. However, leading VQEs are constructed in a modular fashion. A problem-tailored VQE is assembled from parameterised circuit elements that simulate hopping between two or four electronic spin orbitals. We show that these circuit elements can be implemented more efficiently using hardware-tailored pulses. We numerically demonstrate our methodology on a silicon spin-qubit quantum processor. We find that common circuit elements, known as single- and double-qubit excitations, can be implemented in less than 289 ns and 927 ns, respectively. Compared with conventional gate-based implementations, our pulse-accelerated qubit excitations provide a scalable approach for faster and therefore more noise-robust quantum chemistry simulations by reducing VQE runtimes by up to a factor of 15.3.