henrik-dreyer commented 4 years ago

Abstract

It is presumably easier to build ten thousand quantum computers with one hundred qubits than building one quntum computer with a million qubits. Fortunately, the leading candidate for Quantum Chemistry calculations, the Variational Quantum Eigensolver (VQE), could profit massively from this kind of parallelization. In this project, we will solve some easy molecules faster using parallelization.

Description

Quantum Phase Estimation has been historically the first algorithm proposed to compute ground state energies. The long coherence time required, however, put this algorithm out of reach for near-term Quantum devices. The Variational Quantum Eigensolver trades one long coherent evolution for many shorter bits that can individually be run on quantum chips. If they are run on a single processor, these runs have to be performed sequentially, leading to runtimes that way exceed Phase Estimation. Since these short evolution consist of preparing and measuring the same state over and over again, these circuits can be run completely independently. In the limiting case of optimal parallelization, even large molecules can be solved with around 100 qubits. We are going to build a parallelization module for the Qiskit stack. Since currently we don't have access to many quantum computers, we will parallelize the VQE on a single chip. Doing so will hopefully give us interesting information about the amount of crosstalk between different quantum registers on current IBM devices.