Quantum Chemistry
STAQ Collaboration
Our group leads the applications component of STAQ (Software-Tailored Architecture for Quantum co-design), a seven-university collaboration to design quantum processing hardware together with software interfacing and target applications, all co-optimized to demonstrate quantum advantage in a practical problem.
Hamiltonian term reduction
Variational quantum eigensolvers (VQE) are one of the premier applications for existing and near-term quantum computers (noisy, intermediate-scale quantum computers). The quantum component of a VQE procedure involves separately estimating the expectation values of the terms of some Hamiltonian of interest. Thus, the complexity of the procedure depends on the number of terms in the Hamiltonian (among other things).
We have studied techniques for reducing the number of of terms in Hamiltonians that appear in quantum chemistry and related contexts, and thus reducing the complexity of the associated VQE procedures: the results of this project may be found in arXiv:1908.08067.
Optimization and benchmarking
For quantum simulation to be a viable approach to solving problems in quantum chemistry, a great deal of quantum circuit optimization will be necessary. Our group investigates various avenues for this task. Recent work has included a focus on optimal ordering schemes when performing Trotterization, and on benchmarking the Bravyi-Kitaev mapping as an alternative to the canonical Jordan-Wigner mapping for representing occupation states of fermionic spin-orbitals with qubits. It is known that molecular systems show a greatly variable response to optimization schemes depending on the physical characteristics of the system. As such, our work puts a strong emphasis on the use of large data sets to examine this behaviour.
Novel simulation methods
As we enter the NISQ era, a major source of optimization will be the distribution of computational work between quantum and classical components of the simulation algorithm. Work within the group focuses on reducing the computational load of the quantum device through maximally using the classical device to perform any "classical parts" of the algorithm.
The intersection of these ideas with our work on contextuality has led to our new quantum-classical hybrid algorithm, contextual subspace variational quantum eigensolver, in which a Hamiltonian of interest is partitioned into a noncontextual part that is simulated classically and a contextual correction that is computed on a quantum processor. This reduces the number of qubits required. See https://arxiv.org/abs/2011.10027 for the theory, and https://github.com/wmkirby1/ContextualSubspaceVQE to download our code and simulate or use the method yourself!