How to improve the relative performance of variational algorithms, with strategies for initializing and optimizing larger parameterized quantum circuits.
TITLE: Pushing the capabilities of variational quantum algorithm
TOPIC: Noisy Intermediate-Scale Quantum Computers (NISQ)
SPEAKER: Dr Sukin (Hannah) Sim
AFFILIATION: Zapata Computing, Boston, USA
ABSTRACT:
Variational quantum algorithms are believed to be a promising direction towards achieving quantum advantage in the near-term. However, the bare variational algorithm framework, as initially proposed, is not likely to provide advantage. Fortunately, in recent years, various improvements and modifications have been proposed to extend the capabilities of these near-term algorithms.
In this talk, I will discuss ways to improve relative performance of variational algorithms, namely presenting expressibility as a quantity that can be used to compare among parameterized quantum circuits (PQCs) and introducing strategies for initializing and optimizing larger parameterized quantum circuits. We show that expressibility can be used to rule out unfavorable circuit designs when selecting an ansatz for a variational algorithm.
We additionally demonstrate the capabilities of the Flexible Initializer for Parameterized Quantum Circuits (FLIP), which uses meta-learning for initializing large PQCs, as well as the Parameter-Efficient Circuit Training (PECT) method, which enables optimization of PQCs with hundreds or even thousands of parameters.
https://www.zapatacomputing.com/team/sukin-sim-hannah/
https://www.zapatacomputing.com/
HOSTED BY: Dr Mária Kieferová, Centre for Quantum Software and Information, University of Technology Sydney, Australia.
https://profiles.uts.edu.au/maria.kieferova
