Quantum Memory for Variational Quantum Algorithms.
Navn på bevillingshaver
Lasse Bjørn Kristensen
Beløb
DKK 700,000
År
2021
Bevillingstype
Internationalisation Fellowships
Hvad?
Quantum computing promises to allow problems to be solved which would be intractable even on conventional supercomputers. A popular class of algorithms for the current noisy small-scale machines are variational algorithms, where the quantum computer runs short parametrized computations, and a more dependable classical computer keeps track of the parameters and optimizes them until the quantum computer performs the desired operation. An interesting question is how the results from this framework could be extended once less noisy hardware becomes available by moving the parameters from the classical computer into a quantum state. This project aims to investigate how such quantum memory can be used for variational quantum algorithms, and what benefits would arise from it.
Hvorfor?
Variational quantum algorithms are recieving a lot of attention from the quantum communty, in part due to their simplicity and flexibility, which allows them to be applied to many problems. However, their capacity and the ability to optimize their parameters may be limited by the classical ressources involved. When moving towards larger and less noisy quantum computers, it therefore seems prudent to look for a way to both keep the insights and understandings currently being developed, while at the same time taking advantage of the increased memory capacity and optimization prowess of this future hardware. Understanding how to best do this transition will hopefully lead to a new flexible class of algorithms capable of tackling a wealth of practical problems, e.g. within materials design.
Hvordan?
Due to the current complexity of working with real quantum hardware, the bulk of the work will consist of performing numerical simulations to test the proposed algorithms utilizing quantum memory for variational algorithms. Specifically, the research will consist of constructing a proof of concept algorithm, testing it on small-scale problems, and investigating the best structure and the best learning approach for the quantum parameters. Of specific interest will be comparisons to existing variational algorithms on important parameters such as capacity to solve complex problems and ease of optimization.