Periodic Reporting for period 1 - DQC (Diagrammatic Quantum Computation)
Reporting period: 2022-09-01 to 2024-08-31
The idea that quantum-mechanical computers might outperform classical computers was introduced decades ago. However, it was only late last year that a quantum computer was able to perform a computation that would be intractable on a classical computer. Still, current-generation quantum computers cannot yet solve useful practical problems due to technological limitations, such as the small number of available qubits and the limited maximum number of operations that can be executed before decoherence occurs. While these bounds can be improved by advances in technology, it is equally crucial that we use existing machines to their full extent by running desired computations as efficiently as possible. Doing so will shorten the time to having quantum computers that solve useful problems. This project will build a compiler that uses novel graphical methods to optimise quantum computations in a threefold manner. Firstly, we will develop theoretical insights and practical implementations that help optimise the number of resources needed for a given computation. Secondly, we will find ways to verify the correctness of these optimisations, and lastly we will build a classical simulator to test quantum computations.
In contrast to previous work, which has treated optimisation, verification, and classical simulation as distinct problems in quantum software, this project will advance a new unified approach, revealing previously unforeseen connections and applying the same core techniques to all three problems.
In contrast to previous work, which has treated optimisation, verification, and classical simulation as distinct problems in quantum software, this project will advance a new unified approach, revealing previously unforeseen connections and applying the same core techniques to all three problems.
The project was cut short to a period of 3 months, due to the start of a permanent position by the researcher. However, researcher will continue working in the area of the project.
Prior to the start of the project, significant progress was made in the area of the project:
* We developed a new optimizing quantum compiler, quizx, that is a Rust version or our Python library PyZX. Quizx can be up to a thousand times faster than PyZX.
* We developed new ways to classically simulate quantum computations. These use the stabiliser decomposition approach, and could simulate up to 20 times larger circuits than before.
* We made progress in compiling qutrit unitaries, nearly leading to a full characterisation of which unitaries can be exactly synthesised in the Clifford+T gate set.
* We've proven that the general problem of extracting a quantum circuit from an arbitrary ZX-diagram, the type of structure that this project deals with, is a hard problem. This confirms our suspicion that more specialized methods are necessary.
Prior to the start of the project, significant progress was made in the area of the project:
* We developed a new optimizing quantum compiler, quizx, that is a Rust version or our Python library PyZX. Quizx can be up to a thousand times faster than PyZX.
* We developed new ways to classically simulate quantum computations. These use the stabiliser decomposition approach, and could simulate up to 20 times larger circuits than before.
* We made progress in compiling qutrit unitaries, nearly leading to a full characterisation of which unitaries can be exactly synthesised in the Clifford+T gate set.
* We've proven that the general problem of extracting a quantum circuit from an arbitrary ZX-diagram, the type of structure that this project deals with, is a hard problem. This confirms our suspicion that more specialized methods are necessary.
The project was cut short to a period of 3 months, due to the start of a permanent position by the researcher. However, researcher will continue working in the area of the project. Prior to the start of the project, significant progress was made in the area of the project. In particular, we improved upon the state-of-the-art of simulation using the stabiliser decomposition formalism.