Progress Report
Development of Large-scale Fault-tolerant Universal Optical Quantum Computers[5] Research and development on cloud computing and operation of optical quantum computers
Progress until FY2024
1. Outline of the project
A scalable optical quantum computer can be realized in a compact setup using time-domain multiplexing techniques. Optical quantum computers are considered a promising candidate for realizing practical quantum computers. In collaboration with the R&D themes 1 and 4, we are engaged in research and development for the public release and operation of a cloud-based optical quantum computer. In order to perform arbitrary “computations” on an optical quantum computer, it is necessary not only to build the hardware system, but also to develop a mechanism that allows external users to freely operate the system and enables software developers to easily utilize it. Additionally, we are constructing a platform to be publicly released on the cloud for widespread social implementation.
2. Outcome so far
We have developed a system for publishing the optical quantum computer developed by the R&D theme 4 on the cloud, as shown in Figure 1.
When making the system available on the cloud, we considered future expandability and usability and implemented a multi-layered system design, shown in Figure 2. In this optical quantum computer, the most abstract programming model is a continuous quantum circuit. This can be described in the same way as a normal Python program through an SDK (software development kit) installed on the user's computer. The input program is transferred to the cloud server, converted into a format executable on the actual machine, and passed to the optical quantum computer in the laboratory. The processing results are returned to the user in reverse order.
This optical quantum computer generates a two-dimensional cluster state using time domain multiplexing techniques. The method of describing quantum computations using this state is called “graph representation.” Continuous quantum circuits constructed using the SDK are automatically converted into graph representations, shown in Figure 3 and then converted into parameters for direct operation of the actual machine before being sent to it.
We have also developed a framework for describing nonlinear feedforward operations required for next-generation actual machines. We are also developing a simulator so that users can easily try out programs to be implemented in actual machines.
3. Future plans
We will develop software in alignment with the development of next-generation actual machines and verify its functionality. We plan to refine visualization methods to enable easier understanding when creating complex circuits and graphs, and to enable editing of circuits and graphs in the GUI. We will also continue to improve functionality necessary for more practical cloud services. Through these efforts, we will establish an environment for applied research using optical quantum computers, such as optimization problems and neural networks.