Cloud-accessible! Japan's second superconducting quantum computer launched

Overview of the new platform for hybrid quantum computing. The processor's quantum bit control software was developed by NTT, and the platform is accessible in the cloud through serverless computing services such as AWS Lambda.

Access link:

Shintaro Sato, head of Fujitsu's quantum lab, said, "This is just the first or second step; we still have a long way to go." In August 2022, Fujitsu announced the establishment of the RIKEN RQC-Fujitsu Collaboration Center to jointly develop a quantum computer based on superconducting technology - their ultimate goal is to achieve a machine that supports 1,000 quantum bits.

The company claims that the system is capable of realizing 264 quantum superposition and entanglement states, which allows the computer to perform calculations that are extremely difficult for classical computers. Nonetheless, the system will run alongside a classical computer running quantum simulations in order to control it.

Fujitsu explains, "The quantum simulator is able to digitally mimic quantum computation, bridging the gap for the development of practical fault-tolerant quantum computation (FTQC)."

"Quantum computers are very tricky devices that require a lot of error-correction work to be useful; fault tolerance enables computers to operate in a meaningful way, which is a necessity, not a luxury."

For now, pairing a quantum computer with a high-performance computing cluster that simulates 40 quantum bits should help scientists assess the system's ability to reliably generate accurate results.

The problem, Fujitsu explains, is that noise-laden intermediate-scale quantum (NISQ) computers produce computational errors due to the noise of their surroundings. Quantum simulation does not suffer from the same errors because the quantum bits are not real and therefore are not disturbed by noise. The disadvantage of quantum simulation is that it is slower than the real thing and it is not the real thing.

This is not to say that simulations and this system are not useful. RIKEN and Fujitsu claim that hybrid systems have been shown to be more accurate when applying quantum algorithms to chemical calculations; these calculations are typical HPC workloads, which are prime candidates for quantum acceleration in the near future.

One experiment involving such a hybrid device calculated the ground-state energy of a molecule containing 12 hydrogen atoms. Using a combination of quantum algorithms, density matrix embedding theory (a method of breaking up large molecules into smaller pieces) and artificial intelligence models to help mitigate the effects of noise, Fujitsu and the RIKEN Institute of Science and Technology in Japan said they were able to complete these energy calculations with greater precision than they could using classical algorithms alone.

As the system is deployed, Fujitsu and RIKEN are opening it up to outside companies and organizations, including Fujifilm, Tokyo Electron, Mizuho Financial Technologies and Mitsubishi Chemical, to conduct joint research.

--With this new machine, these researchers will have the opportunity to see how their algorithms perform on an actual quantum processor.