Why Fujitsu is Working on both Quantum Computers and Simulators

Fujitsu and the Barcelona Supercomputing Center recently signed an agreement to collaborate using quantum and AI to develop personalized medicine. They started work in May on two research projects to exploit clinical data and advance quantum simulation technologies using tensor networks.

Developing innovative and effective quantum simulation techniques is not just important for the project, it’s integral to Fujitsu’s strategy for developing quantum computers and ensuring its customers are prepared for the advent of commercial-scale quantum.

In this Q&A, the head of the Fujitsu Research quantum laboratory Shintaro Sato and the senior director of Fujitsu Global’s HPC office Ingolf Staerk explain the goals of the project and why Fujitsu’s quantum roadmap incorporates developing quantum simulators alongside quantum computers.

Enter Quantum: What is the background to Fujitsu’s partnership with the Barcelona Supercomputing Center?

Ingolf Staerk: Fujitsu and Barcelona Supercomputing Center (BSC) have a very long history of working together in various fields. We always aim for a close customer context in Europe to understand customer problems and processes and therefore adopt our technologies and services properly to what's needed in the market to provide value.

Related:Fujitsu, Barcelona Center to Collaborate on Quantum Simulation, Personalized Medicine

First contact with BSC about these new quantum technologies and AI technologies came about in very fruitful meetings last year at the International Supercomputing Center in Hamburg with our CTO, Vivek Mahajan, and director of BSC, Mateo Valero.

At that point, we decided our interests and our work fitted together nicely and we have joint targets. We then decided to work on proposals for quantum and AI. It took quite some time to nail down the concrete research topics we wanted to work on jointly because it took time to really understand in more detail the needs and capabilities of both sides and the results we wanted to achieve.

In April this year we signed a contract in Barcelona and since then work is progressing heavily in both the quantum and AI teams and people are looking forward to achieving compelling results.

In the field of quantum simulation, we wanted to work on new technologies based on tensor networks compared to the state vectors which in some cases allow larger problem sizes and better scalability with more qubits. In AI we want to progress to work on personalized medicine and to do joint work in the field of personalized genomics.

How did you decide on the use approach for this collaboration?

Shintaro Sato: The current use cases are quantum chemistry calculation for material sciences or drug discovery and quantum machine learning on financial problems like predicting the price of derivatives. We need a really large-scale quantum computer, but current quantum computers produce a lot of errors. To develop applications, we need a more reliable machine, which is why we are working on the quantum computer simulator.

We have developed a quantum computer simulator in Japan using feedback loop technology using the same chip as the Fugaku supercomputer, which simulates a 40-qubit quantum computer. But to be able to solve a real-world problem we need a larger-scale quantum computer or quantum simulator.

The reason we decided to work with BSC is that BSC is working on a simulation technique using what is called a tensor network, which can handle a larger number of qubits compared with the standard simulation method.

Even though the special tensor network methods can handle a larger number of qubits, we have to compare the result obtained by the new method with the existing method to see whether that result is really good or not. We believe that it's very important to use secure simulation techniques and that’s one of the things BSC is working on, so that's why we decided to collaborate and develop the application using that simulator.

At the same time, we are developing the 64-qubit quantum computer at Riken [research institute] in Japan. We may use that on the project in the future, but we are starting on the simulation method with BSC.

What is on Fujitsu’s quantum computing roadmap?

Shintaro Sato: Last year we released a 39-qubit Fugaku simulator, and we plan to release a 40-qubit simulator this month.

We have also developed a 64-qubit quantum computer that is going to be released and integrated with the supercomputer at the Riken RQC-Fujitsu Collaboration Center.

Then in two years we will release the largest superconducting quantum computer with 256 qubits and try to implement error correction technology so that we don't have to worry about errors anymore.

Later we are going to release a superconducting quantum computer with more than 1,000 qubits. But while we work on the actual quantum computer, we will also work on the quantum simulator and let users use the quantum simulators so that they have more weapons to develop applications.

At this moment quantum computers are not reliable at all, there are a lot of errors, so to make sure that our quantum computer is correct we need to use a superconductor simulator.