Japan’s Quantum Computer Prototype Is 100 Times Faster Than Supercomputers

In November 2017, Japan unveiled it’s first quantum computer prototype that is opened up for free to the public over Internet for trials. With this machine, Japan has joined the race of building world’s most powerful computer with larger brute force, which is the key towards utilizing the full potential of artificial intelligence.

The project is developed by Nippon Telegraph and Telephone Corporation, University of Tokyo, National Institute of Informatics, Stanford University, and financially supported by the Government of Japan’s ImPACT program.

The machine is based on quantum neural network that can theoretically solve complex problems around 100 times faster than traditional supercomputers. What’s more impressive is, it does all of this while consuming only 1 kilowatt of power, rather than 10,000 kilowatts which is used by conventional supercomputers to perform the same task. Let’s find out what exactly they have developed and how does it work.

Quantum Neural Networks

Quantum Neural Networks (QNNs) use optical parametric oscillators as quantum neurons and optical homodyne measurement feedback circuits as quantum synapses. It searches for a solution of several combinations of optimization problems by exploiting collective symmetry braking at the optical parametric oscillators threshold.

Furthermore, users can experience what it’s actually like to carry out experiments with the QNN and simulations based on the quantum theory of optical parametric oscillator networks.

In simple terms, in quantum neural network, researchers try to integrate artificial neural network models utilizing the benefits of quantum information to build more efficient applications. The objective is to use quantum computing features (quantum parallelism, interference, entanglement) as resources. However, it’s quite difficult to train classical neural networks, especially in big data apps. 


If you are interested in the principles and features of quantum neural network, the QNNCloud offers 3 tools –

  1. A white paper elaborating quantum theory
  2. Quantum simulation capability using Shoubu supercomputer
  3. Quantum computation using QNN

The QNNcloud is built on a network of 2000 optical parametric oscillators with programmable all-to-all connections, which enable users to solve NP Hard Max Cut problems of size up to N=2,000 on complete graphs (which is far beyond the limitations of current quantum computers), without putting any effort of embedding a target graph in the machine’s hardware.

Millions of problems exist that involve continuous and combinatorial optimization, such as lead compound optimization in developing frequency band, transmission power in wireless communications, medicine, portfolio optimization in Fin tech, Boltzmann sampling in machine learning, sparse coding for compressed sensing, etc.

Majority of these problems fall under non-deterministic polynomial (NP), NP Complete and NP Hard classes in complexity theory. It requires a huge amount of computational resources to solve them, as the size of the problem increases at every iteration.

The QNN system utilizes quantum parallel searching at below optical parametric oscillators threshold, collective symmetry breaking at threshold, and exponential probability amplification at above threshold, to deal with these limitations.

In near future, the QNNcloud will offer a simulation tool for developing quantum algorithms for real world applications.

Hardware of QNN

The QNN hardware is not as complicated as it sounds. In a one kilometer long fiber ring cavity, N=2,000 pulses of optical parametric oscillator are created concurrently by exciting intra-cavity and periodically poled LiNb03 waveguide instrument using a pulse train with 1 GHz frequency.

Source: QNNcloud

A binary variable is represented as π-phase and 0-phase states of each optical parametric oscillator pulse. All pulses are generated in a π-phase and 0-phase superposition at below threshold, but with either one of the two at above threshold. Any pair of these pulses could be coupled by measuring their amplitude sequentially.

Here, measuring refers to evaluating a suitable feedback pulse amplitude with FPGA (short for field programmable gate array). The feedback is then injected into the target optical parametric oscillator pulse.

All-to-all connections for N=2,000 pluses are executed at each round trip (which lasts for 5 microseconds). When external pump rate rises to above threshold, the solution is obtained as an π-phase or 0-phase configuration after 10 to 1,000 round trips.

QNN Simulator

The QNN dynamics can be predicted theoretically with the help of quantum master equation, considering the wave packet reduction induced by measurements. The model is executed on Shoubu supercomputer, and huge parallel simulation allows it to regenerate the QNN dynamics in significantly less time.

Budget and Future Plans

Currently, United States spends more than $200 million per year in the research and development of quantum computing technology, while China is reportedly building a $10 billion research center for quantum applications.

Japan, on the other hand, has planned to devote nearly $267 million for quantum computing over a decade starting in April 2018. Also, Hitachi is researching quantum computing techniques in partnership with the University of Cambridge.

The algorithms for different real world applications, simulation tools for developing new algorithms and an advanced QNN with recurrent neural network architecture will be released in the future. Currently, they’re targeting commercialization by first quarter of 2020. They’ll be focusing on deeper optimization problems with mobile optimization, urban traffic congestion and discovery of new medicines and chemicals.

Read: 10+ Most Interesting Facts About Quantum Computers

Meanwhile, tech giants like Microsoft, IBM and Google are working on their own quantum machines, and their tests show breakthrough is within reach.

Written by
Varun Kumar

I am a professional technology and business research analyst with more than a decade of experience in the field. My main areas of expertise include software technologies, business strategies, competitive analysis, and staying up-to-date with market trends.

I hold a Master's degree in computer science from GGSIPU University. If you'd like to learn more about my latest projects and insights, please don't hesitate to reach out to me via email at [email protected].

View all articles
Leave a reply