"One billion qubit photon computer" developed by Todai and NTT, will be a paradigm shift | New Switch by Nikkan Kogyo Shimbun

After the electronic computer, it has been several decades since it was called an optical computer. Although the replacement of electronic computers has not progressed, it is regaining investment as a photon computer using quantum entanglement. In Japan, the University of Tokyo and NTT will develop a general-purpose photon computer with a billion qubit scale. The Cabinet Office's large-scale support project "Moonshot R & D Project" aims to realize quantum error correction in 2030. (Takayuki Kodera)

"Who will develop NISQ and buy it? Quantum computers have become a word for collecting gold," laments Professor Akira Furusawa of the University of Tokyo. NISQ is a medium-sized quantum computer with no error tolerance. Since the calculation result contains an error, a majority vote is made with multiple qubits to obtain the correct result. Therefore, a practical computer requires 1 million qubits. Currently, about 100 qubits are realized. This is because it is said that it will take several decades to put it to practical use.

Professor Furusawa points out that "it is not possible unless we can prepare essentially a huge number of qubits." It is necessary to change the paradigm from the current gate method that uses superconductivity. Professor Furusawa is working with NTT on a measurement-induced photon computer. A billion-scale qubit is created from photons in a entangled state, and the calculation result is obtained using a calculated reference table like the look-up table method of a computer.

The method raised $ 665 million (about 76.5 billion yen) from Cyclone Tam in the US and US $ 145 million (about 16.5 billion yen) from Xanadu, Canada. Professor Furusawa explains, "Investors have told Photon that the gate method is impossible. Xanadu is copying our method." Japan is chasing a venture that has secured an order of magnitude of funds as a national professional.

NTT has developed a light source for squeezed light. It has become possible to continuously generate photons in a multiple entangled state by squeezing quantum noise by 75%. This was the first time in the world that a photon computer could be assembled. The next goal is to squeeze 90% of the quantum noise. This is the error tolerance threshold. NTT Research Fellow Takeshi Umeki is confident that "90% squeezing can be achieved by extending the current technology."

He also developed a look-up table method of calculation. An algorithm for quantum computing is also under development, and this will lower the threshold of quantum noise and increase the number of qubits that can be used for calculation. Professor Furusawa asserts, "What is really needed is a general-purpose computer with a fast clock frequency and parallel processing. It is not a computer that cannot correct errors."

The period of the moon shot business is up to 25 years. Once the foundation is laid down in this business, it will need capital to win the investment competition.

Source: Nikkan Kogyo Shimbun December 30, 2021

Development of light source for photon computer Noise suppression by polishing

Source: Nikkan Kogyo Shimbun December 23, 2021

Researcher Takeshi Umeki and Takahiro Kashiwazaki of NTT Advanced Integrated Devices Laboratories, and Professor Akira Furusawa of the University of Tokyo have developed a light source for photon computers. The quantum noise of the light source was squeezed by 75% to realize the multiple quantum entangled state of many photons. This will lead to the construction of error-tolerant parallel quantum computers on the scale of one billion qubits.

A special light called squeezed light is generated using lithium niobate as a waveguide. Squeezed light contains an even number of photons in a special state. When the photons of squeezed light are split in half by the beam splitter, the photons in the entangled state flow through the two optical fibers. Quantum calculation is performed using this entangled state.

This time, the processing of lithium niobate was switched from photolithography to polishing. The surface roughness of the waveguide was improved and quantum noise was suppressed. Then, it was possible to realize multiplexing in which the continuously generated photons are in a quantum entangled state with the previous photon in time.

Nikkan Kogyo Shimbun January 4, 2022

COMMENT

I wonder if a computer that surpasses an electronic calculator will really be realized because it can be either quantum or optical. In this photon computer, I was told why the photons traveling by different optical fibers are in the entanglement, "Einstein couldn't understand, so believe in the observations." No matter how much the impact may change the world, I can't help myself, so I feel uncomfortable. When something goes wrong, I wonder if we can guarantee the quality of something that we do not understand the principle. I'm crazy about VCs that can invest 76.5 billion yen. And I think that computers will not spread unless we train a large number of Hijikata programmers who can engineer like electronic computers without leaving it to geniuses. Which quantum computer will you bet your life on? It's a difficult problem.