Fujitsu has a Digital Annealer with 8192 Qubits and a 1 million qubit system in the lab. This is a classical hardware emulation of quantum annealing (analog) system. This is not a quantum computer but hardware emulation.
David Worral, Head of Quantum Technologies at Cambridge Quantum Computing, indicated that Fujitsu has a 1 million qubit quantum annealing system in a lab at the University of Toronto.
Fujitsu has been a leader in creating lithography machines and supercomputers and FPGAs. Fujitsu chose to leverage this technology and their massive capabilities to develop quantum annealing far beyond the capabilities and scale of D-Wave Systems.
Digital Annealer is a new technology that is used to solve large-scale combinatorial optimization problems instantly. Digital Annealer uses a digital circuit design inspired by quantum phenomena and can solve problems which are difficult and time consuming for classical computers.
Fujitsu’s second-generation quantum-inspired Digital Annealer (8192 qubit) opens doors to previously impossible levels of business and societal problem-solving.
Here is a 51 document describing their digital annealing.
Applications in Decryption, Optimization and More
If the 1 million qubit quantum annealing system works then many levels (ECC) of financial and corporate encryption could be cracked within a year.
Other possibilities for digital annealing in financial services include calculating the optimum amount of cash and the most efficient route for ATM replenishment. Cash replenishment
accounts for up to 60 percent of ATM network operating costs and optimization would improve profitability significantly at a time when ATM network operations are under pressure.
And using Fujitsu’s Quantum-Inspired Digital Annealer, NatWest bank has completed a highly complex portfolio risk optimization calculation that needs to be undertaken regularly by the bank, at 300 times the speed of a traditional computer while providing an even higher degree of accuracy.
Fujitsu Laboratories has shown the capability of the Digital Annealer, Fujitsu’s computational architecture inspired by quantum phenomena that rapidly solves combinatorial optimization problems, to maximize the performance of magnetic devices essential for renewable energy harvesting and other uses. The application of Fujitsu’s next-generation architecture allows for the nearly instantaneous calculation of the optimal arrangement of multiple planar (2D) magnets to maximize the strength of the magnetic field in a device.
Many magnetic devices used for environmental power generation create magnetic flux through the arrangement of a large number of small magnets. The optimal planar (2D) arrangement for maximizing power generation efficiency remains difficult to calculate due to the enormous number of potential combinations of magnet arrangements, however. To overcome this challenge, Fujitsu has developed a technology that utilizes its Digital Annealer to calculate in a matter of seconds how to arrange each individual magnet to achieve maximum magnetic flux density, delivering an efficiency gain of 16%.
This technological breakthrough now makes it possible to quickly calculate the optimal design for magnetic devices with significantly higher power generation efficiency, and will ultimately contribute to the spread of power generation devices that utilize renewable energy such as energy harvesting devices.
Fujitsu claims its Digital Annealer, with 64-bit graduations, is far more accurate than D-Wave’s offering.
Features and Benefits
* Stable operation at normal room temperature and small form factor
* Fully coupled 8,192-bit connectivity that allows for large-scale problem solving
* 64-bit (264) gradations allow high accuracy in expressing combinatorial optimization problems
Fujitsu’s Digital Annealer can solve combinatorial optimization problems instantly using a digital circuit design inspired by quantum phenomena.
The advancement of ICT technology and the realization of Artificial Intelligence (AI) today means that there is a necessity for computers to be able to carry out complex calculations in an instant, with the expectation that the knowledge gained will then be applied to various fields of business such as manufacturing, distribution, retail, automotive and finance, to name a few.
This case study introduces how Fujitsu IT Products Limited deployed Digital Annealer and succeeded in dramatically improving the efficiency of in-warehouse operations.
What is combinatorial optimization?
There is a widespread real-world demand for the ability to choose the optimal solution from a finite set of possibilities, where the scale is typically measured in quintillions. These challenges are classified as combinatorial optimization problems – essentially finding the best combination from an enormous set of potential elements. Example use cases include portfolio optimization and credit risk assessment in financial services; job shop scheduling, car design optimization, robot positioning optimization and many more in manufacturing; drug and materials discovery in life sciences and asset allocation for utilities networks. These problems are difficult to solve optimally in real-time with existing processors, even with the fastest computers, as the number of combinations increases exponentially as the number of factors taken into consideration is increased to obtain precision.
Traffic route optimization, for example, is a particularly difficult arena. Optimizing five pairs of start and destination points has to consider 10^100 possible routes, avoiding overlaps between vehicles and avoiding traffic jams. This use case has been investigated by several global automotive vendors for their autonomous cars and mobility platforms, and by governments as a means to reduce transport’s carbon footprint and for the betterment of the society.
Conventional computing has challenges solving combinatorial optimization challenges optimally in a practical amount of time and relies on approximations. Quantum computing computes all possible solutions simultaneously. When it is eventually ready to move out of the laboratory and solve practical real-world problems, quantum computing may be able to solve such challenges. But it’s not yet usable in the real world. On the other hand, quantum-inspired computing with the Fujitsu Digital Annealer is available today and delivers optimization calculations with the speed, precision and at a scale that true quantum computing is not able to achieve. For example, the Digital Annealer solution can solve the five pair traffic optimization challenge dealing with 10^100 possibilities in one second.
Annealing is a probabilistic technique for approximating the overall optimum result of a given function. Until now, in tackling any combinatorial optimization process with annealing, there has been a trade-off between precision and risk. Seeking high precision used to imply the need for more time to calculate the answer – often more time than was available – while accepting a ‘good enough’ answer introduced an increasing amount of risk and the need for a security buffer. The more precise the calculation you can achieve, the more cost-efficient the final process will be, leading to less waste.
Quantum annealing solves the speed side of this equation, but it is unlikely to be available for solving real-world scenarios or ready for practical enterprise use for at least 10 to 15 years, if at all [They are referring to and dissing real quantum computers and real quantum annealers.]. Fujitsu’s scientists were keen on finding how to solve these critical problems today and realized that the software being developed for quantum computers could be applied to digital architectures.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
12 thoughts on “Fujitsu Has a 1 Million Qubit Digital Emulated Quantum Annealing System in the Lab and 8192 Qubits Commercially Available”
How, when, where and how well? Allmost everything has been done, but usually so poorly that it was useless..
You say that as though powerpoint presentations weren’t the basis for most of the breakthroughs that we see in advanced military tech, space flight, cures for aging and even fusion power.
Clearly a source of pretty .ppt is all you need to have a successful tech project.
Correction: Fujitsu does not make FPGAs. They use Intel’s, which really is rebranded Altera.
Though it does not seem to be truly a quantum computer it’s still great news, a very useful technology for local minima searching algorithms.
One thing. What prevents the system from relaxing to a local minimum rather than an optimum one? Is it dependent on the system being simulated on the annealing system, or is the risk always great/zero?
At least in neural networks, it is known that the initial random values of the weights affect the trained network final efficiency, i.e. certain networks start “closer” to the good/optimal (or has a viable training path to) solutions, whereas other random combinations of initial weights start farther from (or doesn’t have a viable training path to) a good/optimal solution. This can be interpreted as there being local minima in the ANN that cannot be overcome with normal back propagation.
So, are there similar local minima in the simulations for this system? When optimizing the best traffic route (see above), will the annealer get “stuck” in a so-so-solution rather than finding the optimum one?
I have one possible killer application for this system: calculating the correct weight for a ANN. If a part of the neural network can be formulated as an annealing network, then one would be able to calculate the near optimum weights by performing one annealing cycle rather than thousands or millions of back-propagation cycles.
Wow, this looks amazing. Thanks, Brian.
Here is that claim.
According to Dwave public press released, they were on track to achieve the computing power of the entire universe by 2015.
So they have ascended to godhood and left this mortal plane long ago. That’s why we don’t hear about them any more.
Just because you left the gate first does not mean you’ll be the first at the finish line.
Put enough into r&d and you can make up any deficit.
Wait….wtf happned to Dwave? I thought they were on the cutting edge of quantum annealing?
Comments are closed.