Just as candle and electric light bulbs are two different lighting technologies, classical and quantum computing are two different computing technologies.
In the evolution of lighting technology, there was a period when electricity was not accessible, reliable and affordable. It was the time when mankind was making rapid progress, though people had graduated from candle light, they couldn’t really leverage the potential of electric light bulbs. That was when “Graetzin-Licht” provided the inspiration for a completely new type of light called Petromax lantern. Depending on the actual size and configuration of the individual unit, effective light output levels from 200 to 1000 candlepower were possible.
This was a leap forward in lighting technology at a time when electric bulb had just been invented (1879) and acted as a bridge before electricity and electric powered lighting became mainstream. Petromax light technology reigned supreme with 90% plus market share for close to 50 years. Today, even after 100 years since its invention, it continues to be an effective source of light for ~500 Million people in countries that are yet to achieve 100% electrification and in developed counties, supporting people in their lifestyle activities like camping and bushwalking.
Today’s candles, i.e. classical computers have the power to put a man on moon. But these computers which are based on classical physics demand that a bit (smallest unit of information) should either be a 1 or a 0 which limits our ability to solve large scale problems involving complex combinatorial optimisations in a reasonable time. Combinatorial optimisation is about finding the best solution from a set of many possible (feasible) solutions for a given context.
Examples of business problems that involve combinatorial optimisation include, drug discovery, route planning, financial portfolio management etc. To illustrate further, in a conventional drug discovery process researchers have to find target compounds that have similar molecular structures to that of the base compound. The way classical computers support this process today is by comparing partial molecular structures of compounds. The search for target compounds takes weeks/months to complete and efficacy is also compromised because computation comes to a halt before actual conclusion.
However, in quantum physics and hence quantum computing, a quantum bit (Q-bit) can be both 0 and 1 and this superposition enables a quantum computer to effectively search for all possible optimal solutions simultaneously before offering the best optimal solution. This makes the quantum computer much faster and effective than a classical computer!
The issue however is that, the industry is yet to build commercial grade Quantum computers.
Fujitsu’s Digital Annealer is a computational architecture offering near Quantum computing capability that can solve large scale business problems involving combinatorial optimisation. Fujitsu has developed a quantum computing inspired (superposition, quantum tunnelling and entanglement) circuitry in a chip called Digital Annealing Unit. This in conjunction with 1-Qbit’s (our partner) quantum computing software achieves accessible, reliable and affordable near quantum capability that is required to solve large scale business problems involving combinatorial optimisation.
In the case of a drug discovery problem we discussed before, Fujitsu’s Digital Annealer can now find target compounds much quicker by comparing the whole of molecular structures and searching all possible optimal compounds. Thus Fujitsu Digital Annealer can not only accelerate drug discovery process, it can also improve the efficacy through high accuracy.
Fujitsu’s Digital Annealer has so far been deployed and trialled with customers in…
Financial services, where a large Bank uses Digital Annealer to help identify new and profitable investment opportunities while achieving full regulatory risk compliance. Fujitsu, sees Digital Annealer becoming the tool of choice for companies in work through their financial portfolio allocation challenges.
Automotive and discrete manufacturing, where companies look to Digital Annealer to optimise their job shop scheduling and smart mobility services (dynamic route planning). In our own manufacturing plant we have seen a 45% reduction in pick distance travelled because of optimised routing and stock placement powered by Digital Annealer
Digital Annealer can be put to work now. It is commercially available as of April 2019.
Our customers typically engage Fujitsu in a 1 day Co-creation workshop where we help identify the candidate business problems, associated constraints and intended business outcomes. The workshop is followed by a proof of concept stage, where for a duration of 3 months, Fujitsu and customer teams (mathematicians, consultants and personnel from customer’s business) work together to translate the business challenge into a mathematical model (Ising Model) and a 1-Qbit based software program that can run on Digital Annealer cloud. The results are evaluated and optimised till the intended business outcomes are achieved. This proof of concept is then easily scaled / productionised on a Cloud hosted Digital Annealer instance or an on-premise infrastructure as desired by the customers. The whole process takes about 6 months from ideation through to scale out!