From the 1940s until the present, there have been four major iterations in the way computing is enabled, starting with the humble vacuum tube and progressing toward the microprocessors that power just about every computer in the world today. Since these early days, the world's appetite for computing has only grown exponentially, and shows few signs of stopping or slowing down.
Quantum Computing is currently being hailed as a likely candidate for the next generation of technologies that will enable computing in the future, including for applications in emergent fields such as Artificial Intelligence (AI), which has increased in prominence in recent times. Many predict that the impact of AI promises to trigger rapid technological advancements and cause far reaching cultural shifts ultimately resulting in "technological singularity--"an event which every business may need to prepare for irrespective of industry or sector they belong to.
In these dynamic times, businesses and societies must confront this tide of technological change-- computers that can perform more complex calculations, faster than ever before will remain key to staying competitive and relevant in the market. Quantum computing is a promising technology that could deliver unprecedented computational power, offering the capability to serve the demanding market needs of the 21st Century.
Quantum Computing boasts obvious advantages over classical computing, including the ability to solve extremely complex problems in a fraction of the time it would take a conventional computer. The technology is not without its limitations, however. Presently, Quantum Computing -remains very expensive and extremely difficult to run as a practical technology, and requires very "special conditions" in order to compute and provide output. In other words, the associated costs drive up CAPEX and OPEX, not to mention the added technical demands and expertise needed to maintain the complex infrastructure that is needed to run a Quantum Computer.
For example, in order to get the correct output for a problem, quantum bits need to stay in superposition at near zero temperatures free of any outside interference, including cosmic or magnetic rays. If either of these conditions are not met, the "qubits" freeze out of their delicate state into something known as classical state which yields incorrect results. The challenges associated with maintaining these precise conditions ensure that for the time being, at least, Quantum Computing remains complex and impractical to many sectors and industries.
Digital Annealer by Fujitsu was developed as an alternative to Quantum Computing technology, taking into account these technical challenges, while also drawing inspiration from theoretical principles of quantum phenomena. This article introduces the Digital Annealer, and shows how its use can transform businesses and society.
The Limitations of Classical Computers Encountered in Practical Use of AI
Leveraging AI fully requires an enormous amount of computing resources, effort, and time. And the need for faster computers to carry out operations quickly to meet this demand requires greater scaling of integrated circuits or integrated circuits with three-dimensional structures.
While semiconductor scaling has progressed in accordance with Moore's Law -- which predicted that the density of transistors in an integrated circuit would double every 18 months -- this scaling has nearly reached its limit.
Beyond the Limits of Moore's Law
Quantum computers, which are based on completely different principles than classical computers, are attracting attention as a key element to liberate us from the limits of Moore's Law and to build platforms for leveraging AI to its full potential.
Quantum computers are based on the principles of quantum theory, which explains phenomena that occur at the atomic and subatomic levels, such as the behavior of electrons. In quantum mechanics, the phenomenon known as "superposition" refers to the simultaneous appearance of two different states, "0" and "1." This phenomenon forms the basis of "quantum bits," which can dramatically improve operation speeds.
Despite their enormous potential, quantum computers remain in the nascent stages of development, and many countries and regions invest heavily in research and development to further realize this radical approach to computing. While the widespread use of this technology is still years away, analyzing big data and leveraging AI to solve today's most challenging problems represents an essential objective for forward-thinking businesses.
Keeping these challenges in mind, Fujitsu has developed a futuristic computing technology with a new architecture that embraces the advantages of both quantum and classical computers.
Digital Annealer Finds Optimal Solutions Instantly
Implementing Future Quantum Computing Technology Today――Digital Annealer
Among the various quantum computing methods that exist in the market today, Digital Annealer is categorized as an example of the annealing method, which focuses on the solution of combinatorial optimization problems and the achievement of successful results through its rapid operational capabilities. Unlike classical computers, the Digital Annealing method does not require programming; instead, simply setting the parameters causes calculations to be performed.
With the superconducting annealing method, which employs pure quantum computing devices, it is difficult to maintain quantum states and these machines are inevitably large as a consequence. In addition, quantum bits can only be connected to adjacent ones because of the need for physical connections between the bits.
To solve these challenges, Fujitsu used conventional semiconductor technology and developed Digital Annealer, a new computing architecture that can quickly solve combinatorial optimization problems without the added costs and complications that are typically associated with Quantum Computing.
Available Today without Waiting for Quantum Computers
Digital Annealer uses a Quantum-inspired digital circuit that enables it to rapidly solve complex problems.
Digital Annealer can compute more complex problems than other quantum annealing machines since its fully connected structure enables devices to freely exchange signals inside the computer. Moreover, it can operate at normal temperatures without the need for complex cooling solutions to reach absolute zero (-273.15℃) which result in complexity and further drive up costs.
Digital Annealer is also small enough to fit in a datacenter rack unlike pure quantum computers that require considerable underlying infrastructure to function. Fujitsu also provides Digital Annealer as a cloud service that anyone can use readily starting tomorrow.
Digital Annealer is the ideal computing technology that can be leveraged today instead of waiting for the development of practical quantum computers.
Collaboration with 1Qbit and University of Toronto
To ensure further progress in this area, Fujitsu has concluded strategic partnership agreements with 1QBit (a Canada-based company) and the University of Toronto to achieve future development of hardware and software in this space.
1Qbit Information Technologies Inc. is a quantum computing software company that will collaborate with Fujitsu on applying quantum-inspired technology to the field of AI, focusing on the areas of combinatorial optimization and machine learning.
The University of Toronto is among the world's top ranked research universities with internationally recognized strengths in quantum-inspired computing and artificial intelligence, such as the Centre for Quantum Information and Quantum Computing (CQIQC). Fujitsu Laboratories has established a new research center in Toronto focusing on R&D into breakthrough quantum computing technologies.
Innovating in Every Line of Business with Combinatorial Optimization
Combinatorial optimization problems cannot be solved using classical computing on a practical level since the amount of computation increases exponentially when the number of elements in an optimal combination increase by even a small fraction.
One of the most well-known examples of this type of problem is the traveling salesman problem which states - "When a salesman must visit several cities and then return to the origin city, which visit order minimizes the salesman's total traveling distance?"
When the number of cities is just two or three, the solution can be found in a moment. However, when the number is increased to say 30 cities, it takes a classical computer approximately 800 million years to perform round-robin calculations in order to find the optimal solution.
Such severe limitations are easily solved by Digital Annealer, which can quickly and efficiently solve this problem type, making it an ideal solution of combinatorial optimization problems that businesses are facing today.
Application in Various Business Fields, Including Manufacturing, Distribution, and Finance
Optimized Placement of In-Stock Parts in Factories and Distribution Warehouses
In real business scenarios, many problems can be solved with combinatorial optimization. For example, in factories and distribution warehouses, the aforementioned traveling salesman problem can be applied to optimize the work routes and placement of in-stock parts, thereby minimizing the time it takes workers to walk around picking parts. In fact, introduction of Digital Annealer has successfully reduced workers' traveling distances by 45% per month at a Fujitsu group factory.
Personalized Digital Marketing Advertisements
In the field of digital marketing, advertisements and content on web pages can be altered according to viewers' attributes (e.g., age and gender) when known in advance. Providing attribute data to each webpage part allows more appropriate and precisely personalized information to be shown to viewers. This enables specific advertisements to be shown to 30-year-old single women or particular content to be displayed to 53-year-old married men.
With six attributes, there are only 720 permutations that can be served to the viewer. However, with 20 attributes, there are as many as 2.43 quintillion (quintillion: 1,000 times a quadrillion) permutations or possible recommendations that can be served to the viewer. Using Digital Annealer, this complex calculation can be resolved instantly.
Such usage can be extended to a variety of fields, such as proposing the optimal itineraries for the day's weather and modes of transportation and so on.
Low-Risk, Diversified Financial Investments
Digital Annealer can also be applied to challenges such as reducing risks by optimizing investment portfolios. For example, 20 or more stock names result in an astounding number of permutations-- more than one quintillion-- which no conventional computer can solve in a practical amount of time. By contrast, Digital Annealer can instantly find the least risky, diversified investment allocation that maximizes returns, even from among 500 stock names, by grouping stocks that correlate with price variations.
Faster Molecule Similarity Searching in Drug Development
Chemical laboratories and pharmaceutical companies often use "molecule similarity searching," which extracts molecules' characteristics partially to perform searches, as an effective means for finding new substances and developing new drugs. If two molecules can be broken down to atoms and the match rate can be found, similarity searching can have very high accuracy. Unfortunately, this approach takes a considerable amount of time to perform even for high-performance computers.
However, in case of the Digital Annealer, searching based on entire molecular structures can be performed quickly without limiting which characteristics can be extracted, thereby facilitating accurate, and speedy drug discovery.
Digital Annealer Opens Up the Future
As big data analysis and AI become increasingly important themes in business innovation, some companies have started to invest in AI and data science. However, the difficulty in optimizing business processes is that many problems discourage any attempt to solve them because they require an immense amount of computation. Such problems can now be solved thanks to the development of quantum computing technology.
Digital Annealer makes it possible to easily verify hypotheses that may have been abandoned due to immense computational costs and time, thereby steadily advancing business objectives.
The year 2018 will mark the beginning of businesses' utilization of quantum computing technology. As the first and foremost application, Fujitsu Digital Annealer stands ready to open up the future.
1Qbit's CEO Speaks on His Expectations for Digital Annealer
1QBit is the world's first company to specialize in software for quantum computers. 1QBit has collaborated with many companies that focus on manufacturing quantum hardware; in particular, we think Fujitsu Digital Annealer is the first real piece of hardware that can truly leverage the research we have conducted thus far.
Digital Annealer beneficially allows not only utilization of methods for conventional computers that have been accumulated over many years but also research into advanced technology related to the quantum annealing method. Use of Digital Annealer makes it possible to respond to a variety of challenges that exceed the scale today's quantum computers can handle.
1QBit, a leading developer of software that leverages the new paradigm of quantum computing, highly appreciates our collaboration with Fujitsu, which has technological advantages in hardware. We expect to continue to have a long-term, fruitful relationship going forward.