Kathy Gibson at Fujitsu Forum, Munich — Common sense tells us that encryption is the best way to safeguard our data. And it is, because it’s based on the mathematical function of prime factorisation, which is incredibly difficult to solve.
Dr Joseph Reger, chief technology officer: EMEIA at Fujitsu, explains that a public key is the product of two numbers. This means a 248-bit code will take 6-quadrillion years on a PC, or 21-billion years on the fastest supercomputer we have.
“So is this safe for the future?” he asks. It would appear that it will be, since new algorithms or radically-improved hardware will still be unable to break this code.
“But there is a third possibility: that a third type of computer comes up, with exponential performance increase, that will let us break this code,” Dr Reger says.
Quantum computing can do more than break codes, however. Because quantum computing uses quantum physics, it allows us to do all possibilities at the same time — not one by one like we do it now.
Quantum computing is not feasible now, but it will come, Dr Reger says. “The most detailed studies about quantum computing are being conducted by investment companies — they are already trying to figure out what companies to invest in.”
In fact, commercially-viable quantum computing is likely to be with us in anything between five and 10 years, Dr Reger believes.
“There is already a machine that is commercially available, but only two have been sold. However, this demonstrates that the development efforts are real. We probably won’t see quantum computing at scale this decade, but we will in the next.”
Quantum computing is based on quantum physics. And, even if we didn’t want to build a quantum computer, quantum physics are affecting us anyway.
This is because traditional computing is using smaller and smaller transistors that no longer behave as we expect because they are subject to quantum physics happening at a molecular level.
The phenomenon of quantum tunnelling, based on a series of papers by Erwin Schrodinger that describes particles as waves and, because it is not deterministic, the particles can exist in all states simultaneously.
“As we shrink transistors, there is a probability for the electron going through it, tunnelling through it. So you can forget deterministic computing – or sometimes not. But it means that computing is no longer deterministic.”
What this means is, like it or not, quantum effects are coming, Dr Reger says. “So why don’t we use them?”
According to the Schrodinger equation, waves can combine well. This means that you can have all possible solutions to a problem at the same time, which allows for computation using parallelism like never before.
“It means you can compute all possibilities at the same time,” Dr Reger says. “But, of course, in the quantum space this doesn’t tell you anything.
“Once you have performed the calculation in the quantum world, you have to bring it down to the classical world. And the moment you do that, the quantum nature is destroyed.”
Quantum gates would appear to be the answer, but Dr Reger says they are incredibly hard to build, and the algorithms to use them are also hard.
“Quantum computing could be farther out than we think.”
However, once they are built, they will quickly solve seemingly-impossible problems, including RSA encryption.
“So maybe we should start changing the methods of encryption,” Dr Reger says. “Because Fujitsu has a technology that is not quantum, but has learned the tricks and is learning how to mimic a quantum tunnelling approach and put it into the commercial world.”
Dr Reger describes the Digital Annealer as having learned quantum tricks to create a computer that is thousands of times faster than anything available today.
Dr Hirotaka Tamura, Fellow of Fujitsu Laboratories and chief architect of the Digital Annealer, describes it as an accelerator for combinatorial optimisation problems.
Parallel search, modelled on quantum tunnelling, lets it escape from local minimums and radically speed computation.
“Basically, things happen at the same time,” he says.
There are many good reasons to build a quantum-inspired computer rather than a quantum computer itself. “It is easier to build because it is based on traditional technology.”
The Digital Annealer runs at least 10 000 times faster than the fastest computers today.
Tamura shows the example of the travelling salesman problem. In this equation, where the salesman needs to decide on the best order to visit 32 cities, the DA is 17 000 times quicker than anything else on the market.
The DA is optimised for particular use cases in life sciences, financial services, energy, and retail and distribution.
Programming is by 1Qbit Software.
“We can now do molecular design, materials design, drug design, disaster management, radiation therapy, traffic optimisation, financial portfolios, IoT applications, optimisation problems, and artificial intelligence (AI) or machine learning.
“But this is not just about quantum computing,” Dr Reger say. “There is the quantum communication part too — and we will be able to build completely safe communication channels that cannot be copied.
“So we can have safer networks and, so, a safer world, and a quantum-secure Internet.”