A bridge to Quantum Computing - and why we need one

Delivering real results with quantum-inspired computing

Main visual : A bridge to Quantum Computing - and why we need one

I attended hub.berlin last week, Europe's interactive business festival for digital movers and makers. I had the opportunity to talk about quantum computing - a technology that has enormous potential but that is also currently subject to a great deal of hype.

I set out to cut through that hype and share how we have built a technical ‘bridge’ that allows us to immediately access many of the accelerated compute benefits of quantum technology, today.

Quantum Computing 101

Before I talk about this bridge, I want to explain what it is that we are bridging to: quantum computing itself – a technology that promises the next generation of computing, possibly one million, 10 million or even 100 million times faster than what we have today.

However, these numbers themselves are meaningless. The real question is: “Does mankind have problems today that need this level of compute power?”

The answer is that not only can we use this power, but the problems we can apply it to are almost countless and they exist in all industries.

For example, quantum computing has the potential to transform materials design – enabling us to design molecules or compounds according to precise specifications and requirements.

The larger the molecules become, the more complex the problem that needs solving – and traditional computers soon run out of the horsepower needed. Solving this sort of challenge will revolutionize medical drug design and the treatment of diseases, as well, not just materials design.

Another industry that can benefit immediately from next generation computer technology is manufacturing – a sector that is particularly dear to those of us who live in Germany.

Here, we are not talking about solving theoretical mathematical challenges like the classic travelling salesman problem or the knapsack problem, rather about optimizing manufacturing processes in real-time or near-time.

That’s because multiple speedups and efficiencies gained by improving single steps or individual processes add up to deliver significant benefits in terms of production time and cost.

Now that we’ve covered what our goal looks like, I need to talk about the technology it is based on. Bear with me, as I am going to condense a several-semester course on quantum physics into a few paragraphs. Obviously, I will have to simplify but I hope not to over-simplify.

The first crucial concept to understand is superposition – which is what makes quantum computing so powerful. This is a bizarre property of quantum systems to be in multiple states at the same time.

Whereas in classical (i.e. today’s) computing we have the simple unit called a bit that can have the value of one or zero, in quantum computing, quantum bits can also have the value of one or zero, or any mixture of those, but at the same time.

1 and 0, not 1 or 0. That means that a rather small 64-qbit system can be in 2 to the power of 64 states simultaneously, which is a very large number (≈ 1019) This means that even a relatively slow quantum machine can beat a much more powerful traditional machine by processing 1019 options at once.

If we can keep our computer in a quantum state, we can do enormous amounts of computation at one go.

Another key concept, no less strange, is the phenomenon of quantum tunneling. Translating this into a real-world concept would equate to the very real possibility of an object that you throw at a wall, flying through that wall to the other side.

Strictly speaking, in the traditional (a.k.a. classical) world we live in, this possibility also exists. It’s just so infinitesimally small that it is pointless to attempt to throw things through walls. But in the quantum world, quantum objects can tunnel through barriers with a considerable probability.

The third concept – which is unsurprisingly equally peculiar – is quantum entanglement. This is when quantum particles, once entangled, remain connected so that actions performed on one affect the other, even when they are separated by great distances. In fact, distance doesn’t play any role.

Being able to harness these quantum characteristics and phenomena for computation is the Holy Grail. That’s what we strive to achieve. But despite the hype by some companies, we’re not there yet with quantum computing.

Today’s quantum computers can solve toy problems only. Interesting, but far from practical. And they currently only exist in laboratories. This is because they (the most advanced ones anyway) only run close to absolute zero temperature – that is minus 273.15 degrees Celsius - a temperature colder than outer space.

As you can imagine, it takes a significant amount of energy to maintain that temperature.

What’s more, these early quantum systems also need electromagnetic shielding, as the slightest disturbance will cause a quantum system to decompose – this essentially means that it gets bumped out of its quantum state, causing it to lose both its superposition capabilities and the benefits that come with them.

Universal quantum computers are hard to build, hard to run and hard to program. But there’s a way of applying quantum capabilities in a simpler, perhaps less capable quantum computer called a quantum annealer – which has advantages but unfortunately, also needs similar laboratory conditions.

These systems essentially solve complex combinatorial optimization problems by exploring a huge number of possibilities to find the best possible value, that is, the optimal solution.

Fujitsu’s Digital Annealer Takes Inspiration from Quantum Technology

As we’ve seen, quantum capabilities aren’t yet available to us in their full strength. But Fujitsu has used guidance from this approach to create a new bridge technology.

We’ve taken inspiration from superposition, and have mimicked tunneling and entanglement digitally and in hardware to create the Digital Annealer.

So, now you might be thinking, we have a solution that isn’t even as good as a quantum annealer – never mind the full quantum computing experience with universal quantum computers. What’s the point of doing that?

Well, it is hugely beneficial, as every industry from automotive, finance, healthcare, retail to the public sector faces countless complex combinatorial optimization problems.

For example, coming back to the field of molecular design – starting with just 50 atoms – we must explore ≈ 1048 (10 to the power of 48) combinations. In finance, if we want to optimize a portfolio of high value liquid assets in a volatile market, we must explore the optimal combination of 500 stocks by considering 10150 (10 to the power of 150 combinations). And there only some 1077 atoms in the universe, for comparison.

Back to manufacturing, my favorite, we can calculate the most efficient sequence of welding points and seams on a car chassis for a robotic arm, also specifying whether seams should be welded right to left, or left to right, to reduce the amount of idle travel the robot arm needs to make.

This is essentially the manufacturing version of the travelling salesperson (TSP) route optimization problem. Over a whole day, month or year, these small savings of time add up to a substantial competitive advantage in cost.

In the depicted case, optimizing a 64-seam welding job enables the automotive manufacturer to produce more vehicles in the same timeframe using the same resources (cost down).


Once we look at the potential applications of the Digital Annealer in this way, we start to realize that the technology isn’t really limited by the restriction to combinatorial optimization problems.

Its dramatic speedup of up to 4 orders of magnitude (104) is worth talking about, whether real quantum or not. The key is to be able to translate the problem into a particular mathematic form known as QUBO, or “quadratic unconstrained binary optimization”. The Digital Annealer can take it from there.

In summary, our technology isn’t quantum as it isn’t in a quantum state. And it only does one thing – and that’s solving combinatorial optimization problems. But it does that really well.

We can’t achieve the possible 100,000x boost of quantum computing, but we can solve these combinatorial problems up to ten thousand times faster to a very good accuracy than anything else on the market today – outperforming even the lab-bound proto quantum computers of today.

This means Fujitsu’s “quantum inspired” bridge technology is currently more powerful than the technology it is bridging to.

We could say, tongue in cheek, that it has taken inspiration from quantum tunneling to pass right through the quantum computing hype to emerge on the plateau of productivity for industrial applications today!

So, the question is – do we wait until quantum computers are ready?

Or leverage some of the quantifiable benefits you can already see with quantum algorithms mimicked on a quantum-inspired solution, the Digital Annealer?

Find out more here.