Kathy Gibson is at Fujitsu Forum in Munich – The era of quantum and quantum-inspired computing will open up new opportunities that we probably don’t fully recognise today.
New levels of optimisation are already having a massive impact on a number of workloads and industries, says Joseph Reger, Fujitsu Fellow and chief technology officer at Fujitsu.
He reminds delegates that optimisation is the act of making something as good as possible according to circumstance, time and the means available.
“We human beings do that all the time. We optimise without even thinking about it much.”
We have been optimising since the beginning of time, and today have optimised very complex physical and virtual systems, Reger says.
“So where is the problem?” he asks. “We are hitting limits to what we can do. It turns out that the mathematical nature of the problems today become exponential in nature. And the problems become exponentially difficult and we are hitting the limits.
“But we have to carry on optimising.”
Using science and mathematics we can determine where to optimise, he explains. But the existing models fall short when it comes to the very complex problems we face today.
When models fail and brute force is impractical, Reger advocates the application of what he calls “smart force”.
Probabilistic methods play an important role in industrial optimisation, he says, but using physics, and annealing, is a more efficient way of achieving results.
Hiding complexity from the system lets it jump over states to achieve simulated annealing.
But cutting through barriers – or quantum tunneling – is a much more efficient way to achieve this, Reger says. And quantum annealing is proving itself in a number of industrial workloads.
“I know quantum annealing is not full quantum computing, and we will need full quantum computing some day. But industrial optimsisation problems are very well-suited to quantum annealing – so we shouldn’t hesitate to use this technology today, because it is available.”
Fujitsu’s Digital Annealer isn’t a quantum computer, but is rather a specialised computer that does things like quantum tunneling, like super-position, like quantum mechanics.
“It works – and does it very well,” Reger adds. “It is very efficient and the software is almost identical to what we use for quantum annealing.
“And this system is without the barriers to quantum – so it is a bridge to quantum computing, or quantum inspired.”
Anyone can participate in the optimisation race, Reger adds – and the solution may not necessarily be quantum computing.
“This Digital Annealer makes it even more democratic because it is available now. Anyone can use it immediately and apply it to real industrial problems – now.”
Some use cases already in the field include work that Deutsche Telecom is doing.
Marc Geitz, innovation architect at Deutsche Telecom Innovation Laboratories, says the organisation has started applying the Digital Annealer to network configuration – possibly one of the most difficult problems the telco faces.
“We want to optimise networks in a way to provide the best service for our customers.” The cost of rolling out networks costs billions of euros, and DT wants to make the best use of that investment by optimising network performance.
The initial phases of the project have been to find optimal routs that use the best energy function
A future use case for the Digital Annealer could be in the field of artificial intelligence (AI) for customer satisfaction, Geitz says.
Traffic is a constant problem around the world, and the Hamburg Port Authority is using the Digital Annealer to optimise traffic flows.
The biggest ships in the world use the port, totaling 14-million tonnes, 8,82-million containers, 72 600 trains and 2,92-million trucks per year.
Managing all of this traffic is an enormous challenge, measuring stress levels and aiming to keep these at the lowest possible levels.
The Digital Annealer is used to run and test the necessary models and has resulted in optimisation of the streets modelled so far.
Back in June 2017, BMW identified quantum computing as a direction to consider. During its proof of concept on the technology, the company found that a quantum annealer it tested could not do the required job – at which point the company turned to Fujitsu’s Digital Annealer.
Oliver Wick, quantum computing lead at BMW, points out that many use cases are suited to digital annealing technologies. These include projects in development and production.
A proof of concept that has been completed is the BMW paint shop, where the Digital Annealer was used to increase the optimisation of PVC painting.
The robot programs have now been optimised, with theoretical time savings proved. The system is now being validated at the Munich paint plant.
“We need to optimse more,” Reger adds. “We are not optimal or even close to it – therefore we must do more.
“Our job is to blend natural and artificial resources so we can optimise what’s limited by using the unlimited.”