What's behind quantum computing and why Daimler is researching it

Picture a new, incredibly powerful race car that can accelerate at an unheard-of speed right out of the gate, leaving the competition in the dust. But that velocity and power come at a price. The driver can only keep the machine under control for a few short moments before it veers off course and runs off the track, unable to complete a full lap let alone a race.

This scene sums up nicely where things stand with the latest and most promising type of computing out there: quantum computing. It’s still a nascent technology with its own evolving set of hardware and software that scientists and engineers are trying to understand and master. But they hope that in a few years’ time quantum computing will be able to solve problems that would even take a state-of-the-art supercomputer hundreds or thousands of years to crack. It could become the best way to discover new, more efficient battery technology, simulating aerodynamic shapes for better fuel efficiency and a smoother ride, or to optimize manufacturing processes with myriad variables.

Daimler and the quantum leap

Daimler is hard at work with a global research program headquartered in Silicon Valley to explore how it can take computing to the next level. At Mercedes-Benz Research and Development North America (MBRDNA), a team of experts together with their counterparts in Sindelfingen is coordinating the company’s efforts to get up to speed with this technology, better understand its potential and current shortcomings and conduct basic research in close collaboration with industry leaders from the high-tech world and academia.

The Quantum Computing Initiative was established within Daimler by 2015 when the company started to explore, evaluate and benchmark this novel technology. “We always try to pick up emerging trends and understand the future of technology. And quantum computing is obviously one of those technologies where we still need to learn a lot as a company,” explains Ben Boeser, who heads up Innovation Management at MBRDNA. “It’s a very research-oriented activity, looking at things that happen somewhere 10 to 15 years out, but we want to understand the basics as a new universe is created — and we as a company want to be a part of it.”

To that end, Boeser in early 2018 assembled a team of specialists in Silicon Valley who work with a global network of in-house experts, renowned tech heavyweights IBM and Google and external researchers around the world to develop software or algorithms that could help solve previously unanswerable problems in record time. “It makes a lot of sense to have the nucleus of this global research effort in Silicon Valley,” Boeser says. “The U.S. is leading in this emerging field, and we are in close proximity to Google and IBM, the two companies that have the most advanced technology in the market right now when it comes to quantum computing. Our partnerships for fundamental research collaboration make a big difference because we can discuss approaches and develop algorithms together, not as a customer talks to a vendor but at eye level.”

Learning from competent partners

Both IBM and Google have developed their own distinct quantum computing architecture that Boeser’s team can access through the cloud to run experiments and gain insights into the basics behind this new method of computation. And while the innovation manager cautions that the path from running theoretical experiments to actually useful, business-related queries is long, building up the knowledge and skills early on has the potential to speed up and improve not only battery research but other areas that will propel Daimler as a company and transportation in general forward.

Today, the race car called quantum computer is only stable and stays reliably on course for a few moments. That has to do with the way a quantum computer works. While classical computers store information in an unambiguous pattern of bits containing either a one or a zero (or an on or off switch that lets a current flow), a quantum machine follows the principles of quantum mechanics. It employs so-called qubits as the essential units or building blocks of information, which can store any combination of one and zero at the same time, while a classical bit can store either one or zero. Experts call this superposition, and it is one of two features that gives quantum computers the ability to potentially be much faster than today’s mainframes and servers.

The second principle is entanglement, the interaction of one qubit with another one. While today’s computers are expressly designed to keep their bits separated to avoid errors and perform one calculation for one result, a quantum machine’s power lies in the ability to conduct multiple calculations at once. In technical jargon, a certain number “N” of qubits can describe a superposition of 2N states — an exponentially larger range of operations and results than today’s machines. So a fully scaled version of a quantum computer loaded with enough qubits could — in theory — run highly complex calculations in mere seconds that would take a classical computer millennia.

Unbridled power

But quits have the nasty habit of uncontrollably interacting with each other, which makes them highly unstable or volatile. It’s an inherent feature computer scientists obviously don’t like since it means inputs can be lost or altered and calculations either incomplete or erroneous. “We are at the very infancy of the technology. Today, you don’t have long before you lose control of the quantum system and its interaction with the environment takes over. And now the results from the quantum computer are no longer useful,” says Dr. Tyler Takeshita, Daimler’s quantum technology expert. He admits that controlling a quantum system has been extremely hard so far. “The goal everyone is chasing is to have enough control over these quantum systems to perform reliable calculations on meaningful applications.”

While the general concept has been around since the 1980s, only recently did researchers show that this vision can be accomplished. Search giant Google garnered a lot of attention when it announced what it called “quantum supremacy” in October 2019 in a paper published in the science journal Nature . Scientists at its research lab in Santa Barbara, California managed to have their quantum system perform a mathematical calculation in three minutes and 20 seconds that today’s most powerful supercomputers would need more than 10,000 years to complete.

“It clearly shows that the quantum computing community is making outstanding progress,” explains Takeshita who left the University of California, Berkeley, to join the Daimler initiative. “A quantum computer hasn’t done something that will change the industry — yet. This announcement,” he adds, “gives you the needed hope. Quantum computing is not new, but it wasn’t clear for a long time if it can be done at scale. Experiments like this are a major milestone on the path to scalable quantum computing.” He points out that scientists don’t even know yet how many qubits a fully functional system should contain. “We’ve learned along the way that the sheer number of qubits alone is not enough. We also need to understand what problems are good applications for this technology and possibly design new algorithms to solve these problems. You only discover these important details if you invest time and money in this technology from the very beginning.”

Luckily, the worldwide community of quantum computing experts and aficionados is relatively small and keeps in touch about their ideas and accomplishments. The scene reminds one of the 1950s when the first computers started coming out and people only had a rudimentary understanding how to control and use them. It’s a small, young community that has grown over the last three years and is well connected. One estimates that the global quantum community to number no more than 2,000 people. So far, there is little competition and the spirit of collaboration is prevalent. Daimler, for its part, has been conducting first experiments with both IBM and Google and has established a growing network of universities in Germany and the U.S. where Ph.D.s work in conjunction with the company’s own scientists. “We are open for new ideas all the time,” says Russell Seeman who is the program manager of the quantum initiative, tasked with making sure it runs smoothly. As he tells it, enthusiasts inside Daimler and in academia offer input in the form of contributions and their time. “People in many disciplines — material scientists, physicists, chemists, mathematicians — face problems that they have been trying to solve their entire careers. Yet they couldn’t figure out whether the technology exists to help them do it. We can start a dialogue and can bring them into our program if there’s a fit.” The outreach includes keeping track of and scouting for startups in the field that explore their own approaches to quantum machines.

Hardware from the cold

The research partnership with both IBM and Google is particularly valuable to drive the field forward by exploring what-if scenarios. “We made contact very early and really work together. It’s not about testing their hardware. Instead, we bring real-world problems to them and work together to figure out how we can best approach this technology,” says his program partner Takeshita. IBM runs quantum computers in its historic headquarters of Yorktown, New York, while Google’s systems stand in a lab in Santa Barbara in Southern California and Venice Beach by Los Angeles. While the machines differ in the architectural and operational details, these quantum computers rely on superconductivity. That means they have to be chilled down to minus 272.9 degrees Celsius (minus 454 on the Fahrenheit scale), making them one of the the coldest spots in the known universe.

Daimler researchers access those systems remotely over the internet like an employee would access a productivity suite or database in the cloud. The work involves tossing basic questions around such as what queries to run on a quantum computer once it reaches a certain level of maturity or stability. “We have an ongoing discussion about what we can do on such a system and develop the next level. You cannot just take existing algorithms and run them on a quantum machine, you have to formulate the problem differently and think about how it could be computed.” By early 2020, this basic research work with the tech companies had resulted in three peer-reviewed papers published in scientific journals.

Progress on batteries

Battery technology is the most promising exploration of quantum computing’s potential to date for Daimler. Namely, using a powerful computer to simulate what new battery chemistry is more efficient and long-lived, made with more abundant or less harmful ingredients. The way nature connects or binds two atoms is not dissimilar to how a quantum computer works. So it was an obvious choice to look into how this new computing paradigm can be used as an advantage. In this particular case, a team of scientists at Daimler and IBM looked into whether a quantum machine can compute and accurately simulate the fundamental behavior of lithium battery materials. Explains Daimler scientist Dr. Eunseok Lee, who is also part of the core quantum team in Silicon Valley: “Traditionally, you had to predict chemical reactions by either running experiments and observing the results or calculating them as best you could. There is a new trend in materials science, called Materials Informatics, that combines machine learning and chemical simulation. This new method hopefully can save us time and money.” The key question here is whether the quantum computer can replace or supplement some of the classical calculations that battery researchers are performing now. If so, it can be a valuable shortcut: “A quantum computer, if developed successfully, will accelerate the advance in Materials Informatics,” says Lee.

Fast and economical

Over time, this tool stands to supercharge the standard workflow to improve batteries for electric vehicles. Quantum simulations could be used to run through hundreds of thousands of chemical reactions and quickly identify a few hundred that human experts would then examine more closely. The shortcut would come with a whole list of benefits. Today Lithium-ion batteries are based on a chemistry which includes the precious and toxic elements Cobalt and Nickel. Replacing these elements with an environmentally abundant and non-toxic element such as Sulfur requires typically hundreds of thousands of experiments in laboratories. Computers can already help to make predictions which of the experiments might be the most promising ones. Nonetheles, today's computers are too slow to predict complex chemical reactions. This is where Quantum Computing could help in the future: speeding up the progress in chemistry and material by replacing experimental work in a lab with simulation in computers to get results much faster at higher accuracy.

Quantum computers also translate into significant energy savings compared to running calculations on today’s server farms. One estimates that, depending on the architecture, a quantum machine may consume 25 times less energy for the same task than a supercomputer just by being so fast. Daimler AG is pushing ahead with the transformation to zero-emission mobility. IT also makes an important contribution here.

It’s the first, promising step on a longer journey, to be sure. Today’s quantum machines are not advanced and stable enough to simulate molecules consisting of more than two to three atoms, thereby excluding a sizable portion of the periodic table from closer scrutiny. The same goes for other practical applications on the horizon, say crash test simulations or recommendation engines for configuring complex, feature-rich products which require crunching many variables. But the experts at MBRDNA expect rapid progress to get there. “In several years or so we will have a whole new generation of quantum computers. We don’t know what they will look like and what they will be able to do, but we have already made a lot of progress in only a few years we can build on,” says Tyler Takeshita.

The chemist doesn’t expect a future where problems are handled by either classical or quantum systems, but rather a hybrid approach that fluidly blends the computing resources depending on the problem at hand. Under this scenario, big platforms such as Amazon, Google, IBM or Microsoft in conjunction with specialized quantum startups would provide cloud access to a new form of hybrid computing on demand.

“What matters is that we build our core competency around quantum computing now, which will allow us to grow with the field over time a be able to solve problems efficiently in the future. It can be a catalyst for Daimler as a whole,” Takeshita explains. That way, engineers, chemists and programmers will soon be able to just tap into a new resource to do their work, regardless of what type of computer solves their problem. What matters is performance, speed and useful applications.