Quantum Computing Accelerates Materials Design & Discovery

Quantum computing has the potential to disrupt almost all industries, with exponential increases in performance compared to classical computing. The interest in quantum computing comes from the considerable amount of computing potential in quantum bits (qubits), which are exceedingly difficult to manage, both in terms of quantity and quality. OTI Lumionics is implementing quantum methods to improve its Materials Discovery Platform that applies state of the art high-performance computing (HPC) simulations along with machine learning (ML) algorithms to design production-ready advanced materials without a wet-lab. The platform provides faster material simulations, more accurate property predictions, mapping of excited states and modeling of chemical reactions. Over the past year, OTI has demonstrated its ability to simulate commercially relevant problems, such as the light-emitting metal complexes used in OLED displays, using quantum methods. In an interview with EE Times Europe, Michael Helander, CEO and co-founder of OTI, stated how OTI’s latest results show that quantum methods can exceed the capabilities of traditional simulations when run on both quantum hardware and quantum-inspired systems. OTI is now working to package these algorithms into a general purpose quantum computational chemistry tool. In addition, Microsoft recently presented its Azure Quantum service, designed to provide early users with a scalable path to quantum computing while other companies enter the emerging quantum computing software market. OTI currently runs its quantum algorithms on Microsoft and D-Wave systems.

Quantum technology 

“The principle of material discovery is explained in the name; it is discovery,” said Helander. The main objective of OTI’s platform is to run simulations and ML algorithms while creating statistical probability maps, showing what properties the candidate material may have and what problems it may encounter. “This allows our scientists and engineers to make informed decisions about which materials to synthesize and distribute in production tests,” said Helander. For the study of the properties of materials such as OLEDs in particular, a multitude of equations, such as Schrodinger’s, involves the use of considerable computation. Most of the phases of these simulations involve the calculation of electron integrals, the inversion of matrices, and the optimization of coefficients. The results of these simulations make it possible to define the optoelectronic properties of materials prior to synthesis, thus helping the company save costs. Helander highlights how the ML approach supports well-defined combinations such as the use of deep neural networks, already used for facial recognition. “Our algorithms map molecular information to define candidate materials to determine how to best organize our experiments to maximize success in finding a material,” said Helander. “We have spent a lot of time working with most major hardware suppliers to identify the best process. Some of them are Fujitsu and D-Wave,” said Helander. He added, “The hardware has to be mature scale, low error rate. At the moment, the only real quantum hardware we can actually use for these large scale problems is the D-wave machine. There’s a lot of excitement and other hardware being built by IBM, Google and Honeywell computing, which all present a ton of potential, yet we are still many, many years away from having useful qubits and low enough error rate. Until we make that system, we would need thousands of qubits or 10s of thousands of qubits in a normal kind of gate, basically quantum computing hardware to make it clickable.”

Conclusion

The use of quantum algorithms in artificial intelligence techniques will increase the learning capabilities of machines. “I think the whole challenge for quantum computing as a whole is, being a totally new paradox for computing as research to define a standard. So if we look at classical computing, we know we have companies like AMD and Intel that are very mature, like Microsoft doing the operating system apart. For quantum, it is not clear what the offer will be and what kind of programming language we will use, which makes it difficult for anyone to bet on who will be a winner. We think that, over time, there will certainly be a need for some consolidation of the industry and some sort of closer integration between hardware and software. But that’s a bit too much for the industry as a whole at the moment,” said Helander. The involvement of household names like IBM, Google or Amazon not only adds credibility to quantum computing but also spreads its fame. With such firms involved in this market, it is easy to get quantum computers onto the homepages of news outlets that ordinarily don’t cover advanced physics or supercomputing.