Calculating the unimaginable

Imagine you are spinning a basketball on your finger. Under the laws of nature, the ball spins either to the right or to the left, but never in both directions at the same time. The world of quantum computing is different: Here, the ball can revolve in both directions simultaneously.

The smallest units in quantum computers are qubits. Unlike classic bits, which can have a value of either 0 or 1, qubits can be both 0 and 1 at the same time. This state of being two incompatible things at the same time is also known as a ‘superposition’. Whereas classic bits operate independently, qubits interact with one another, creating interdependencies known as ‘entanglement’. These qualities enable complex operations and exponential growth of states: one qubit represents two states, 0 and 1, two qubits four states, three qubits eight states etc. As a result, quantum computers can solve certain types of problems much quicker than classic computers. They are not constrained to stepwise calculations but, rather, can compute a vast number of possible solutions simultaneously—and at a speed that is far beyond anything we can imagine.

Here’s an example: Imagine an archipelago with thousands of islands connected by bridges. You are looking for a route that crosses each island only once. Assuming there were a million possible routes and only one correct one, a classic computer would take 500,000 attempts on average to find the right solution. A quantum computer, by contrast, would solve the problem in just 1,000 attempts, or 500 times faster.1

Quantum computing is difficult to comprehend for most people—even Bill Gates described it as being like ‘hieroglyphics’ to him. For Martin Strahm, Head of Data Science in pRED (Pharma Research and Early Development), that is precisely what makes it so fascinating: “We cannot understand that something can be at two locations at the same time. Children can do that, they are capable of magical thinking, but we are not anymore. For me, quantum computing is a little bit like magic. It is unimaginable and yet we can calculate it and we can even build machines to do things that are unthinkable.”

Bryn Roberts, Head of pRED Operations, is also excited by the possibilities of quantum computing. Under his leadership, a task force was set up several months ago with the aim of monitoring the field, developing collaborations and piloting early applications.

Mariëlle van de Pol, Global Area Head, Technical Solution Delivery & Architecture in pRED, heads up the task force and says she sometimes feels like a surfer watching the swells: “We are scanning the horizon, waiting for the big wave, but we don’t know how big it is going to be, or when it will come. But if you see how much tech companies are investing in this topic, and how quickly the whole landscape is evolving, you realise that it will come and it will be a game-changer.”

Being prepared—this is the mission of the task force. They have adopted a two-pronged approach: an academic and a business track. On the one hand, they collaborate with three doctoral students from the University of Oxford, who apply quantum computer simulations, for example, to calculate the energy of molecules. On the other hand, with the support of Roche Partnering, they investigate possible partnerships with big tech companies which possess both the know­ledge and the hardware and software required. Yet Roche has something that the tech companies of this world do not: an in-depth understanding of biomedicine and, conse­quently, the ability to develop specific software and algorithms for applying quantum technology in pharma. This competitive advantage is something Mariëlle and her team are looking to defend and extend.

One of the most promising applications is the simulation of molecules and their chemical behaviour, which would enable faster and more precise development of new medicines. Quantum technology could also be used for quantum-­powered neural networks in machine learning, allowing us to solve optimisation problems, for instance in protein folding. In biomedical image analysis, quantum computers could help detect topological changes that are caused by the disease. And there are many other applications beyond R&D, for instance in production, finance and IT.

There is still a long way to go before quantum computing will be ready for prime-time, however. Optimists cite a figure of five years, realists cite a figure closer to 15 years. A major challenge is the hardware. Qubits remain extremely unstable and susceptible to external influences. In comparison with our conveniently sized laptops and smartphones, quantum computers look downright old-fashioned—occupying large rooms with racks of specialised equipment and huge refrigerators to super-cool the quantum chips, containing only a handful of qubits. A much higher number of qubits will be required to achieve meaningful impact for Roche. The key question is, therefore, which hardware solution, from the many being explored, will realise both qubit stability and scalability. Andreas Maunz, Scientific Application Developer in pRED Informatics, sums up the situation as follows: “We’re really waiting for the physical implementation. The theory is there, but the machine still has to be built.”

One thing is certain, however: Quantum computers have enormous potential. Some people even believe that this technology will be capable of solving humanity’s big problems, from climate change to the battle against various diseases. Data scientist Martin concludes: “I do not think that quantum computing will solve all the world’s problems. But what is really fascinating is that even quite a small quantum computer can do calculations that would take the lifetime of the universe on a conventional computer. What you can calculate is really unthinkable. But will it really help huma­nity? That depends on us.”

One of the most promising applications is the simulation of molecules and their chemical behaviour, which would enable faster and more precise development of new medicines. Quantum technology could also be used for quantum-­powered neural networks in machine learning, allowing us to solve optimisation problems, for instance in protein folding. In biomedical image analysis, quantum computers could help detect topological changes that are caused by the disease. And there are many other applications beyond R&D, for instance in production, finance and IT.

1. BCG Henderson Institute [Internet; cited 2018 December]. Available from