This article is taken from the monthly Sciences et Avenir – La Recherche n°902, dated April 2022.
It’s a revolution that is coming, but it will be done gently. The good old classic bit computer, perfectly suited to its current office automation or entertainment tasks, will be enthroned in our offices and living rooms for a long time to come. It is in major research laboratories that the quantum computer should make its appearance in the next few years. From chemistry to finance, passing through fundamental physics, a short overview – not exhaustive – of the playgrounds of the computer of the 21st century.
Health: tailor-made medicines
High-performance computing, using supercomputers, is already at the heart of chemical research. “For about twenty years, chemistry has fully embraced the digital, explains Jean-Philip Piquemal, director of the Theoretical Chemistry Laboratory (Sorbonne University, CNRS) and co-founder of the start-up Qubit Pharmaceuticals. Molecules and drugs are partly created by computer. For this, theoretical chemists try to solve quantum mechanical equations numerically. These calculations make it possible to determine the properties of molecules, which can be very varied: the elasticity and resistance of a material, the interaction of a drug with a protein, etc. “But to simulate the quantum nature of matter, what better than a quantum computer?
This idea finds a particular echo among chemists as great progress is constantly announced around the design of this machine. “What is important in chemistry is to be predictive, to be able to compare the theoretical calculation with the experimental results. We know what level of precision must be had in the equations to be close to reality. But we use in many many cases of approximations to run the models, because more precision would make the calculations too long and complicated for classical machines. With a quantum computer, it would be possible to limit these approximations “. Powerful quantum computers would thus make it possible to explore the world of chemistry like never before.
“The big winner will be the pharmacy. When you have a target, for example SARS-CoV-2, you want to know if a drug is going to latch onto it to neutralize it. And for that you have to solve a physical equation. More we solve it exactly, the better the prediction. What we are going to try to do is to predict a lot of interactions between molecules, at very high speed and very high precision “, says Jean-Philip Piquemal. Because there are potentially billions of possible combinations to find “the right molecule” which will be used against a specific disease. A quantum computer capable of performing an astronomical number of simulations in record time would change the given. “The preclinical phase, where we design the molecules, would be much more efficient, not only to find the good molecule but also eliminate those which would have prohibitive side effects. This can save several years in the development of a drug! “
Matter: the nucleus of atoms revealed
Fair return, the quantum computer should play a big role in fundamental physics, from which it is derived. As described by Denis Lacroix, researcher at the Irène-Joliot-Curie 2 Infinite Physics Laboratory (Paris-Saclay University, CNRS), “atomic nuclei are made up of protons and neutrons, the number of which can range from two to several hundred. In addition, they are in very intense interaction. If many phenomena occurring in nuclei, such as radioactive decay, are quite well understood thanks to models resulting from experimentation, the tendency is to study them by making exact calculations, starting from the strong and weak interaction at work in the atomic nucleus. problems of 15 to 20 particles maximum “. A quantum computer will simulate systems with more particles.
The possibilities for nuclear physics do not stop there: “The nuclei are like liquid drops which can take many forms (rugby ball, pear, etc.). Today, it is possible to describe systems with two or three types of deformations at most. Quantum machines will exceed this limit and will explore many more configurations, with repercussions in the description of the dynamics of nuclei: collisions between heavy ions, nuclear fission, etc. “, explains the physicist. Will this work have consequences on nuclear technologies? It is still too early to say.
On the other hand, here is an area where quantum is already there: data processing. Scientists from Cern in Geneva, in collaboration with companies like IBM, have shown that quantum algorithms can already compete with the best classical algorithms. “To find information among the gigantic amount of data produced during collisions between particles at the LHC (1 petabyte (1015 bytes, editor’s note) per second!), we use machine learning, or machine learning. These are algorithms allowing an artificial intelligence to “train” to carry out a task (for example sorting data) using already known data. It has been shown that a generalist algorithm adapted to quantum performs as well as the classical algorithms of Cern specially designed to detect particles “, indicates Xavier Vasques, of IBM France. Results which are not only theoretical, but well tested thanks to the first quantum processors out of the laboratories of IBM and its competitors (Intel, Google), and by means of simulators of quantum calculations.
Simulation of the atomic nucleus deuterium (formed of a proton, in red, and a neutron, in blue) obtained using a quantum computer. Credits: ANDY SPROLES/OAK RIDGE NATIONAL LABORATORY, US DEPT. OF ENERGY
Materials: very small-scale simulations
Manufacture new materials with properties chosen according to their applications. A dream that the quantum computer will make it possible to touch. Examples abound: “cleaner” fertilizers, catalysts for capturing carbon dioxide, and even new batteries, as illustrated by Xavier Vasques of IBM: “We work with Daimler (the parent company of Mercedes-Benz, editor’s note) to design the batteries of the future. For example, we are interested in lithium-sulphur, a much-studied couple. Our teams have shown that it is possible to simulate these molecules and their interactions, with a considerable time saving on quantum machines since we go from 290 days to 7 hours for a simulation that we have actually tested! With a better understanding of these molecules, it is hoped to design batteries with more recharge cycles and which also last longer, made with less toxic materials than those currently used. “
A simulation of the behavior of matter on very small scales can also have effects on very large scales, as summed up by Marc Porcheron, head of EDF’s IT and quantum technologies project: “We use simulations a lot to analyze the evolution over time of structures (metals, concretes, etc.) found in nuclear power plants. They can be subjected to very high temperatures or radiation, and we have need to predict what is happening inside over the long term, to take the appropriate security measures. “Ideas that can also be applied to hydroelectric dams.
Networks: optimization at all costs
The capacity of the quantum computer to manage many parameters is also used at our level. Marc Porcheron, from EDF, explains: “All power plants do not operate permanently, their activity is linked for example to the consumption of households and industries at time T. To decide whether to stop or start them, it is necessary to take into account the demand, the constraints of the plants (maintenance, reloading, etc.), production from renewable sources, etc. Based on these parameters, we must decide whether to open the valve of a dam, whether to start this or that gas-fired power plant… Quantum algorithms could enable faster and more accurate decision-making, which saves time, money, and increases confidence in the safety and reliability of the electrical system.”
They should also improve the precision of probability studies, for example on the failure of power plants: “These studies answer questions such as ‘what is the probability that such an event will happen (a failure for example) if a first event occurs?’ We build ‘event trees’ which quickly become gigantic. Quantum algorithms allow you to explore them and calculate the probabilities associated with each event more quickly and efficiently. “Another example of optimization is fast charging stations for electric vehicles. “With hundreds of thousands of vehicles and thousands of terminals, being able to tell each driver which way to go, depending on the charge of the batteries, the position of the vehicle, is very complex to solve by calculation. Here again, quantum algorithms will solve these problems more efficiently. “Finally, this type of application is of interest far beyond EDF, as evidenced by IBM’s partnership with ExxonMobil to improve the maritime transport of fuel. Road traffic, production lines… This is undoubtedly only the beginning of a vast optimization of networks, of all kinds!
Finance: ever faster
IT has a central place in the world of finance: transactions (trade) at high frequency, consisting of buying and selling shares in very short periods of time to make the most of the market, use complex algorithms that require a lot of computing power. Analyze stock market fluctuations even faster, simulate market behavior faster than competitors from a growing amount of data. seem so in the quantum computer’s strings…
These assets have not escaped the financial players. The teams of the Spanish bank BBVA and the start-up Multiverse Computing, for example, published an article at the beginning of this year on the optimization of investments using quantum algorithms. This is probably only the beginning of a wave of publications that will develop “quantum finance”.
Quantum programming… at home!
What if tomorrow you learned to program in quantum? Resources already exist to practice it, accessible through a simple search on the Internet! Microsoft and its Q# language (pronounced Q sharp), IBM’s Qiskit and Google’s Cirq tools (which allow quantum programming in Python language): they are open-source and have many online tutorials (mostly in English). Of course, having solid notions in mathematics (in linear algebra in particular) and in programming is essential, especially if you want to develop your own algorithms.
But there are also simplified uses such as the Drag & Drop interface at IBM, or quantum algorithms already built and simply “to be integrated” into larger programs: enough to create your own random number generator or implement a says Grover’s search for finding a keyword quickly among an entire list. And why not, then, test its programs on real quantum computers (or simulators). IBM Quantum, Microsoft Azure Quantum, Google Quantum Computing service… Solutions for cloud allow access to these machines. The field is thus experiencing an emulation similar to the beginning of computing, with unprecedented (relative) ease of access.
By Corentin Paillassard