- #Introduction -
France ranks among the world’s
top leaders in mathematics, with
globally renowned universities
and institutions that produce
high-caliber scientific talent in
the field of mathematics and related
fields in computer science.
École Polytechnique has its part in this historical tradition of excellence that emerged during the Age of Enlightenment, remaining at the cutting edge of disciplines whose applications are shaking up the world of today and tomorrow. With their tremendous innovative potential, mathematics and computer science are omnipresent in our society, sparking breakthrough innovations to transform our daily lives.
École Polytechnique researchers in these different fields work from cross-disciplinary perspectives, conducting cutting-edge research that relies on interdisciplinarity to arrive at new and original discoveries, advance our understanding of the world, and design new technology.
In doing so, they follow in the footsteps of the most illustrious contributors of the institution’s 225-year history.
Since its creation, École Polytechnique has had, among its alumni and professors, numerous scientists, captains of industry, and high-ranking government officials, whose legacy lives on.
Today, from the work carried out by l’X researchers and students at the intersection of various scientific disciplines, innovations are arising across all areas of high growth, such as numerical simulation, cybersecurity, additive manufacturing, sound spatialization, augmented reality, blockchain, 3D modeling and e-health.
This publication aims to put the spotlight on the undergraduate students, PhD students, professors and researchers of École Polytechnique for their contributions to the world and to science.
They are the next generation that will drive innovation in France and abroad. We invite you to discover ten flagship projects that allow us to imagine the science of tomorrow, as well as texts by five experts that provide insight into the disciplines of priority at l’X.
Through these innovative projects, École Polytechnique reaffirms its ambition to understand and change the world of today and tomorrow, and to focus on subjects with a strong social impact.
- #DigitalSimulation -
Researchers from l’X and Inria have managed
to simulate a beating heart numerically and
in real time.
To achieve this, they created a simplified reduced-dimensional model of the 3D physical model produced four years earlier; the latter was too complex to be applied in real time, since it took about six hours to simulate numerically a single heartbeat.
This breakthrough opens up prospects for predictions in the medical field. “By integrating signals such as electrocardiograms or pressure measurements, this model, once it has been adapted to the individual patient’s data, could be used as a tool to diagnose the current clinical status of a person’s heart, and more importantly to predict future anomalies or diseases,” explains Philippe Moireau, head of the joint l’X / Inria project team.
In 2018, the researchers filed two patents which are currently being evaluated.
The first is to detect day‑to‑day abnormalities by seismocardiograms, measuring cardiac vibrations as they affect the thorax.
This lends itself to integration with wearables and other connected devices. The second patent concerns monitoring during anesthesia.
As part of a partnership between l’X, Inria and Lariboisière Hospital (AP-HP), two anesthetists are working with the team on the application of this digital model to monitoring.
The aim is to reduce the risk of rare events by making use of models and data collected in real time during anesthesia.
In the future, according to Philippe Moireau, this approach could be combined with a predictive model and big data to yield even more accurate predictions.
- #Cybersecurity -
With the rapid development of connected
devices, ensuring security has
become a priority.
Cryptography protects the authenticity and integrity of communications by preventing an attacker from interfering with access to data, computer accounts or the physical resources of hacked machines.
Cryptography researchers at the l'X Computer Science Laboratory are working on new methods of ensuring the security of communications.
When two communicating parties need to share an encryption key, one method is to hide it in the solution to a complex mathematical problem.
Since the 1970s, arithmetic problems were used for this purpose but the very large keys they yielded are not suitable for current communication tools. The new solutions developed by l’X researchers are based on elliptic curves whose points, characterized by their own algebra, can be reinjected into the arithmetic problems as if they were numbers.
This yields very concise mathematical problems, extremely difficult to solve, which are suitable for current devices such as cell phones.
To further reduce the size of encryption keys for small connected devices, researchers at the Computer Science Laboratory are studying problems that will be used in the future, shifting from curves to surfaces, such as Kummer surfaces, to create even more efficient cryptosystems with compact and secure keys.
- #VideoGames -
He’d like his mobile game to become the “Candy Crush” of science.
Raphaël Granier de Cassagnac, a researcher in particle physics at CNRS, introduced video games to École Polytechnique.
Research director at the Leprince-Ringuet Laboratory, he works with a small team of designers and artists to develop a game dedicated to particle physics.
The idea of this game is to immerse players in a "fantasy" universe where they explore the world of elementary particles.
“As they progress, they discover one by one the bricks of the standard model, making machines, using telescopes, and observing particle showers,” explains the physicist.
In addition to his role as a scientific expert, Raphael Granier de Cassagnac co-wrote the script for the game. For many years, he has also been writing science fiction novels, role-playing sketches, and different film projects. At the same time, he has been active in a joint project between École Polytechnique and Ubisoft, creating a research chair in “science and video games.”
Signed in early 2019, it aims to introduce more science into video games and train professionals by offering new courses at l’X. “It could be a game design or storytelling course in the context of human science education, or a course geared towards game engines in connection with computer graphics and 3D animation,” he explains.
Another aim of the chair is to study education methods through the use of games and bring together scientists and industry at a symposium devoted to science and video games.
- #AdditiveManufacturing -
The success of 3D printing has led to growing
interest in the shape and topology optimization
algorithms developed by École Polytechnique’s
Center for Applied Mathematics.
These algorithms, used to optimize the design of industrial parts, have been around for some twenty years but their field of application was limited, until recently, by traditional manufacturing processes.
“The inventiveness of the algorithms was limited by the need to respect such design constraints as demold restrictions,” explains mathematician Grégoire Allaire, a specialist in structural optimization.
Thanks to additive manufacturing, especially metal AM, components created by these algorithms, which were previously too difficult to build, are now conceivable.
But this new freedom goes hand in hand with novel constraints. “The process, which consists of melting at high temperature successive layers of metal powder using a laser heat source generates constraints linked to the risk of deformation when the metal cools,” says the mathematician.
To counter this risk, optimization software starts from an initial shape and calculates, through an iterative process, a better solution than the previous one. This software works to minimize deformations and improve such criteria as the weight of the structure, its strength and the resistance of the materials, until the best possible shape is generated.
These metal additive manufacturing processes are paving the way for new, original applications, but some obstacles subsist, including cost.
To address this, l’X is participating in the Sofia program, alongside industrial partners and academic institutions, to provide research tools that contribute to the development of the French metal additive manufacturing sector throughout the world.
- #SoundSpatialization -
Researchers in applied mathematics have developed a new type of audio
navigation technology for the visually impaired.
The goal is to make it possible for the latter to practice sports with near autonomy. Using simple headphones and the 3D sound spatialization software, they can follow the movement of a sound object in space.
Complex mathematical calculations were used to precisely model the functions for the construction of a realistic sound space. By numerically calculating the propagation of sound waves in a given environment, say an auditorium, a historical monument, or the human ear, mathematicians strive to reproduce the characteristics of waves in each point of space.
Coupled with a positioning system on the head and body, the software can help the blind navigate safely.
During experiments conducted in partnership with the NGO MixHandi Cap sur la vie, Stéphane le Sueur, a blind athlete, managed to run and rollerblade by following a virtual sound guide in front of him.
This research is meant to enable the visually impaired to practice sports activities with greater autonomy and thereby offer a solution to an important social problem.
- #ArtificialIntelligence -
The cost of losses due to shoplifting in the
retail industry is estimated between
1% and 2% of turnover, representing
$120 billion per year in the world.
Three students in the Data Science for Business joint l’X/HEC master’s program decided to help retailers tackle this problem.
They created the start-up Veesion to market their video gesture recognition technology, an artificial intelligence program they developed to automatically detect thefts in real time.
“Our technology is based on a deep learning algorithm, a technique that uses neural networks to analyze sequential images,” explains Thibault David, co-founder of Veesion.
Their artificial intelligence can connect images together to identify the sequence of movements.
A large store with a hundred cameras commonly invests up to several million euros is video surveillance systems, which are not cost-effective given the low rate of shoplifting actually avoided.
“An operator can simultaneously process the images of up to six cameras, but they are being asked to analyze images from a hundred screens,” says Thibault David.
The start-up proposes a human-machine interactive system, with an AI filter that continuously analyzes all of the screens for theft probability, and only alerts the human operator in the case of a sequence with a high probability.
Their technology will be deployed in France and internationally in 2019.
The team plans to continue its research on video motion detection by recruiting PhD students to improve its software.
- #AugmentedReality -
Five students from the Ingénieur Polytechnicien
Program have launched the start-up Theatrall
which proposes individualized, automated
captioning of theatrical performances in real
Their innovation grew out of a second-year group science project and won the French final of the Microsoft Imagine Cup 2018.
It is based on voice recognition. “Our algorithm can listen to the actors, understand the overall meaning of their words, situate it in the script and communicate it to viewers in the language of their choice,” says Daniel Huynh, one of the founders of the start-up. Based on artificial intelligence, the system is designed for use by the hearing-impaired, among others.
The program is available through an app on a smartphone or tablet or on smartglasses for the display of surtitles.
This innovation, which has already been tested in partnership with the Comédie Française, is likely to interest many other institutions. The system has already attracted the interest of a luxury brand and the students are thinking about applications abroad.
“The musicals market is very well developed in England, and the appetite for theater is very widespread in Germany and Italy too,” says Daniel Huynh who also plans to provide the system to movie theaters to reach foreign audiences.
- #Blockchain -
Fifteen years ago, before the concept even
existed, professor of innovation management
Thierry Rayna was already working on a forebear
of the blockchain.
Nowadays, his research is devoted to studying the impact of digital technologies on business models.
After studying other emerging technologies, such as cloud computing, 3D printing and the IoT, his research now focuses on assessing the relevance of use cases of blockchain for companies.
Although each type of digital technology is unique, it is still possible to find common trends.
Nowadays, consumers have increasingly become producers; they create digital content by putting videos and photos online, deliver services by renting out their apartments, and in the future 3D printing will even enable them to manufacture objects. But for that to happen, a trusted intermediary needs to emerge.
Blockchain frees them from this burden, leading to far more opportunities (and challenges).
By accelerating the emergence of new markets, blockchain causes disruption and shakes up existing markets. “For start-up companies, it provides enormous opportunities that do not necessarily involve competing head-on with, but rather circumventing already established players in their market.
The latter, as a result, will be faced with a need for transformation and to adapt their skills and products to these new uses,” explains Rayna.
Bearing this in mind, he has developed a diagnostic tool to assess digital technologies and their impact on existing business models.
The aim is to grasp both the technology and the market ‘rules’ while establishing a connection between the two to anticipate the disruptions likely to occur with the emergence of a technology. Because sometimes, there is no need to actually use a technology for a disruption to occur; the mere threat may suffice.
Leonardo da Vinci, Clément Ader and the
Montgolfier brothers all shared an interest in
designing flying machines.
Over the past two years, second year l’X students have been devising their own flying objects, modeling them in virtual and then printing a 3D prototype during lab sessions as part of an Applied Laboratory Module (ModaL).
“This module relies on advances in computer graphics towards virtually modeling and then materializing an imaginary shape.
It draws on video games and special effects while addressing a key engineering issue: representing and animating a mental picture,” comments Marie-Paule Cani, Computer Science professor at l’X and head of the module “Expressive modeling for 3D manufacture.”
These sessions led last year to the creation of spacecrafts inspired by Star Wars, the films of Hayao Miyazaki, or nature.
Using selft-designed computer-aided creation software, based on a form of artificial intelligence called “creative AI’, the students sketched their concept and then “sculpted” it virtually in 3D.
The principle involves coding the knowledge needed to infer a 3D model in real-time from a 2D sketch, to deform it and to add details using hand gestures.
This methodology is being designed by Marie-Paule Cani and her team at LIX in their research.
For example, their mountain sculpting project, accessible at Mus’X, École Polytechnique’s museum space, utilizes geological constraints, such as the motion of tectonic plates or erosion, for users to animate in real time through hand gestures the formation and evolution of a mountain range.
Students in this module thus have the opportunity to gain a firm grasp of research being conducted in the field of expressive 3D modeling.
- #E-health -
It is estimated that nearly 640 million adults
will have diabetes by 2040.
The start-up Healsy, with the help of École Polytechnique’s start-up accelerator, has developed the first mobile platform to help people with diabetes to make their own treatment decisions.
This is important because blood glucose measurements alone do not suffice for an individual to determine the right dose of insulin.
As co-founder Nicolas Caleca (X2004) explains, “there are so many factors to consider, including meals, physical activity, stress levels, alreadyinjected insulin, that diabetics make mistakes, which may cause dramatic short- or medium-term complications.”
Healsy's goal is to provide patients with the information they lack –that is, the prediction of changes in their blood glucose levels.
To make this prediction, patients connect their smartphone to their devices –blood glucose sensors and insulin pump– and to an activity tracker.
“On the basis of all this data, the app can simulate the evolution of blood glucose using a mathematical model coupled with artificial intelligence that we develop,” says Nicolas Caleca. A first version of the app will be available on the market in 2019.
To help the startup develop its technology, as part of their second-year science project, École Polytechnique students designed a glucose monitoring and treatment armband providing patients with continuous glucose levels in real time on their smartphone, then Healsy built a database with the information collected.
For those who may still have entertained any doubts, the year 2018
demonstrated once again that the digital threat is anything but
insubstantial and that challenges to the security and stability of
cyberspace remain immense.
More sophisticated, better planned, more destructive and with an impact on the whole of society, from the individual to big business to our democratic institutions, computer attacks have entered a new dimension.
All connected, all concerned, all responsible: this is the fundamental approach that we strive to bring to bear. Security has to come out of its restricted field and involve all architects of the digital society.
For beyond the threats to society, the economy, and the sovereignty and stability of cyberspace, the very development of technologies is at stake. Indeed, the impressive uses made possible by digital technology can only be sustained if they are trusted by users.
Cybersecurity: more than a cost
This involves first and foremost a change in our outlook on cybersecurity. It can no longer be considered a cost component or a kind of patch to be applied at the end of the innovation process. Ask ANSSI's experts: cybersecurity is in itself an exciting field of innovation, scientifically rich and highly transdisciplinary, which brings together a wide variety of players, both public and private, in France and abroad. It poses major intellectual challenges for innovators of all kinds.
A challenge beyond the realm of engineering
How can future technologies, such as machine learning, connected health, and quantum computing, be secured?
The field of autonomous connected vehicles, which has experienced significant progress in recent years, provides a good illustration of the overlapping of uses and security requirements.
The presence of driverless cars on our roads remains largely determined by pressing questions of trust and social acceptability. And there is much that remains to be done: the recognition system of these vehicles can still be easily fooled by a slight alteration of traffic signs, leading them to confuse a stop sign with a priority route sign.
These challenges naturally concern engineers, but also architects of public policy, lawmakers and experts in international relations. What can be done to achieve cyberspace stability?
Should we allow private players to take justice into their own hands, to respond to attacks in a context where businesses themselves have become “battlefields”? The stability of cyberspace is a subject that is impacting political, diplomatic and military habits. It brings up numerous issues along with exciting, fascinating, structuring prospects.
Engineers, lawyers, designers, experts in public policy, and international relations, ergonomists, start-ups, large groups, individual citizens – digital security is definitely everyone's business.
"The stability of cyberspace is a subject that is impacting political, diplomatic and military habits."
Technological innovations made possible by the recent spectacular
advances in materials science—touch screens, sensors, electronic components—
are everywhere around us.
The “wish list” for future innovations is long and varied. To cite just two: we are looking for a non-toxic, chemically stable and, of course, low-cost material that can be used as a permanent magnet in electric car engines or in wind turbines; and we are hoping to find materials to manufacture a battery that can store the energy production of an entire photovoltaic farm.
But how are we to proceed? Are we to test each and every new chemical compound, at the risk of losing our way? The ideal would be to develop methods to predict the properties of materials based on a combination of theoretical considerations and computations!
Digital technology for materials science
This is where condensed matter theorists from the Center for Theoretical Physics come in. The development of numerical tools to describe the properties of materials from “first principles”, that is, without adjustable parameters, is precisely at the heart of our project, “Predictive electronic structure calculations for materials with strong electronic correlations: Long-range Coulomb interactions and quantum dynamical screening”, funded by the European Research Council.
The task is not easy. The properties of materials are the result of a subtle arrangement of their components, nuclei and electrons, and to understand their magnetic, optical, or conductive properties requires solving the equations of quantum mechanics that govern the behavior of their constituents at the microscopic scale.
Electrons are by no means individualistic, but interact strongly with each other. So we are dealing with what physicists call a huge “quantum many-body problem.”
To make matters worse, the most interesting compounds are those most impacted by the complex collective effects induced by the electrostatic repulsion between electrons.
These materials are said to be “correlated” because the behavior of an electron is dependent on that of all others. More specifically, theorists at the Center for Theoretical Physics have recently taken an interest in the retarded character of the repulsion between electrons.
Indeed, when a charge is added to a solid, the electrons rearrange in order to find a more stable position. This rearrangement reduces the repulsion between two charges, but it is not instantaneous. Thus, if an electron moves within the solid, it will be followed by a sort of charge cloud that reduces its effective charge, but at the same time makes it “heavier.”
Predicting and understanding the properties of correlated materials with precision requires taking into account this “dynamical screening”.
This can be done using methods developed very recently at the Center for Theoretical Physics. Theorists at the Center have performed numerical simulations on a dedicated computer cluster and found, much to their great delight, that their results matched up with measured experimental data!
These findings are a very promising step on the path to “first principles” calculations.
On the horizon is the prospect of one day achieving computer alchemy, even in the most difficult case; that of correlated materials.
Digital data has become ubiquitous in scientific, industrial, and personal
spheres. Massive amounts of data of all types—text, image, sound,
video, internet browsing history, sensors, etc.—are part and parcel of
New challenges driven by development
This data requires increasingly complex automated processing, beyond indexing or simple statistical calculations, in order to recognize objects in an image, for example, or translate a text from one language to another, or predict which product to propose to customers.
For all these socalled artificial intelligence tasks, which seemed too difficult for a machine only a few years ago, current performances allow for widespread use. Recent advances are due to significant increases in computing capacity and in the volume of data, along with the development of new algorithmic architectures in machine learning that capitalize on these.
Media hype aside, the development of artificial intelligence raises many questions and new scientific issues. What are its limits? What interactions will there be with other scientific disciplines? What are its future industrial and social applications? What impact will it have on employment and, more broadly, on people’s lives?
The French advantage
France will have its full place in tackling these challenges if it leverages its chances.
At the scientific level, the education of French engineers is world renowned and French research scientists are present at the top in the various AI-related scientific communities.
However, the brain drain to industrial sectors and academia abroad, mainly due to the comparatively low salaries proposed to young researchers in France, poses a serious threat to maintaining this excellence.
At the industrial level, there are few large French technology companies of the FAAMG* type, but many start-ups and major world industrial champions in all sectors of activity.
Harvesting the fruits of artificial intelligence and thereby avoiding absorption by these technological giants can only be achieved by engaging in a strategic reflection on the importance of data in business, and not simply by elaborating a technical approach to software solutions or equipment purchase.
In particular, given the speed of technology transfer between university laboratories and industry, it will require a massive recruitment of experts in artificial intelligence. And, for this, doctoral degrees and competitive remunerations are essential. This is something the FAAMGs* have understood perfectly well.
"French research scientists are present at the top in the various AI-related scientific communities"
The sharing economy has built a social network of trust founded on
data: when users of a collaborative service enter their personal information
– name, age, address, photo, etc. – they are entering into a relationship
of transparency, one that invites strangers to trust each other.
The more information they provide, the greater the trust and the more they will transact.
This is the foundation which ensures open access to a multitude of services: ridesharing, equipment borrowing, peer-to-peer financing, house or skill swapping, etc. Trust is remarkably high within collaborative ecosystems: according to a poll conducted by BlaBlaCar among its members, the level of trust between users matches that of family members and is far higher than the level between colleagues. Such is the power of data!
This then triggers a second virtuous behaviour: users of collaborative services, once they’ve grown accustomed to trusting strangers, will gradually extend this instinct to other sharing activities.
It is easier to make one’s home available to a stranger when one has already had experience with the sharing economy, through carpooling, for example.
Digital stengthens social cohesion
This flies in the face of the mainstream notion that digital technology damages social cohesion. To the contrary: in the collaborative economy, it strengthens social bonds previously weakened by technological advancements. This is because it puts interactions between people back at the center of our everyday activities by making services conditional on social encounters.
It redefines the way we are connected to each other and reproduces, in larger communities, the circle of trust that we previously shared on the scale of a village or association.
The two major issues of data
But trust needs a framework to exist. Data privacy scandals involving major digital technology companies have generated growing distrust among users. At a time when data has become the lifeblood of businesses in the internet era, two major issues have arisen: on the one hand, companies that collect data must be transparent about its security and use, but on the other hand, users must have a clear and immediate understanding of the added value that comes with the transmission of their data.
These two conditions are imperative for companies to link data and trust. If they are not met, there is a risk that access to consumer data will be curtailed. Soon the issue of responsible data use by companies may become a question of fundamental human rights in the digital era for increasingly knowledgeable and recalcitrant users.
The fundamental question of online technology remains. For as long as we don’t have large European technology platforms to write our own rules governing data use and security, we will not be in charge of the game –and this will be to the detriment of individuals and companies alike.
"Soon the issue of responsible data use by companies may become a question of fundamental human rights"
In the time it takes to read this entry, 20,000 objects will have been
connected to the internet around the world. That translates into 30 million
It is a massive transformation that is underway. The Internet of Things began gaining momentum more than a decade ago, under the “Machine to Machine” concept.
After the internet of people, this second wave is picking up speed, with more than 50 billion objects expected to be connected by the end of 2020, which amounts to ten times more objects than people!
The scope of application is immense. In buildings, for example, hundreds of objects and sensors are now connected, with digital technologies being used to view, correlate, and manage.
Dynamic and intelligent control of buildings radically changes energy use, putting the user at the heart of the system. At the same time, the development of new digital applications around digital architectural models facilitates the coordination of all professionals, from design to maintenance, and revolutionizes the entire industry.
The digital transformation of objects will have structuring implications for the world of tomorrow, be they economic or social, some of which are not even conceivable today! We would have to imagine the intersection of every type of application, existing or to be developed, by micro- market segment to begin to glimpse the immensity of this systemic revolution. Without forgetting the implications for the security of the people and goods that are connected to it.
Take the example of the 4th industrial revolution: after the automation of production, it is now the exploitation that is being transformed thanks to the analysis of data collected on the machines, as well as the traceability of products and materials through the different production chains.
Marked by artificial intelligence, the industrial world of tomorrow –or should I say of today– will accelerate the revolution of supply chains, personalized production, and improvements in the safety of people.
Machines on production lines are now connected, responsive to their environment, capable of adapting, flexible and communicating. This is paving the way for an industry that respects the environment and people.
To be sure, the Internet of Things is a combination of multiple technologies, but it also opens up possibilities that we didn’t imagine along with new responsibilities.
It changes our economy and our way of perceiving the world. Our way of working will move towards greater transversality, more collaboration, with platforms, start-ups, and actors from all horizons.
It changes the way we make decisions along with management methods, which become more responsive and flexible. And it is the people behind the technology who must grasp the opportunity of the Internet of Things to accelerate the revolutions it allows, be it in terms of energy transition or simply to create the working methods of the future.
"The digital transformation of objects will have structuring implications for the world of tomorrow"