Three PhD Thesis Awards

At the first PhD graduation ceremony held online on June 17th, three thesis prizes were awarded to Clémence Tricaud, Ambre Bouillant and Boris Muzellec for their doctoral work at the Institut Polytechnique de Paris. Mentions were also awarded to Jaouad Mourtada, Guillaume Graciani, Vincent Corlay and Emile Parolin. This is the first edition of this PhD Thesis Award. For Benoît Deveaud, who heads the research committee of the Institut Polytechnique de Paris, "the work rewarded is of high quality. The winners have shown great originality in their research and have obtained outstanding results in their respective fields.”

Clémence Tricaud. Political economy and public economy

How can we better understand the behaviour of voters and elected officials? This is the question at the heart of Clémence Tricaud's PhD completed at the Ecole Polytechnique under the supervision of Pierre Boyer and Bruno Crépon. Based on a large-scale collection of administrative data and the use of quantitative methodologies, Clémence Tricaud examined, for example, the question of expressive voting, i.e. the fact of choosing one's ballot according to one's preferences, as opposed to a strategic or 'useful' vote. By comparing the results of two and three candidates in the second round of the French legislative and departmental elections - elections in which there may be trilateral contests - her work shows that a significant fraction of voters vote expressively and that this has a considerable impact on the outcome of elections. In particular, the presence of a third competitor in the second round leads to the loss of the candidate who, among the first two, was ideologically closest to the third, in almost 20% of cases. Clémence Tricaud also studied the effects of past election results on present voting choices. The results show that the ranking of candidates in the first round has a strong impact on their chances of victory in the second round, with voters having a preference to vote for the person who came first. Finally, through the analysis of public policies, another part of Clémence Tricaud's thesis provides explanations for the reluctance of mayors to integrate intermunicipal cooperation. The opposition of municipalities is attributable to the local consequences of this integration rather than to ideological reasons: an increase in construction in the case of urban municipalities and the loss of public services for rural ones. Clémence Tricaud's work has resulted in an article published in the international journal Econometrica, which is recognised as one of the top five economics journals in the world. Her work on voting has important implications for the choice of electoral rules, while her work on intermunicipal cooperation could usefully advise the implementation of new public policies. Two notes for the Institute for Public Policy have been published by the researcher, who has since been appointed assistant professor at the University of California at Los Angeles (UCLA).

Boris Muzellec, optimal transport for data sciences

In 1781, in his “Mémoire sur la théorie des déblais et des remblais” (treatise on the theory of cut and fill) the french mathematician Gaspard Monge asked himself how to move earth from one place to another in the fastest-or least tiring-way possible for the men engaged in the task. He thus posed the difficult problem of "optimal transport". The mathematical tools of optimal transport were later developed by mathematicians in the 20th century and applied, among other things, in logistics and economics. In his thesis, carried out at ENSAE under the supervision of Marco Cuturi, Boris Muzellec perfected these tools for data sciences, and more particularly for machine learning. The latter requires the ability to compare two probability distributions - the equivalent of cuttings and fillings - that describe data. This involves, for example, comparing the distribution of observed data, such as images, with the distribution of a model that generates similar images. However, optimal transport has its limitations, as the computation is particularly difficult for high-dimensional continuous distributions. Boris Muzellec's work helps to overcome these difficulties, in particular by using the so-called Bures-Wasserstein geometry to regularise the optimal transport problem or to regularise the data. His work has been presented at the most renowned conferences in the field, such as NeurIPS and ICML. He is now pursuing a post-doctorate at Sierra, a research group based in the Laboratoire d'Informatique de l'École Normale Superiéure.

Ambre Bouillant. A better understanding of Leidenfrost effect

In 1756, the German doctor Johann Gottlob Leidenfrost discovered a surprising effect: a drop of water placed on a support heated to more than 200°C do not come into contact with it. It levitates on a small cushion formed by its own steam, which gives it outstanding mobility. Although almost 300 years later we can all observe this effect on our stoves, some aspects of the phenomenon are still poorly understood. During her PhD under the supervision of David Quéré and Christophe Clanet at the Hydrodynamics Laboratory, Ambre Bouillant explored multiple facets of the Leidenfrost effect. The major contribution of her work concerns the dynamic properties of these Leidenfrost drops, whose unusual mobility is not simply explained by the absence of contact permitted by the vapour cushion. By spreading tracers within the drops, we can see that very intense internal flows are set up: when this drop is rounded off by evaporation, these flows form a single roll which turns as a wheel would do. In order for a wheel to move forward, it must take advantage of the friction with the ground. However, Leidenfrost drops are frictionless because they levitate. Ambre Bouillant's work has shown that the internal flows tilt the base of the drops. Everything happens as if the drops were sliding down an endless slope, which it creates spontaneously by tilting its base. This work has been awarded two video prizes and published in Nature Physics. Other contributions from Ambre Bouillant's thesis, notably on the star-like pulsations of Leidenfrost drops, are still in the process of being published, and she has meanwhile joined the University of Twente (Netherlands) as a post-doctoral fellow.

About the PhD Thesis Award

The Institut Polytechnique de Paris will award annually three prizes to the students who have produced the best PhD thesis and made the most outstanding contributions in one of the following fields: Physics; Biology and Chemistry; Information, Communications, Electronics; Computer Science, Data, and Artificial Intelligence; Engineering, Mechanics, and Energetics; Economics, Management, and Social Sciences; and Mathematics.