The goal of the 4DPlants project is to enable biologists to successfully analyse and predict a real plant's growth at the organ scale in 3D. To do so, we first plan to generate high-quality annotated 4D datasets of growing plants, both virtual and real, as sequences of 3D point clouds. 4DPlants will then develop compact representations that consider spatial alignment information over a sequence. Such annotated datasets and compact representations will be used to study spatiotemporal segmentation and growth prediction with the help of deep neural networks that learn the plant growth information. Tools for automated analysis by comparisons with ground truth data will be developed to speed up the validation process. All three partners have previous experience in either learning moving shapes and/or digitized plant processing and analysis.

This collaborative project focuses on derived algebraic geometry and its interactions with other fields, such as geometric representation theory, Donaldson-Thomas theory and singularities of schemes. Derived algebraic geometry is a rich and vibrant field that lies at the crossroad between algebraic geometry and homotopy theory. It was developed starting from 2000 thanks to the efforts of J. Lurie, B. Toën and G. Vezzosi, although one can date back several central ideas back to the work of Serre, Quillen and Illusie. Nowadays, derived geometry has become a widespread toolkit and found applications in a variety of subjects, ranging from symplectic geometry to p-adic Hodge theory. With this project we plan to push further the boundaries of derived geometry and to develop new applications to other subjects: we will study variations of cohomological Hall algebras, categorifications of Donaldson-Thomas invariants as well as derived foliations in positive characteristic and their applications to the theory of singularities of schemes.

The CosMoPoly project is aimed to design, synthesize and evaluate the biological properties of a portfolio of innovative biobased polyfunctional molecules dedicated to cosmetics and obtained from a building-blocks library available from agro-industries, biorefineries and chemical companies. The strategy that will be developed during the project is based on : (1) the use of powerful predictive tools (QSAR and artificial intelligence) to select the building-blocks to combine together to achieve polyfunctionality of the molecules, (2) the use of green chemistry and biocatalysis approaches to synthesize the polyfunctional molecules, (3) the evaluation of the targeted properties of the synthesized molecules.

Understanding the properties and evolution of interstellar dust is a key to our knowledge of galaxy evolution. A promising way to progress in this field is to model the dust properties of galaxies at different stages of their evolution (i.e. at different metallicities). Contemporary dust models are however constrained solely by the observations of the Milky Way (MW), specific to Solar metallicity systems. Their current application to lower-metallicity objects is thus biasing our knowledge of galaxy evolution. Fortunately, we now have, in the Large and Small Magellanic Clouds (LMC and SMC), the set of observations necessary to properly constrain the grain properties (abundance, composition, size distribution), similarly to what is done in the MW: the emission, extinction and elemental depletions of their diffuse interstellar medium. It is therefore timely to take this opportunity. We propose to hire two postdoctoral researchers at DAp and IRAP, to simultaneously model the dust properties in the MW, LMC and SMC, by inferring metallicity-dependent grain properties. We will take particular care in estimating the different sources of observational and experimental uncertainties and propagate them through the Bayesian inference of the dust parameters. We will consequently deliver the first metallicity-dependent, probabilistic, fully-constrained dust model. This model will allow the community to: (i) more reliably interpret observations of galaxies; (ii) benchmark cosmic dust evolution models; and (iii) improve the accuracy of numerical simulations of the star formation process and of galaxy evolution.

Solvophobic effect is a prevalent driving force for host-guest chemistry. Like H2O, perfluorocarbons (PFCs) are attractive solvents for supramolecular receptor confinement. A new poly-acridinium receptors will be confined in PFCs and in H2O and will be exploited to give rise to the selective extraction of polycyclic aromatic hydrocarbons (PAHs) from an organic source phase to a receiving phase across a transporting phase (PFC or H2O). In this fundamental proposal, light will used to drive the selective transportation of the PAH guests out-of-equilibrium to enrich the receiving phase with the desired solute. Finally, a systematic comparative study between PFCs and H2O will allow the estimate of the enthalpic and entropic contributions of both solvents in order to rationalize their solvophobic effect.