Publications

2021

• Unbalanced Optimal Transport through Non-negative Penalized Linear Regression
  
  • Chapel Laetitia
  • Flamary Rémi
  • Wu Haoran
  • Févotte Cédric
  • Gasso Gilles
  , 2021, 34. This paper addresses the problem of Unbalanced Optimal Transport (UOT) in which the marginal conditions are relaxed (using weighted penalties in lieu of equality) and no additional regularization is enforced on the OT plan. In this context, we show that the corresponding optimization problem can be reformulated as a non-negative penalized linear regression problem. This reformulation allows us to propose novel algorithms inspired from inverse problems and nonnegative matrix factorization. In particular, we consider majorization-minimization which leads in our setting to efficient multiplicative updates for a variety of penalties. Furthermore, we derive for the first time an efficient algorithm to compute the regularization path of UOT with quadratic penalties. The proposed algorithm provides a continuity of piece-wise linear OT plans converging to the solution of balanced OT (corresponding to infinite penalty weights). We perform several numerical experiments on simulated and real data illustrating the new algorithms, and provide a detailed discussion about more sophisticated optimization tools that can further be used to solve OT problems thanks to our reformulation. (10.48550/arXiv.2106.04145)
  DOI : 10.48550/arXiv.2106.04145
• Transcriptional regulation of photoprotection in dark-to-light transition - more than just a matter of excess light energy
  
  • Petroutsos Dimitris
  • Redekop Petra
  • Sanz-Luque Emanuel
  • Yuan Yizhong
  • Villain Gaelle
  • Grossman Arthur
  , 2021. In nature, photosynthetic organisms are exposed to different light spectra and intensities depending on the time of day and atmospheric and environmental conditions. When photosynthetic cells absorb excess light, they induce non-photochemical quenching to avoid photo-damage and trigger expression of ‘photoprotective’ genes. In this work, we used the green alga $Chlamydomonas\ reinhardtii$ to assess the impact of light intensity, light quality, wavelength, photosynthetic electron transport and CO 2 on induction of the ‘photoprotective’ genes ( LHCSR1 , LHCSR3 and PSBS ) during dark-to-light transitions. Induction (mRNA accumulation) occurred at very low light intensity, was independently modulated by blue and UV-B radiation through specific photoreceptors, and only LHCSR3 was strongly controlled by CO$_2$ levels through a putative enhancer function of CIA5, a transcription factor that controls genes of the carbon concentrating mechanism. We propose a model that integrates inputs of independent signaling pathways and how they may help the cells anticipate diel conditions and survive in a dynamic light environment. (10.1101/2021.10.23.463292)
  DOI : 10.1101/2021.10.23.463292
• A drug repurposing screen identifies altiratinib as a selective inhibitor of a key regulatory splicing kinase and a potential therapeutic for toxoplasmosis and malaria
  
  • Swale Christopher
  • Bellini Valeria
  • Bowler Matthew
  • Nardella Flore
  • Brenier-Pinchart Marie-Pierre
  • Cannella Dominique
  • Belmudes Lucid
  • Mas Caroline
  • Couté Yohann
  • Laurent Fabrice
  • Scherf Artur
  • Bougdour Alexandre
  • Hakimi Mohamed-Ali
  , 2021. The apicomplexa comprise a large phylum of single-celled, obligate intracellular protozoa that infect humans and animals and cause severe parasitic diseases. Available therapeutics against these devastating diseases are limited by suboptimal efficacy and frequent side effects, as well as the emergence and spread of resistance. Here, we use a drug repurposing strategy and identify altiratinib, a compound originally developed to treat glioblastoma, as a promising drug candidate with broad spectrum activity against apicomplexans. Altiratinib is parasiticidal and blocks the development of intracellular zoites in the nanomolar range and with a high selectivity index. We have identified TgPRP4K of T. gondii as the primary target of altiratinib by genetic target deconvolution, highlighting key residues within the kinase catalytic site that, when mutated, confer resistance to the drug. We have further elucidated the molecular basis of the inhibitory mechanism and species selectivity of altiratinib for TgPRP4K as well as for its P. falciparum counterpart PfCLK3. Our data also point to structural features critical for binding of the other PfCLK3 inhibitor, TCMDC-135051. Consistent with the role of this kinase family in splicing in a broad spectrum of eukaryotes, we have shown that altiratinib causes global disruption of splicing, primarily through intron retention in both T. gondii and P. falciparum. Thus, our data establish parasitic PRP4K/CLK3 as a promising pan-apicomplexan target whose repertoire of inhibitors can be expanded by the addition of altiratinib. (10.1101/2021.11.03.467097)
  DOI : 10.1101/2021.11.03.467097
• Estimation of the Uncertainty due to Each Step of Simulating the Photovoltaic Conversion under Real Operating Conditions
  
  • Migan-Dubois Anne
  • Badosa Jordi
  • Bourdin Vincent
  • Torres Aguilar Moira
  • Bonnassieux Yvan
  International Journal of Photoenergy, Hindawi Publishing Corporation, 2021, 2021, pp.4228658. The simulation of photovoltaic installations is a major issue for their sizing, their smart grid operation, and their fault detection and diagnosis. In this article, we study in detail every step of the simulation chain, either from the global horizontal irradiance and the ambient temperature (i.e., 4 steps of simulation) or considering the global in-plane irradiance and the module operating temperature (i.e., 1 step of simulation). The average quality estimation of the models is made through the calculations of average annual error between estimations and measurements, from 2016 to 2020. We have shown that the most uncertain step is the conversion of the global irradiance in its diffuse and direct components (17.2%, 2 models tested). If the model goes up to the in-plane irradiance, the average annual error decreases to 5.3% (6 models tested). The photovoltaic module temperature calculation induces an error of less than 2°C (4 models tested with 2 configurations). Meanwhile, the photoelectrical conversion shows a 3.5% error, similar to the measurement uncertainties, considering as input, the modules temperature, and the in-plane irradiance. If the simulation goes from the global irradiance and the ambient temperature measured locally, the estimation leads to a 6.7% average annual error. If the local measurements are not available, we can use the closest meteorological station’s records (13 for our study), and the error becomes 12.1%. Finally, we can also use the satellite images that lead to a 15.2% error, for average per year. The impact of available input shows that modeling the DC photovoltaic production, using global horizontal irradiance and ambient temperature, gives rise to an error of 6.6% for local measurements, 12.1% for weather station measurements, and 15.2% for satellite images estimations. This article thus draws up a review of the existing models, allowing to calculate the DC production of a photovoltaic module, depending on the atmospheric conditions, and highlights the most precise or most critical steps, considering in situ and weather station ground-based measurements, and also estimation from satellite images. (10.1155/2021/4228658)
  DOI : 10.1155/2021/4228658
• Embedding ML algorithms onto LPWAN sensors for compressed communications
  
  • Bernard Antoine
  • Dridi Aicha
  • Marot Michel
  • Afifi Hossam
  • Balakrichenan Sandoche
  , 2021, pp.1539-1545. LPWANs are networks characterized by the scarcity of their radio resources and their limited payload size. To extend the efficiency of the data transmission by decreasing the traffic sent from sensors, this paper proposes a lossy compression method using known ML techniques. We embedded a pre-trained neural network directly on constrained LoRaWAN devices and we tested the trade-off between compression ratio and accuracy of the compression algorithm. This paper studies multiple aspects of the system-energy consumption, error rate due to the lossy compression, compression ratio and the impact of LSTM parameter quantization-to measure the possible strengths and weaknesses of using a dual prediction system in order to reduce transmission costs. Surprisingly, machine learning used in this context does not consume a lot of energy and it even leads to energy saving in the very constrained devices which are the sensors. (10.1109/PIMRC50174.2021.9569714)
  DOI : 10.1109/PIMRC50174.2021.9569714
• Electrochemical Active Surface Area Determination of Iridium‐Based Mixed Oxides by Mercury Underpotential Deposition
  
  • Duran Silvia
  • Elmaalouf Marine
  • Odziomek Mateusz
  • Piquemal Jean‐yves
  • Faustini Marco
  • Giraud Marion
  • Peron Jennifer
  • Tard Cédric
  ChemElectroChem, Weinheim : Wiley-VCH, 2021, 8 (18), pp.3519-3524. The electrochemical surface area (ECSA) is a critical property to describe, analyze and compare electrocatalysts. The determination of the mass activity of a given catalyst is associated with this parameter which can thus lead to materials benchmarking. Reliable and robust methods to measure ECSA are needed, and those have to accommodate different structures, morphologies and compositions. In this study, we investigate mercury underpotential deposition (HgUPD) as a way to estimate ECSA for ultraporous electrocatalysts based on iridium and iridium-molybdenum electrocatalysts for the oxygen evolution reaction. Results reveal a clear agreement between physisorption measurements and HgUPD with excellent reproducibility. The method shows also that pre- and post-catalysis surface area measurements are not affected by the catalytic process on short timescale, opening the possibility of electrocalyst stability and degradation monitoring. (10.1002/celc.202100649)
  DOI : 10.1002/celc.202100649
• Mixing nanostructured Ni/piezoPVDF composite thin films with e-beam irradiation: A beneficial synergy to piezoelectric response
  
  • Potrzebowska Natalia
  • Cavani Olivier
  • Oral Ozlem
  • Doaré Olivier
  • Melilli Giuseppe
  • Wegrowe Jean-Eric
  • Clochard Marie-Claude.
  Materials Today Communications, Elsevier, 2021, 28, pp.102528. (10.1016/j.mtcomm.2021.102528)
  DOI : 10.1016/j.mtcomm.2021.102528
• Unbalanced minibatch Optimal Transport; applications to Domain Adaptation
  
  • Fatras Kilian
  • Séjourné Thibault
  • Courty Nicolas
  • Flamary Rémi
  , 2021. Optimal transport distances have found many applications in machine learning for their capacity to compare non-parametric probability distributions. Yet their algorithmic complexity generally prevents their direct use on large scale datasets. Among the possible strategies to alleviate this issue, practitioners can rely on computing estimates of these distances over subsets of data, {\em i.e.} minibatches. While computationally appealing, we highlight in this paper some limits of this strategy, arguing it can lead to undesirable smoothing effects. As an alternative, we suggest that the same minibatch strategy coupled with unbalanced optimal transport can yield more robust behavior. We discuss the associated theoretical properties, such as unbiased estimators, existence of gradients and concentration bounds. Our experimental study shows that in challenging problems associated to domain adaptation, the use of unbalanced optimal transport leads to significantly better results, competing with or surpassing recent baselines. (10.48550/arXiv.2103.03606)
  DOI : 10.48550/arXiv.2103.03606
• Integration of climate variability and climate change in renewable energy planning
  
  • Drobinski Philippe
  • Tantet Alexis
  Physics-Uspekhi, Turpion, 2021. The trajectory outlined in the Paris Agreement to keep global warming below 2°C dictates not only the timing but also the speed at which the transformation of our energy system must take place to decarbonize energy production. Complying with the Paris Agreement requires reducing the carbon content of energy by about 75% and therefore making a rapid transition from fossil production to production based on low-carbon technologies. Among these technologies are those based on renewable energies. The variability of the climate itself induces a fluctuating or even an intermittent production of variable renewable energy (solar, wind, marine), challenging the balance of the electricity grid. In this context, to speak of energy transition is to face the problem of increasing the penetration of low-carbon energy production while limiting the variability while ensuring socio-technical feasibility and economic viability. The problem is not simple and the delicate balance between urgency (drastic reduction in emissions) and utopia (what strategy for low carbon energies, opportunities and obstacles) needs to be clearly defined. (10.3367/UFNe.2021.07.039080)
  DOI : 10.3367/UFNe.2021.07.039080
• Chiral Radical Cation Salts of Me-EDT-TTF and DM-EDT-TTF with Octahedral, Linear and Tetrahedral Monoanions
  
  • Mroweh Nabil
  • Bogdan Alexandra
  • Pop Flavia
  • Auban-Senzier Pascale
  • Vanthuyne Nicolas
  • Lopes Elsa B
  • Almeida Manuel
  • Avarvari Narcis
  Magnetochemistry, MDPI, 2021, 7 (6), pp.87. Methyl-ethylenedithio-tetrathiafulvalene (Me-EDT-TTF (1) and dimethyl-ethylenedithio-tetrathiafulvalene (DM-EDT-TTF (2) are valuable precursors for chiral molecular conductors, which are generally obtained by electrocrystallization in the presence of various counter-ions. The number of the stereogenic centers, their relative location on the molecule, the nature of the counter-ion and the electrocrystallization conditions play a paramount role in the crystal structures and conducting properties of the resulting materials. Here, we report the preparation and detailed structural characterization of the following series of radical cation salts: (i) mixed valence (1)2AsF6 as racemic, and (S) and (R) enantiomers; (ii) [(S)-1]AsF6·C4H8O and [(R)-1]AsF6·C4H8O where a strong dimerization of the donors is observed; (iii) (1)I3 and (2)I3 as racemic and enantiopure forms and (iv) [(meso)-2]PF6 and [(meso)-2]XO4 (X = Cl, Re), based on the new donor (meso)-2. In the latter, the two methyl substituents necessarily adopt axial and equatorial conformations, thus leading to a completely different packing of the donors when compared to the chiral form (S,S)/(R,R) of 2 in its radical cation salts. Single crystal resistivity measurements, complemented by thermoelectric power measurements in the case of (1)2AsF6, suggest quasi-metallic conductivity for the latter in the high temperature regime, with σRT ≈ 1–10 S cm–1, while semiconducting behavior is observed for the (meso)-2 based salts. (10.3390/magnetochemistry7060087)
  DOI : 10.3390/magnetochemistry7060087
• The Tuning Strategy of IPSL‐CM6A‐LR
  
  • Mignot Juliette
  • Hourdin Frédéric
  • Deshayes Julie
  • Boucher Olivier
  • Gastineau Guillaume
  • Musat Ionela
  • Vancoppenolle Martin
  • Servonnat Jérôme
  • Caubel Arnaud
  • Cheruy Frédérique
  • Denvil Sébastien
  • Dufresne Jean-Louis
  • Éthé Christian
  • Fairhead Laurent
  • Foujols Marie-Alice
  • Grandpeix Jean-Yves
  • Levavasseur Guillaume
  • Marti Olivier
  • Menary Matthew B
  • Rio Catherine
  • rousset clement
  • Silvy Yona
  Journal of Advances in Modeling Earth Systems, American Geophysical Union, 2021, 13 (5), pp.e2020MS002340. Climate change is a serious issue for humanity with important ramifications for policy and decision making. Robust and cost-efficient policies on mitigation and adaptation require assessments of current and future risks for natural and human systems under a range of socioeconomic scenarios. Those assessments rely on numerical simulations performed with state-of-the-art climate models. Simulations are coordinated at an international level within the Coupled Model Intercomparison Project (CMIP) which provides the bedrock for a substantial part of the publications synthesized in the Intergovernmental Panel on Climate Change (IPCC) reports. Such projects are fundamental in order to document the robust features as well as the relatively large uncertainties in the future climate projections. Among others, these uncertainties come from the various assumptions made by the ∼30 teams that develop CMIP-class models. In particular, because of (10.1029/2020ms002340)
  DOI : 10.1029/2020ms002340
• Charge-Transfer Chemical Reactions in Nanofluidic Fabry-Pérot Cavities
  
  • Mauro L
  • Caicedo K
  • Jonusauskas G
  • Avriller Rémi
  Physical Review B: Condensed Matter and Materials Physics (1998-2015), American Physical Society, 2021. We investigate the chemical reactivity of molecular populations confined inside a nanofluidic Fabry-Pérot cavity. Due to strong light-matter interactions developing between a resonant electromagnetic cavity-mode and the electric dipole moment of the confined molecules, a polariton is formed. The former gets dressed by environmental vibrational and rotational degrees of freedom of the solvent. We call the resulting polariton dressed by its cloud of environmental excitation a "reacton", since it further undergoes chemical reactions. We characterize how the reacton formation modifies the kinetics of a photoisomerization chemical reaction involving an elementary charge-transfer process. We show that the reaction driving-force and reorganization energy are both modulated optically by the reactant concentration, the vacuum Rabi splitting and the de-tuning between the Fabry-Pérot cavity frequency and targeted electronic transition. Finally, we compute the ultrafast picosecond dynamics of the whole photochemical reaction. We predict that despite optical cavity losses and solvent-mediated non-radiative relaxation, measurable signatures of the reacton formation can be found in state-of-the-art pump-probe experiments. (10.1103/PhysRevB.103.165412)
  DOI : 10.1103/PhysRevB.103.165412
• POT : Python Optimal Transport
  
  • Flamary Rémi
  • Courty Nicolas
  • Gramfort Alexandre
  • Alaya Mokhtar Zahdi
  • Boisbunon Aurélie
  • Chambon Stanislas
  • Chapel Laetitia
  • Corenflos Adrien
  • Fatras Kilian
  • Fournier Nemo
  • Gautheron Léo
  • Gayraud Nathalie T H
  • Janati Hicham
  • Rakotomamonjy Alain
  • Redko Ievgen
  • Rolet Antoine
  • Schutz Antony
  • Seguy Vivien
  • Sutherland Danica J
  • Tavenard Romain
  • Tong Alexander
  • Vayer Titouan
  Journal of Machine Learning Research, Microtome Publishing, 2021. Optimal transport has recently been reintroduced to the machine learning community thanks in part to novel efficient optimization procedures allowing for medium to large scale applications. We propose a Python toolbox that implements several key optimal transport ideas for the machine learning community. The toolbox contains implementations of a number of founding works of OT for machine learning such as Sinkhorn algorithm and Wasserstein barycenters, but also provides generic solvers that can be used for conducting novel fundamental research. This toolbox, named POT for Python Optimal Transport, is open source with an MIT license.
• Utility-Scale PV-Battery versus CSP-Thermal Storage in Morocco: Storage and Cost Effect under Penetration Scenarios
  
  • Bouramdane Ayat-Allah
  • Tantet Alexis
  • Drobinski Philippe
  Energies, MDPI, 2021, 14 (15), pp.4675. In this study, we examine how Battery Storage (BES) and Thermal Storage (TES) combined with solar Photovoltaic (PV) and Concentrated Solar Power (CSP) technologies with an increased storage duration and rental cost together with diversification would influence the Moroccan mix and to what extent the variability (i.e., adequacy risk) can be reduced; this is done using recent (2013) cost data and under various penetration scenarios. To do this, we use MERRA-2 climate reanalysis to simulate hourly demand and capacity factors (CFs) of wind, solar PV and CSP without and with increasing storage capabilities—as defined by the CSP Solar Multiple (SM) and PV Inverter Loading Ratio (ILR). We adjust these time series to observations for the four Moroccan electrical zones over the year 2018. Our objective is to maximize the renewable (RE) penetration and minimize the imbalances between RE production and consumption considering three optimization strategies. We analyze mixes along Pareto fronts using the Mean-Variance Portfolio approach—implemented in the E4CLIM model—in which we add a maximum-cost constraint to take into account the different rental costs of wind, PV and CSP. We propose a method to calculate the rental cost of storage and production technologies taking into account the constraints on storage associated with the increase of SM and ILR in the added PV-BES and CSP-TES modules, keeping the mean solar CFs fixed. We perform some load bands-reduction diagnostics to assess the reliability benefits provided by each RE technology. We find that, at low penetrations, the maximum-cost budget is not reached because a small capacity is needed. The higher the ILR for PV, the larger the share of PV in the mix compared to wind and CSP without storage is removed completely. Between PV-BES and CSP-TES, the latter is preferred as it has larger storage capacity and thus stronger impact in reducing the adequacy risk. As additional BES are installed, more than TES, PV-BES is favored. At high penetrations, optimal mixes are impacted by cost, the more so as CSP (resp., PV) with high SM (resp., ILR) are installed. Wind is preferably installed due to its high mean CF compared to cost, followed by either PV-BES or CSP/CSP-TES. Scenarios without or with medium storage capacity favor CSP/CSP-TES, while high storage duration scenarios are dominated by low-cost PV-BES. However, scenarios ignoring the storage cost and constraints provide more weight to PV-BES whatever the penetration level. We also show that significant reduction of RE variability can only be achieved through geographical diversification. Technological complementarity may only help to reduce the variance when PV and CSP are both installed without or with a small amount of storage. However, the diversification effect is slightly smaller when the SM and ILR are increased and the covariances are reduced as well since mixes become less diversified. (10.3390/en14154675)
  DOI : 10.3390/en14154675
• A Minimal System Cost Minimization Model for Variable Renewable Energy Integration: Application to France and Comparison to Mean-Variance Analysis
  
  • Tantet Alexis
  • Drobinski Philippe
  Energies, MDPI, 2021, 14 (16), pp.5143. The viability of Variable Renewable Energy (VRE)-investment strategies depends on the response of dispatchable producers to satisfy the net load. We lack a simple research tool with sufficient complexity to represent major phenomena associated with the response of dispatchable producers to the integration of high shares of VRE and their impact on system costs. We develop a minimization of the system cost allowing one to quantify and decompose the system value of VRE depending on an aggregate dispatchable production. Defining the variable cost of the dispatchable generation as quadratic with a coefficient depending on macroeconomic factors such as the cost of greenhouse gas emissions leads to the simplest version of the model. In the absence of curtailment, and for particular parameter values, this version is equivalent to a mean-variance problem. We apply this model to France with solar and wind capacities distributed over the administrative regions of metropolitan France. In this case, ignoring the wholesale price effect and variability has a relatively small impact on optimal investments, but leads to largely underestimating the system total cost and overestimating the system marginal cost. (10.3390/en14165143)
  DOI : 10.3390/en14165143
• Impact of natural disasters on consumer behavior: case of the 2017 El Niño phenomenon in Peru
  
  • Alatrista-Salas Hugo
  • Gauthier Vincent
  • Nunez-Del-Prado Miguel
  • Becker Monique
  PLoS ONE, Public Library of Science, 2021, 16 (1), pp.e0244409:1-e0244409:23. El Niño is an extreme weather event featuring unusual warming of surface waters in the eastern equatorial Pacific Ocean. This phenomenon is characterized by heavy rains and floods that negatively affect the economic activities of the impacted areas. Understanding how this phenomenon influences consumption behavior at different granularity levels is essential for recommending strategies to normalize the situation. With this aim, we performed a multi-scale analysis of data associated with bank transactions involving credit and debit cards. Our findings can be summarized into two main results: Coarse-grained analysis reveals the presence of the El Niño phenomenon and the recovery time in a given territory, while fine-grained analysis demonstrates a change in individuals’ purchasing patterns and in merchant relevance as a consequence of the climatic event. The results also indicate that society successfully withstood the natural disaster owing to the economic structure built over time. In this study, we present a new method that may be useful for better characterizing future extreme events. (10.1371/journal.pone.0244409)
  DOI : 10.1371/journal.pone.0244409
• Deep Recurrent Learning versus Q-Learning for Energy Management Systems in Next Generation Network
  
  • Dridi Aicha
  • Boucetta Chérifa
  • Moungla Hassine
  • Afifi Hossam
  , 2021, pp.1-6. (10.1109/GLOBECOM46510.2021.9685620)
  DOI : 10.1109/GLOBECOM46510.2021.9685620