LTT-9779 b is an ultra-hot Neptune (Rp ~ 4.7 Re, Mp ~ 29 Me) orbiting its Sun-like host star in just 19 hours, placing it deep within the "hot Neptune desert," where Neptunian planets are seldom found. We present new JWST NIRSpec G395H phase-curve observations that probe its atmospheric composition in unprecedented detail. At near-infrared wavelengths, which penetrate the high-altitude clouds inferred from previous NIRISS/SOSS spectra, thermal emission reveals a carbon-rich atmosphere with opacity dominated by carbon monoxide (CO) and carbon dioxide (CO2). Both species are detected at all orbital phases, with retrieved mixing ratios of 10^-1 for CO and 10^-4 for CO2, indicating a globally well-mixed reservoir of carbon-bearing gases. We also moderately detect water vapor (H2O) and tentatively detect sulfur dioxide (SO2), providing insight into its chemistry and possible photochemical production under intense stellar irradiation. From these detections we infer a carbon-to-oxygen ratio near unity (C/O ~ 1) and a metallicity exceeding 500X Solar, consistent with equilibrium chemistry predictions for high-temperature atmospheres. This enrichment raises the mean molecular weight, reducing atmospheric escape, and likely helps LTT-9779 b retain a substantial atmosphere despite extreme irradiation. Our findings show that LTT-9779 b survives where few planets can, maintaining a carbon-rich atmosphere in a region where hot Neptune-class worlds are expected to evaporate. This makes LTT-9779 b a valuable laboratory for studying atmospheric escape and chemical processes under extreme conditions, offering new insight into the survival of planets in the hot Neptune desert.

We propose a method to represent bipartite networks using graph embeddings tailored to tackle the challenges of studying ecological networks, such as the ones linking plants and pollinators, where many covariates need to be accounted for, in particular to control for sampling bias. We adapt the variational graph auto-encoder approach to the bipartite case, which enables us to generate embeddings in a latent space where the two sets of nodes are positioned based on their probability of connection. We translate the fairness framework commonly considered in sociology in order to address sampling bias in ecology. By incorporating the Hilbert-Schmidt independence criterion (HSIC) as an additional penalty term in the loss we optimize, we ensure that the structure of the latent space is independent of continuous variables, which are related to the sampling process. Finally, we show how our approach can change our understanding of ecological networks when applied to the Spipoll data set, a citizen science monitoring program of plant-pollinator interactions to which many observers contribute, making it prone to sampling bias.

In biomedical research and artificial intelligence, access to large, well-balanced, and representative datasets is crucial for developing trustworthy applications that can be used in real-world scenarios. However, obtaining such datasets can be challenging, as they are often restricted to hospitals and specialized facilities. To address this issue, the study proposes to generate highly realistic synthetic faces exhibiting drug abuse traits through augmentation. The proposed method, called "3DG-GA", Deep De-identified anonymous Dataset Generation, uses Genetics Algorithm as a strategy for synthetic faces generation. The algorithm includes GAN artificial face generation, forgery detection, and face recognition. Initially, a dataset of 120 images of actual facial drug abuse is used. By preserving, the drug traits, the 3DG-GA provides a dataset containing 3000 synthetic facial drug abuse images. The dataset will be open to the scientific community, which can reproduce our results and benefit from the generated datasets while avoiding legal or ethical restrictions.

The Josephus problem is a well--studied elimination problem consisting in determining the position of the survivor after repeated applications of a deterministic rule removing one person at a time from a given group. A natural probabilistic variant of this process is introduced in this paper. More precisely, in this variant, the survivor is determined after performing a succession of Bernouilli trials with parameter $p$ designating each time the person to remove. When the number of participants tends to infinity, the main result characterises the limit distribution of the position of the survivor with an increasing degree of precision as the parameter approaches the unbiaised case $p=1/2$. Then, the convergence rate to the position of the survivor is obtained in the form of a Central-Limit Theorem. A number of other variants of the suggested probabilistic elimination process are also considered. They each admit a specific limit behavior which, in most cases, is stated in the form of an open problem.

Timed languages contain sequences of discrete events ("letters'') separated by real-valued delays, they can be recognized by timed automata, and represent behaviors of various real-time systems. The notion of bandwidth of a timed language defined in a previous paper characterizes the amount of information per time unit, encoded in words of the language observed with some precision {\epsilon}. In this paper, we identify three classes of timed automata according to the asymptotics of the bandwidth of their languages with respect to this precision {\epsilon}: automata are either meager, with an O(1) bandwidth, normal, with a {\Theta}(log (1/{\epsilon})) bandwidth, or obese, with {\Theta}(1/{\epsilon}) bandwidth. We define two structural criteria and prove that they partition timed automata into these three classes of bandwidth, implying that there are no intermediate asymptotic classes. The classification problem of a timed automaton is PSPACE-complete. Both criteria are formulated using morphisms from paths of the timed automaton to some finite monoids extending Puri's orbit graphs; the proofs are based on Simon's factorization forest theorem.

We extend the De Giorgi--Nash--Moser theory to a class of kinetic Fokker-Planck equations and deduce new results on the Landau-Coulomb equation. More precisely, we first study the H{ö}lder regularity and establish a Harnack inequality for solutions to a general linear equation of Fokker-Planck type whose coefficients are merely measurable and essentially bounded, i.e. assuming no regularity on the coefficients in order to later derive results for non-linear problems. This general equation has the formal structure of the hypoelliptic equations "of type II" , sometimes also called ultraparabolic equations of Kolmogorov type, but with rough coefficients: it combines a first-order skew-symmetric operator with a second-order elliptic operator involving derivatives along only part of the coordinates and with rough coefficients. These general results are then applied to the non-negative essentially bounded weak solutions of the Landau equation with inverse-power law $\gamma$ $\in$ [--d, 1] whose mass, energy and entropy density are bounded and mass is bounded away from 0, and we deduce the H{ö}lder regularity of these solutions.

We prove that certain nonlocal functionals defined on partitions made of measurable sets Gamma-converge to a local functional modeled on the perimeter in the sense of De Giorgi. Those nonlocal functionals involve generalized surface tension coefficients, and are lower semicontinuous even if the coefficients do not satisfy the triangular inequality. It implies a relaxation process in the limit, and provides a novel effect compare to the known gamma-convergence of the fractional perimeter towards the standard perimeter.

Knee osteoarthritis is a degenerative joint disease that induces chronic pain and disability. Bone morphological analysis is a promising tool to understand the mechanical aspect of this disorder. This study proposes a 2D bone morphological analysis using manually segmented bones to explore morphological features related to distinct pain conditions. Furthermore, six semantic segmentation algorithms are assessed for extracting femur and tibia bones from X-ray images. Our analysis reveals that the morphology of the femur undergoes significant changes in instances where pain worsens. Conversely, improvements in pain may not manifest pronounced alterations in bone shape. The few-shot-learning-based algorithm, UniverSeg, demonstrated superior segmentation results with Dice scores of 99.69% for femur and 99.60% for tibia. Regarding pain condition classification, the zero-shot-learning-based algorithm, CP-SAM, achieved the highest accuracy at 66% among all models. UniverSeg is recommended for automatic knee bone segmentation, while SAM models show potential with prompt encoder modifications for optimized outcomes. These findings highlight the effectiveness of few-shot learning for semantic segmentation and the potential of zero-shot learning in enhancing classification models for knee osteoarthritis diagnosis.

Buses and heavy vehicles have more blind spots compared to cars and other road vehicles due to their large sizes. Therefore, accidents caused by these heavy vehicles are more fatal and result in severe injuries to other road users. These possible blind-spot collisions can be identified early using vision-based object detection approaches. Yet, the existing state-of-the-art vision-based object detection models rely heavily on a single feature descriptor for making decisions. In this research, the design of two convolutional neural networks (CNNs) based on high-level feature descriptors and their integration with faster R-CNN is proposed to detect blind-spot collisions for heavy vehicles. Moreover, a fusion approach is proposed to integrate two pre-trained networks (i.e., Resnet 50 and Resnet 101) for extracting high level features for blind-spot vehicle detection. The fusion of features significantly improves the performance of faster R-CNN and outperformed the existing state-of-the-art methods. Both approaches are validated on a self-recorded blind-spot vehicle detection dataset for buses and an online LISA dataset for vehicle detection. For both proposed approaches, a false detection rate (FDR) of 3.05% and 3.49% are obtained for the self recorded dataset, making these approaches suitable for real time applications.

We complete our study of the three dimensional Ginzburg--Landau functional with magnetic field, in the asymptotic regime of a small inverse Ginzburg--Landau parameter $\varepsilon$, and near the first critical field $H_{c_1}$ for which the first vortex filaments appear in energy minimizers. Under a nondegeneracy condition, we show a next order asymptotic expansion of $H_{c_1}$ as $\varepsilon \to 0$, and exhibit a sequence of transitions, with vortex lines appearing one by one as the intensity of the applied magnetic field is increased: passing $H_{c_1}$ there is one vortex, then increasing $H_{c_1}$ by an increment of order $\log |\log\varepsilon|$ a second vortex line appears, etc. These vortex lines accumulate near a special curve $\Gamma_0$, solution to an isoflux problem. We derive a next order energy that the vortex lines must minimize in the asymptotic limit, after a suitable horizontal blow-up around $\Gamma_0$. This energy is the sum of terms where penalizations of the length of the lines, logarithmic repulsion between the lines and magnetic confinement near $\Gamma_0$ compete. This elucidates the shape of vortex lines in superconductors.

Given a prime $p$, the $p$-adic Littlewood Conjecture stands as a well-known arithmetic variant of the celebrated Littlewood Conjecture in Diophantine Approximation. In the same way as the latter, it admits a natural function field analogue depending on the choice of an irreducible polynomial $P(t)$ with coefficients in a field $\mathbb{K}$. This analogue is referred to as the $P(t)$-adic Littlewood Conjecture ($P(t)$-LC for short). $P(t)$-LC is proved to fail for any choice of irreducible polynomial $P(t)$ over any ground field $\mathbb{K}$ with characteristic $\ell \equiv 3\pmod{4}$. The counterexample refuting it is shown to present a local arithmetic obstruction emerging from the fact that -1 is not a quadratic residue modulo a prime $\ell\equiv 3\pmod{4}$. The theory developed elucidates and generalises all previous approaches towards refuting the conjecture. They were all based on the computer-assisted method initiated by Adiceam, Nesharim and Lunnon (2021) which has been able to establish that $P(t)$-LC fails in some small characteristics (essentially up to 11). This computer-assisted method is, however, unable to provide a general statement as it relies on ad hoc computer verifications which, provided they terminate, refute $P(t)$--LC in a given characteristic. This limitation is overcome by exhibiting an arithmetic obstruction to the validity of $P(t)$--LC in infinitely many this http URL leaves the remaining case of odd characteristics $\ell\equiv 1\pmod{4}$ dependent on their full determination.

Time series forecasting task predicts future trends based on historical information. Transformer-based U-Net architectures, despite their success in medical image segmentation, have limitations in both expressiveness and computation efficiency in time series forecasting as evidenced in YFormer. To tackle these challenges, we introduce Kernel-U-Net, a flexible and kernel-customizable U-shape neural network architecture. The kernel-U-Net encoder compresses the input series into latent vectors, and its symmetric decoder subsequently expands these vectors into output series. Specifically, Kernel-U-Net separates the procedure of partitioning input time series into patches from kernel manipulation, thereby providing the convenience of customized executing kernels. Our method offers two primary advantages: 1) Flexibility in kernel customization to adapt to specific datasets; and 2) Enhanced computational efficiency, with the complexity of the Transformer layer reduced to linear. Experiments on seven real-world datasets, demonstrate that Kernel-U-Net's performance either exceeds or meets that of the existing state-of-the-art model in the majority of cases in channel-independent settings. The source code for Kernel-U-Net will be made publicly available for further research and application.

LTT-9779 b is an ultra-hot Neptune (Rp ~ 4.7 Re, Mp ~ 29 Me) orbiting its Sun-like host star in just 19 hours, placing it deep within the "hot Neptune desert," where Neptunian planets are seldom found. We present new JWST NIRSpec G395H phase-curve observations that probe its atmospheric composition in unprecedented detail. At near-infrared wavelengths, which penetrate the high-altitude clouds inferred from previous NIRISS/SOSS spectra, thermal emission reveals a carbon-rich atmosphere with opacity dominated by carbon monoxide (CO) and carbon dioxide (CO2). Both species are detected at all orbital phases, with retrieved mixing ratios of 10^-1 for CO and 10^-4 for CO2, indicating a globally well-mixed reservoir of carbon-bearing gases. We also moderately detect water vapor (H2O) and tentatively detect sulfur dioxide (SO2), providing insight into its chemistry and possible photochemical production under intense stellar irradiation. From these detections we infer a carbon-to-oxygen ratio near unity (C/O ~ 1) and a metallicity exceeding 500X Solar, consistent with equilibrium chemistry predictions for high-temperature atmospheres. This enrichment raises the mean molecular weight, reducing atmospheric escape, and likely helps LTT-9779 b retain a substantial atmosphere despite extreme irradiation. Our findings show that LTT-9779 b survives where few planets can, maintaining a carbon-rich atmosphere in a region where hot Neptune-class worlds are expected to evaporate. This makes LTT-9779 b a valuable laboratory for studying atmospheric escape and chemical processes under extreme conditions, offering new insight into the survival of planets in the hot Neptune desert.

Forecasting multivariate time series is a computationally intensive task challenged by extreme or redundant samples. Recent resampling methods aim to increase training efficiency by reweighting samples based on their running losses. However, these methods do not solve the problems caused by heavy-tailed distribution losses, such as overfitting to outliers. To tackle these issues, we introduce a novel approach: a Gaussian loss-weighted sampler that multiplies their running losses with a Gaussian distribution weight. It reduces the probability of selecting samples with very low or very high losses while favoring those close to average losses. As it creates a weighted loss distribution that is not heavy-tailed theoretically, there are several advantages to highlight compared to existing methods: 1) it relieves the inefficiency in learning redundant easy samples and overfitting to outliers, 2) It improves training efficiency by preferentially learning samples close to the average loss. Application on real-world time series forecasting datasets demonstrate improvements in prediction quality for 1%-4% using mean square error measurements in channel-independent settings. The code will be available online after 1 the review.

The performance of Human Activity Recognition (HAR) models, particularly deep neural networks, is highly contingent upon the availability of the massive amount of annotated training data which should be sufficiently labeled. Though, data acquisition and manual annotation in the HAR domain are prohibitively expensive due to skilled human resource requirements in both steps. Hence, domain adaptation techniques have been proposed to adapt the knowledge from the existing source of data. More recently, adversarial transfer learning methods have shown very promising results in image classification, yet limited for sensor-based HAR problems, which are still prone to the unfavorable effects of the imbalanced distribution of samples. This paper presents a novel generic and robust approach for semi-supervised domain adaptation in HAR, which capitalizes on the advantages of the adversarial framework to tackle the shortcomings, by leveraging knowledge from annotated samples exclusively from the source subject and unlabeled ones of the target subject. Extensive subject translation experiments are conducted on three large, middle, and small-size datasets with different levels of imbalance to assess the robustness and effectiveness of the proposed model to the scale as well as imbalance in the data. The results demonstrate the effectiveness of our proposed algorithms over state-of-the-art methods, which led in up to 13%, 4%, and 13% improvement of our high-level activities recognition metrics for Opportunity, LISSI, and PAMAP2 datasets, respectively. The LISSI dataset is the most challenging one owing to its less populated and imbalanced distribution. Compared to the SA-GAN adversarial domain adaptation method, the proposed approach enhances the final classification performance with an average of 7.5% for the three datasets, which emphasizes the effectiveness of micro-mini-batch training.

Activated Random Walks, on $\mathbb{Z}^d$ for any $d\geqslant 1$, is an interacting particle system, where particles can be in either of two states: active or frozen. Each active particle performs a continuous-time simple random walk during an exponential time of parameter $\lambda$, after which it stays still in the frozen state, until another active particle shares its location, and turns it instantaneously back into activity. This model is known to have a phase transition, and we show that the critical density, controlling the phase transition, is less than one in any dimension and for any value of the sleep rate $\lambda$. We provide upper bounds for the critical density in both the small $\lambda$ and large $\lambda$ regimes.

Accurate modeling of human mobility is critical for understanding epidemic spread and deploying timely interventions. In this work, we leverage a large-scale spatio-temporal dataset collected from Peru's national Digital Contact Tracing (DCT) application during the COVID-19 pandemic to forecast mobility flows across urban regions. A key challenge lies in the spatial sparsity of hourly mobility counts across hexagonal grid cells, which limits the predictive power of conventional time series models. To address this, we propose a lightweight and model-agnostic Spatial Neighbourhood Fusion (SPN) technique that augments each cell's features with aggregated signals from its immediate H3 neighbors. We evaluate this strategy on three forecasting backbones: NLinear, PatchTST, and K-U-Net, under various historical input lengths. Experimental results show that SPN consistently improves forecasting performance, achieving up to 9.85 percent reduction in test MSE. Our findings demonstrate that spatial smoothing of sparse mobility signals provides a simple yet effective path toward robust spatio-temporal forecasting during public health crises.

Crossing multiple planetary boundaries places us in a zone of uncertainty that is characterized by considerable fluctuations in climatic events. The situation is exacerbated by the relentless use of resources and energy required to develop digital infrastructures that have become pervasive and ubiquitous. We are bound to these infrastructures, dead technologies and negative commons, just as much as they bind us. Although their growth threatens the necessary reduction of our impact, we have a responsibility to maintain them until we can do without them. In university setting, as well as in any public organization, urban mines per se, we propose an IT architecture based on the exclusive use of unreliable waste from electrical and electronic equipment (WEEE) as a frugal alternative to the incessant replacement of devices. Powered by renewable energy, autonomous, robust, adaptable, and built on battle-tested open-source software, we envision this solution for a situation where use is bound to decline eventually, to close this damaging technological chapter. Digital technology, the idol of modern times, is to meet its twilight if we do not want to irrevocably alter the critical zone.

Pollinators play a crucial role for plant reproduction, either in natural ecosystem or in human-modified landscape. Global change drivers,including climate change or land use modifications, can alter the plant-pollinator interactions. To assess the potential influence of global change drivers on pollination, large-scale interactions, climate and land use data are required. While recent machine learning methods, such as graph neural networks (GNNs), allow the analysis of such datasets, interpreting their results can be challenging. We explore existing methods for interpreting GNNs in order to highlight the effects of various environmental covariates on pollination network connectivity. A large simulation study is performed to confirm whether these methods can detect the interactive effect between a covariate and a genus of plant on connectivity, and whether the application of debiasing techniques influences the estimation of these effects. An application on the Spipoll dataset, with and without accounting for sampling effects, highlights the potential impact of land use on network connectivity and shows that accounting for sampling effects partially alters the estimation of these effects.