Search

Job Corps is the largest and most comprehensive education and job training program in the United States for young people ages 16 to 24 who are not in school or working. Most students live at and take classes at its approximately 120 residential centers. To explore ways to improve student outcomes and reduce per-student costs, Job Corps pilot tested two nonresidential models where students received education and training at colleges, supported by intensive personal and career counseling, instead of at a center. The two pilot programs were: (1) The Job Corps Scholars (JCS) pilot program consisted of 26 colleges each delivering its own version of the Job Corps program to eligible young people; and (2) In the Idaho Job Corps (IJC) pilot program, Idaho paired its Boise Job Corps center with a nearby community college and opened three addition fully nonresidential programs at other colleges. The two pilot programs together offered 30 diverse examples of college-partnered Job Corps programs to study. The implementation and outcomes study focused on: (1) The types of students who enrolled in these programs; (2) The ability of colleges to serve young people eligible for Job Corps, who typically face more barriers than other students; and (3) The outcomes achieved by students in the pilot programs.

The Herschel open-time key program Disc Emission via a Bias-free Reconnaissance in the Infrared and Sub-millimeter (DEBRIS) is an unbiased survey of the nearest ~100 stars for each stellar type A-M observed with a uniform photometric sensitivity to search for cold debris disks around them. The analysis of the Photoconductor Array Camera and Spectrometer (PACS) photometric observations of the 94 DEBRIS M dwarfs of this program is presented in this paper, following upon two companion papers on the DEBRIS A-star and FGK-star subsamples. In the M-dwarf subsample, two debris disks have been detected, around the M3V dwarf GJ581 and the M4V dwarf FomalhautC (LP876-10). This result gives a disk detection rate of 2.1^{+2.7}_{-0.7}% at the 68% confidence level, significantly less than measured for earlier stellar types in the DEBRIS program. However, we show that the survey of the DEBRIS M-dwarf subsample is about ten times shallower than the surveys of the DEBRIS FGK subsamples when studied in the physical parameter space of the disk's fractional dust luminosity versus blackbody radius. Furthermore, had the DEBRIS K-star subsample been observed at the same shallower depth in this parameter space, its measured disk detection rate would have been statistically consistent with the one found for the M-dwarf subsample. Hence, the incidence of debris disks does not appear to drop from the K subsample to the M subsample of the DEBRIS program, when considering disks in the same region of physical parameter space. An alternative explanation is that the only two bright disks discovered in the M-dwarf subsample would not, in fact, be statistically representative of the whole population., Comment: 12 pages, long Table 3 included, accepted for publication in Astronomy & Astrophysics

Summary: VTX is a molecular visualization software capable to handle most molecular structures and dynamics trajectories file formats. It features a real-time high-performance molecular graphics engine, based on modern OpenGL, optimized for the visualization of massive molecular systems and molecular dynamics trajectories. VTX includes multiple interactive camera and user interaction features, notably free-fly navigation and a fully modular graphical user interface designed for increased usability. It allows the production of high-resolution images for presentations and posters with custom background. VTX design is focused on performance and usability for research, teaching and educative purposes. Availability and implementation: VTX is open source and free for non commercial use. Builds for Windows and Ubuntu Linux are available at http://vtx.drugdesign.fr. The source code is available at https://github.com/VTX-Molecular-Visualization, Comment: 5 pages, 1 figure, download VTX at http://vtx.drugdesign.fr, follow development at https://github.com/VTX-Molecular-Visualization

We present an all-silicon source of near-infrared linearly-polarized single photons, fabricated by nanoscale positioning of a color center in a silicon-on-insulator microcavity. The color center consists of a single W center, created at a well-defined position by Si$^{+}$ ion implantation through a 150 nm-diameter nanohole in a mask. A circular Bragg grating cavity resonant with the W's zero-phonon line at 1217 nm is fabricated at the same location as the nanohole. Under above-gap continuous-wave excitation, a very clean photon antibunching behavior ($g{^2} \leq 0.06$) is observed over the entire power range, which highlights the absence of parasitic emitters. Purcell-enhancement of W's zero-phonon emission provides both a record-high photoluminescence count rate among Si color centers (ca $1.2 \times 10^{6}$ counts/s) and apparent Debye-Waller factor around 99%. We also demonstrate the triggered emission of single photons with 93% purity under weak pulsed laser excitation. At high pulsed laser power, we reveal a detrimental effect of repumping processes, that could be mitigated using selective pumping schemes in the future. These results represent a major step towards on-demand sources of indistinguishable near-infrared single photons within silicon photonics chips.

Data augmentation is a crucial step in the development of robust supervised learning models, especially when dealing with limited datasets. This study explores interpolation techniques for the augmentation of geo-referenced data, with the aim of predicting the presence of Commelina benghalensis L. in sugarcane plots in La R{\'e}union. Given the spatial nature of the data and the high cost of data collection, we evaluated two interpolation approaches: Gaussian processes (GPs) with different kernels and kriging with various variograms. The objectives of this work are threefold: (i) to identify which interpolation methods offer the best predictive performance for various regression algorithms, (ii) to analyze the evolution of performance as a function of the number of observations added, and (iii) to assess the spatial consistency of augmented datasets. The results show that GP-based methods, in particular with combined kernels (GP-COMB), significantly improve the performance of regression algorithms while requiring less additional data. Although kriging shows slightly lower performance, it is distinguished by a more homogeneous spatial coverage, a potential advantage in certain contexts.

Data augmentation is a crucial step in the development of robust supervised learning models, especially when dealing with limited datasets. This study explores interpolation techniques for the augmentation of geo-referenced data, with the aim of predicting the presence of Commelina benghalensis L. in sugarcane plots in La R\'eunion. Given the spatial nature of the data and the high cost of data collection, we evaluated two interpolation approaches: Gaussian processes (GPs) with different kernels and kriging with various variograms. The objectives of this work are threefold: (i) to identify which interpolation methods offer the best predictive performance for various regression algorithms, (ii) to analyze the evolution of performance as a function of the number of observations added, and (iii) to assess the spatial consistency of augmented datasets. The results show that GP-based methods, in particular with combined kernels (GP-COMB), significantly improve the performance of regression algorithms while requiring less additional data. Although kriging shows slightly lower performance, it is distinguished by a more homogeneous spatial coverage, a potential advantage in certain contexts., Comment: Preprint. in French language

Singular Spectrum Analysis (SSA) occupies a prominent place in the real signal analysis toolkit alongside Fourier and Wavelet analysis. In addition to the two aforementioned analyses, SSA allows the separation of patterns directly from the data space into the data space, with data that need not be strictly stationary, continuous, or even normally sampled. In most cases, SSA relies on a combination of Hankel or Toeplitz matrices and Singular Value Decomposition (SVD). Like Fourier and Wavelet analysis, SSA has its limitations. The main bottleneck of the method can be summarized in three points. The first is the diagonalization of the Hankel/Toeplitz matrix, which can become a major problem from a memory and/or computational point of view if the time series to be analyzed is very long or heavily sampled. The second point concerns the size of the analysis window, typically denoted as 'L', which will affect the detection of patterns in the time series as well as the dimensions of the Hankel/Toeplitz matrix. Finally, the third point concerns pattern reconstruction: how to easily identify in the eigenvector/eigenvalue space which patterns should be grouped. We propose to address each of these issues by describing a hopefully effective approach that we have been developing for over 10 years and that has yielded good results in our research work., Comment: 28 pages, 11 figures

In this article, we derive a congruence property of particular sum rules involving prime numbers. The resulting expression involves Bernoulli numbers and polynomials, for which we obtain, as a consequence, a general congruence relation as well.

We point out that, although SMEFT is constructed to be model-independent, this holds only if complete sets of operators are used. SMEFT analyses that focus on subsets of operators are only semi-model-independent. They probe classes of ultraviolet models, opening a window to the underlying new physics. This model dependence originates from the need to choose a weak basis, usually taken to be the down or up basis. This naturally leads to the introduction of a third SMEFT weak basis, the generic basis, which interpolates between the above two bases. Due to its generality, we argue that the generic weak basis is the most natural for new physics. In the down and up weak bases, only a specific product of the two transformation matrices from the weak to the mass basis is observable -- the CKM matrix. However, in the generic weak basis, both transformation matrices are measurable in principle, possibly up to some phases., Comment: 9 pages

Computational Fluid Dynamics (CFD) is crucial for automotive design, requiring the analysis of large 3D point clouds to study how vehicle geometry affects pressure fields and drag forces. However, existing deep learning approaches for CFD struggle with the computational complexity of processing high-resolution 3D data. We propose Factorized Implicit Global Convolution (FIGConv), a novel architecture that efficiently solves CFD problems for very large 3D meshes with arbitrary input and output geometries. FIGConv achieves quadratic complexity $O(N^2)$, a significant improvement over existing 3D neural CFD models that require cubic complexity $O(N^3)$. Our approach combines Factorized Implicit Grids to approximate high-resolution domains, efficient global convolutions through 2D reparameterization, and a U-shaped architecture for effective information gathering and integration. We validate our approach on the industry-standard Ahmed body dataset and the large-scale DrivAerNet dataset. In DrivAerNet, our model achieves an $R^2$ value of 0.95 for drag prediction, outperforming the previous state-of-the-art by a significant margin. This represents a 40% improvement in relative mean squared error and a 70% improvement in absolute mean squared error over previous methods.

This paper presents BOUQuET, a multicentric and multi-register/domain dataset and benchmark, and its broader collaborative extension initiative. This dataset is handcrafted in non-English languages first, each of these source languages being represented among the 23 languages commonly used by half of the world's population and therefore having the potential to serve as pivot languages that will enable more accurate translations. The dataset is specially designed to avoid contamination and be multicentric, so as to enforce representation of multilingual language features. In addition, the dataset goes beyond the sentence level, as it is organized in paragraphs of various lengths. Compared with related machine translation (MT) datasets, we show that BOUQuET has a broader representation of domains while simplifying the translation task for non-experts. Therefore, BOUQuET is specially suitable for the open initiative and call for translation participation that we are launching to extend it to a multi-way parallel corpus to any written language.

Neoadjuvant chemotherapy (NAC) has become a standard clinical practice for tumor downsizing in breast cancer with 18F-FDG Positron Emission Tomography (PET). Our work aims to leverage PET imaging for the segmentation of breast lesions. The focus is on developing an automated system that accurately segments primary tumor regions and extracts key biomarkers from these areas to provide insights into the evolution of breast cancer following the first course of NAC. 243 baseline 18F-FDG PET scans (PET_Bl) and 180 follow-up 18F-FDG PET scans (PET_Fu) were acquired before and after the first course of NAC, respectively. Firstly, a deep learning-based breast tumor segmentation method was developed. The optimal baseline model (model trained on baseline exams) was fine-tuned on 15 follow-up exams and adapted using active learning to segment tumor areas in PET_Fu. The pipeline computes biomarkers such as maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) to evaluate tumor evolution between PET_Fu and PET_Bl. Quality control measures were employed to exclude aberrant outliers. The nnUNet deep learning model outperformed in tumor segmentation on PET_Bl, achieved a Dice similarity coefficient (DSC) of 0.89 and a Hausdorff distance (HD) of 3.52 mm. After fine-tuning, the model demonstrated a DSC of 0.78 and a HD of 4.95 mm on PET_Fu exams. Biomarkers analysis revealed very strong correlations whatever the biomarker between manually segmented and automatically predicted regions. The significant average decrease of SUVmax, MTV and TLG were 5.22, 11.79 cm3 and 19.23 cm3, respectively. The presented approach demonstrates an automated system for breast tumor segmentation from 18F-FDG PET. Thanks to the extracted biomarkers, our method enables the automatic assessment of cancer progression., Comment: Submit soon to EJNMMI Research

Caribou is a versatile data acquisition system used in multiple collaborative frameworks (CERN EP R&D, DRD3, AIDAinnova, Tangerine) for laboratory and test-beam qualification of novel silicon pixel detector prototypes. The system is built around a common hardware, firmware and software stack shared accross different projects, thereby drastically reducing the development effort and cost. It consists of a custom Control and Readout (CaR) board and a commercial Xilinx Zynq System-on-Chip (SoC) platform. The SoC platform runs a full Yocto distribution integrating the custom software framework (Peary) and a custom FPGA firmware built within a common firmware infrastructure (Boreal). The CaR board provides a hardware environment featuring various services such as powering, slow-control, and high-speed data links for the target detector prototype. Boreal and Peary, in turn, offer firmware and software architectures that enable seamless integration of control and readout for new devices. While the first version of the system used a SoC platform based on the ZC706 evaluation board, migration to a Zynq UltraScale+ architecture is progressing towards the support of the ZCU102 board and the ultimate objective of integrating the SoC functionality directly into the CaR board, eliminating the need for separate evaluation boards. This paper describes the Caribou system, focusing on the latest project developments and showcasing progress and future plans across its hardware, firmware, and software components., Comment: PIXEL-2024 Conference Proceeding

This work focuses on feedback control strategies for applying the sterile insect technique (SIT) to eliminate pest populations. The presentation is centered on the case of mosquito populations, but most of the results can be extended to other species by adapting the model and selecting appropriate parameter values to describe the reproduction and movement dynamics of the species under consideration. In our study, we address the spatial distribution of the population in a two dimensional bounded domain by extending the temporal SIT model analyzed in [2], thereby obtaining a reaction-diffusion SIT model. After the analysis of the existence and the uniqueness of the solution of this problem, we construct a feedback law that globally asymptotically stabilizes the extinction equilibrium thus yielding a robust strategy to keep the pest population at very low levels in the long term.

Cyanopolyynes are among the largest and most commonly observed interstellar Complex Organic Molecules in star-forming regions. They are believed to form primarily in the gas-phase, but their formation routes are not well understood. We present a comprehensive study of the gas-phase formation network of cyanobutadiyne, HC$_5$N, based on new theoretical calculations, kinetics experiments, astronomical observations, and astrochemical modeling. We performed new quantum mechanics calculations for six neutral-neutral reactions in order to derive reliable rate coefficients and product branching fractions. We also present new CRESU data on the rate coefficients of three of these reactions (C$_3$N + C$_2$H$_2$, C$_2$H + HC$_3$N, CN + C$_4$H$_2$) obtained at temperatures as low as 24 K. In practice, six out of nine reactions currently used in astrochemical models have been updated in our reviewed network. We also report the tentative detection of the $^{13}$C isotopologues of HC$_5$N in the L1544 prestellar core. We derived a lower limit of $^{12}$C/$^{13}$C > 75 for the HC$_5$N isotopologues, which does not allow to bring new constraints to the HC$_5$N chemistry. Finally, we verified the impact of the revised reactions by running the GRETOBAPE astrochemical model. We found good agreement between the HC$_5$N predicted and observed abundances in cold ($\sim$10 K) objects, demonstrating that HC$_5$N is mainly formed by neutral-neutral reactions in these environments. In warm molecular shocks, instead, the predicted abundances are a factor of ten lower with respect to observed ones. In this environment possessing an higher gas ionization fraction, we speculate that the contribution of ion-neutral reactions could be significant.

Building neural reward models from human preferences is a pivotal component in reinforcement learning from human feedback (RLHF) and large language model alignment research. Given the scarcity and high cost of human annotation, how to select the most informative pairs to annotate is an essential yet challenging open problem. In this work, we highlight the insight that an ideal comparison dataset for reward modeling should balance exploration of the representation space and make informative comparisons between pairs with moderate reward differences. Technically, challenges arise in quantifying the two objectives and efficiently prioritizing the comparisons to be annotated. To address this, we propose the Fisher information-based selection strategies, adapt theories from the classical experimental design literature, and apply them to the final linear layer of the deep neural network-based reward modeling tasks. Empirically, our method demonstrates remarkable performance, high computational efficiency, and stability compared to other selection methods from deep learning and classical statistical literature across multiple open-source LLMs and datasets. Further ablation studies reveal that incorporating cross-prompt comparisons in active reward modeling significantly enhances labeling efficiency, shedding light on the potential for improved annotation strategies in RLHF.

Large Language Models (LLMs) have made substantial strides in structured tasks through Reinforcement Learning (RL), demonstrating proficiency in mathematical reasoning and code generation. However, applying RL in broader domains like chatbots and content generation -- through the process known as Reinforcement Learning from Human Feedback (RLHF) -- presents unique challenges. Reward models in RLHF are critical, acting as proxies that evaluate the alignment of LLM outputs with human intent. Despite advancements, the development of reward models is hindered by challenges such as computational heavy training, costly evaluation, and therefore poor reproducibility. We advocate for using embedding-based input in reward model research as an accelerated solution to those challenges. By leveraging embeddings for reward modeling, we can enhance reproducibility, reduce computational demands on hardware, improve training stability, and significantly reduce training and evaluation costs, hence facilitating fair and efficient comparisons in this active research area. We then show a case study of reproducing existing reward model ensemble research using embedding-based reward models. We discussed future avenues for research, aiming to contribute to safer and more effective LLM deployments.

Linear Parameter Varying (LPV) Systems are a well-established class of nonlinear systems with a rich theory for stability analysis, control, and analytical response finding, among other aspects. Although there are works on data-driven identification of such systems, the literature is quite scarce in terms of works that tackle the identification of LPV models for large-scale systems. Since large-scale systems are ubiquitous in practice, this work develops a methodology for the local and global identification of large-scale LPV systems based on nonintrusive reduced-order modeling. The developed method is coined as DMD-LPV for being inspired in the Dynamic Mode Decomposition (DMD). To validate the proposed identification method, we identify a system described by a discretized linear diffusion equation, with the diffusion gain defined by a polynomial over a parameter. The experiments show that the proposed method can easily identify a reduced-order LPV model of a given large-scale system without the need to perform identification in the full-order dimension, and with almost no performance decay over performing a reduction, given that the model structure is well-established., Comment: 39 pages, 8 figures. Submitted to Journal of Computational Physics

We construct new invariant measures supported on mass level sets for the cubic defocusing nonlinear Schr\"odinger equation in dimensions $1$ and $2$.

Widefield magnetic imaging using ensembles of nitrogen-vacancy (NV) centres in diamond has emerged as a useful technique for studying the microscopic magnetic properties of materials. Thus far, this technique has mainly been implemented on custom-made optical microscopes. We have developed an add-on for a standard laboratory optical microscope that integrates the NV-diamond sensor and necessary light source, microwave antenna, and bias magnet, enabling NV-based magnetic imaging while retaining the typical optical measurements modes of the microscope. We demonstrate our retrofitted quantum diamond microscope by imaging a magnetic particle sample using brightfield, darkfield, and magnetic imaging modes. Furthermore, we employ an iso-magnetic field imaging technique to visualise the magnetic field of the sample within seconds, and finally demonstrate three-dimensional stray field imaging. Retrofitting existing microscopes exploits the stability and high quality of traditional optical microscope systems while reducing the cost and space requirements of establishing a standalone magnetic imaging system.

The success of Constraint Programming relies partly on the global constraints and implementation of the associated filtering algorithms. Recently, new ideas emerged to improve these implementations in practice, especially regarding the all different constraint. In this paper, we consider the cardinality constraint with costs. The cardinality constraint is a generalization of the all different constraint that specifies the number of times each value must be taken by a given set of variables in a solution. The version with costs introduces an assignment cost and bounds the total sum of assignment costs. The arc consistency filtering algorithm of this constraint is difficult to use in practice, as it systematically searches for many shortest paths. We propose a new approach that works with upper bounds on shortest paths based on landmarks. This approach can be seen as a preprocessing. It is fast and avoids, in practice, a large number of explicit computations of shortest paths., Comment: Published at the 30th International Conference on Principles and Practice of Constraint Programming (CP 2024)

Collaborative Object Detection (COD) and collaborative perception can integrate data or features from various entities, and improve object detection accuracy compared with individual perception. However, adversarial attacks pose a potential threat to the deep learning COD models, and introduce high output uncertainty. With unknown attack models, it becomes even more challenging to improve COD resiliency and quantify the output uncertainty for highly dynamic perception scenes such as autonomous vehicles. In this study, we propose the Trusted Uncertainty Quantification in Collaborative Perception framework (TUQCP). TUQCP leverages both adversarial training and uncertainty quantification techniques to enhance the adversarial robustness of existing COD models. More specifically, TUQCP first adds perturbations to the shared information of randomly selected agents during object detection collaboration by adversarial training. TUQCP then alleviates the impacts of adversarial attacks by providing output uncertainty estimation through learning-based module and uncertainty calibration through conformal prediction. Our framework works for early and intermediate collaboration COD models and single-agent object detection models. We evaluate TUQCP on V2X-Sim, a comprehensive collaborative perception dataset for autonomous driving, and demonstrate a 80.41% improvement in object detection accuracy compared to the baselines under the same adversarial attacks. TUQCP demonstrates the importance of uncertainty quantification to COD under adversarial attacks.

Recent advancements in foundation models have transformed computer vision, driving significant performance improvements across diverse domains, including digital histopathology. However, the advantages of domain-specific histopathology foundation models over general-purpose models for specialized tasks such as cell analysis remain underexplored. This study investigates the representation learning gap between these two categories by analyzing multi-level patch embeddings applied to cell instance segmentation and classification. We implement an encoder-decoder architecture with a consistent decoder and various encoders. These include convolutional, vision transformer (ViT), and hybrid encoders pre-trained on ImageNet-22K or LVD-142M, representing general-purpose foundation models. These are compared against ViT encoders from the recently released UNI, Virchow2, and Prov-GigaPath foundation models, trained on patches extracted from hundreds of thousands of histopathology whole-slide images. The decoder integrates patch embeddings from different encoder depths via skip connections to generate semantic and distance maps. These maps are then post-processed to create instance segmentation masks where each label corresponds to an individual cell and to perform cell-type classification. All encoders remain frozen during training to assess their pre-trained feature extraction capabilities. Using the PanNuke and CoNIC histopathology datasets, and the newly introduced Nissl-stained CytoDArk0 dataset for brain cytoarchitecture studies, we evaluate instance-level detection, segmentation accuracy, and cell-type classification. This study provides insights into the comparative strengths and limitations of general-purpose vs. histopathology foundation models, offering guidance for model selection in cell-focused histopathology and brain cytoarchitecture analysis workflows.

The SPIRAL2 superconducting linear accelerator (LINAC), which has been operational since 2019, employs superconducting, independently phased RF resonators to deliver a wide range of particle beams. Designed for flexibility in particle types, intensities, and energies, it utilizes superconducting quarter-wave resonators (QWRs), whose performances are critically dependent on the reliability of the cryogenic operation. This paper reviews the evolution from commissioning to the routine operation of the SPIRAL2 cryogenic system, initially commissioned in 2017. It highlights the key challenges encountered, including thermo-acoustic oscillations, thermal management, and abnormal behavior of cavities. Furthermore, it explores the integration of thermodynamic modeling and machine learning techniques to enhance system control and diagnose issues. This work serves as a comprehensive resource for advancing the cryogenic operation and performance of superconducting LINACs.

Railway transportation contributes to the objectives of decarbonization but also generates negative externalities, including noise. Energy noise indicators used to characterize population exposure do not adequately reflect the repetitive nature of railway noise peaks. The GENIFER pilot study aims to test a protocol designed to characterize railway noise events according to the instantaneous perceived annoyance when the train is passing, in order to improve understanding of the influence of acoustic factors on annoyance. The first phase of the survey was carried out in 2023 among 62 residents of a pilot site. An electronic device was used to collect around 5,000 ratings, ranging from 1 to 10, assessing the instantaneous annoyance induced by railway noise at passing trains. The site instrumentation included sixteen sound level meters and two video recording systems, enabling annoyance ratings to be associated with the acoustic characteristics of railway noise events. A questionnaire aimed at identifying co-determinants of long-term annoyance was also administered to participants. Feedback on the field implementation of this survey and initial results concerning acoustic measurements, instantaneous annoyance ratings and questionnaire responses will be presented.

Estimating optimal transport maps between two distributions from respective samples is an important element for many machine learning methods. To do so, rather than extending discrete transport maps, it has been shown that estimating the Brenier potential of the transport problem and obtaining a transport map through its gradient is near minimax optimal for smooth problems. In this paper, we investigate the private estimation of such potentials and transport maps with respect to the distribution samples.We propose a differentially private transport map estimator achieving an $L^2$ error of at most $n^{-1} \vee n^{-\frac{2 \alpha}{2 \alpha - 2 + d}} \vee (n\epsilon)^{-\frac{2 \alpha}{2 \alpha + d}} $ up to poly-logarithmic terms where $n$ is the sample size, $\epsilon$ is the desired level of privacy, $\alpha$ is the smoothness of the true transport map, and $d$ is the dimension of the feature space. We also provide a lower bound for the problem.

This paper introduces novel theoretical approximation bounds for the output of quantized neural networks, with a focus on convolutional neural networks (CNN). By considering layerwise parametrization and focusing on the quantization of weights, we provide bounds that gain several orders of magnitude compared to state-of-the art results on classical deep convolutional neural netorks such as MobileNetV2 or ResNets. These gains are achieved by improving the behaviour of the approximation bounds with respect to the depth parameter, which has the most impact on the approximation error induced by quantization. To complement our theoretical result, we provide a numerical exploration of our bounds on Mo-bileNetV2 and ResNets.

A method is presented for the improved calculation of super-shell partition functions which include the repulsive electron-electron interaction energy terms in the Boltzmann factor. Heretofore these interaction terms were approximately treated via the use of Feynman-Jensen inequalities. Such investigations are of particular interest for the super-transition-array approach of hot-plasma radiative opacity., Comment: submitted to "High Energy Density Phys."

The Keck Planet Imager and Characterizer (KPIC) is a series of upgrades for the Keck II Adaptive Optics (AO) system and the NIRSPEC spectrograph to enable diffraction limited, high resolution (R>30000) spectroscopy of exoplanets and low mass companions in the K and L bands. Phase I consisted of single mode fiber injection/extraction units (FIU/FEU) used in conjunction with a H band pyramid wavefront sensor. The use of single mode fibers provides a gain in stellar rejection, a substantial reduction in sky background, and an extremely stable line spread function in the spectrograph. Phase II, deployed and commissioned in 2022, brought a 1000 actuator deformable mirror, beam shaping optics, a vortex mask, and other upgrades to the FIU/FEU. An additional service mission in 2024 extended operations down to y band, delivered an atmospheric dispersion corrector, and provided access to two laser frequency combs. KPIC phase II brings higher planet throughput, lower stellar leakage and many new observing modes which extend its ability to characterize exoplanets at high spectral resolution, building on the success of phase I. In this paper we present a description of the final phase II version of KPIC, along with results of system level laboratory testing and characterization showing the instrument's phase II throughput, stability, repeatability, and other key performance metrics prior to delivery and during installation at Keck. We outlined the capabilities of the various observing modes enabled by the new modules as well as efforts to compensate for static aberrations and non common path errors at Keck, which were issues that plagued phase I. Finally, we show results from commissioning., Comment: 36 pages, 16 figures

We identify a chain of galaxies along an almost straight line in the nearby Universe with a projected length of ~5 Mpc. The galaxies are distributed within projected distances of only 7-105 kpc from the axis of the identified filament. They have redshifts in a very small range of z=0.0361-0.0370 so that their radial velocities are consistent with galaxy proper motions. The filament galaxies are mainly star-forming and have stellar masses in a range of $\rm 10^{9.1}-10^{10.7}\,M_{\odot}$. We search for systems with similar geometrical properties in the full-sky mock galaxy catalogue of the MillenniumTNG simulations and find that although such straight filaments are unusual and rare, they are predicted by $\Lambda$CDM simulations (4% incidence). We study the cold HI gas in a 1.3 Mpc section of the filament through HI-21cm emission line observations and detect eleven HI sources, many more than expected from the HI mass function in a similar volume. They have HI masses $\rm 10^{8.5}-10^{9.5}\,M_{\odot}$ and are mostly within ~120 kpc projected distance from the filament axis. None of these HI sources has a confirmed optical counterpart. Their darkness together with their large HI-21cm line-widths indicate that they contain gas that might not yet be virialized. These clouds must be marking the peaks of the dark matter and HI distributions over large scales within the filament. The presence of such gas clouds around the filament spines is predicted by simulations, but this is the first time that the existence of such clouds in a filament is observationally confirmed., Comment: Published in ApJL. Main figures: Fig. 1 and Fig. 3

In this work, we introduce a novel framework for privately optimizing objectives that rely on Wasserstein distances between data-dependent empirical measures. Our main theoretical contribution is, based on an explicit formulation of the Wasserstein gradient in a fully discrete setting, a control on the sensitivity of this gradient to individual data points, allowing strong privacy guarantees at minimal utility cost. Building on these insights, we develop a deep learning approach that incorporates gradient and activations clipping, originally designed for DP training of problems with a finite-sum structure. We further demonstrate that privacy accounting methods extend to Wasserstein-based objectives, facilitating large-scale private training. Empirical results confirm that our framework effectively balances accuracy and privacy, offering a theoretically sound solution for privacy-preserving machine learning tasks relying on optimal transport distances such as Wasserstein distance or sliced-Wasserstein distance.

We prove that globally hyperbolic compact anti-de Sitter (2+1)-spacetimes with strictly convex spacelike boundary that is either smooth or polyhedral and whose holonomy is close to Fuchsian are determined by the induced metric on the boundary.

String-net models describe a vast family of topological orders in two spatial dimensions, but fail to produce all the expected anyonic excitations. Following arXiv:1502.03433, we consider an extended string-net model by attaching one tail to each plaquette of the lattice, allowing all anyons to emerge as elementary plaquette excitations for arbitrary input categories. The corresponding tube algebra is the mathematical tool needed to construct the anyons from the input category and to obtain their internal multiplicities. We use them to compute the energy level degeneracies and the partition function. In the thermodynamic limit, the latter is dominated by the trivial (vacuum) anyon, so that the topological order is destroyed at any non-zero temperature. In a finite-size system, order survives up to a finite temperature, similarly to the one-dimensional classical Ising model. We confirm this by computing thermal averages of topological projectors, Wegner-Wilson loops and the topological mutual information. The results are also generalized to models with multiple tails per plaquette., Comment: 19 pages, 6 figures

Music is a deeply personal experience and our aim is to enhance this with a fully-automated pipeline for personalized music video generation. Our work allows listeners to not just be consumers but co-creators in the music video generation process by creating personalized, consistent and context-driven visuals based on lyrics, rhythm and emotion in the music. The pipeline combines multimodal translation and generation techniques and utilizes low-rank adaptation on listeners' images to create immersive music videos that reflect both the music and the individual. To ensure the ethical use of users' identity, we also introduce CHARCHA (patent pending), a facial identity verification protocol that protects people against unauthorized use of their face while at the same time collecting authorized images from users for personalizing their videos. This paper thus provides a secure and innovative framework for creating deeply personalized music videos., Comment: NeurIPS 2024 Creative AI Track

The International Liquid Mirror Telescope (ILMT) project was motivated by the need for an inexpensive 4 metre diameter optical telescope that could be devoted entirely to astronomical surveys. Its scientific programmes include the detection and study of transients, variable objects, asteroids, comets, space debris and low surface brightness galaxies. To this end, a collaboration was formed between the Institute of Astrophysics and Geophysics (Li\`ege University, Belgium), several Canadian universities (University of British Columbia, Laval University, University of Montreal, University of Toronto, York University, University of Victoria) and the Aryabhatta Research Institute of Observational Sciences (ARIES, India). After several years of design work in Belgium and construction in India on the ARIES Devasthal site, the telescope saw its first light on 29 April 2022. Its commissioning phase lasted from May 2022 until June 2023 (beginning of the monsoon). The ILMT was inaugurated on 21 March 2023 and has been in regular operation since October 2023. The telescope continuously observes the sky passing at the zenith using the SDSS g', r', and i' filters. This paper describes the ILMT, its operation, performance and shows some initial results., Comment: 12 pages, 18 figures, accepted for publication in Astronomy & Astrophysics

The measurement tasks involved in evaluating generative AI (GenAI) systems are especially difficult, leading to what has been described as "a tangle of sloppy tests [and] apples-to-oranges comparisons" (Roose, 2024). In this position paper, we argue that the ML community would benefit from learning from and drawing on the social sciences when developing and using measurement instruments for evaluating GenAI systems. Specifically, our position is that evaluating GenAI systems is a social science measurement challenge. We present a four-level framework, grounded in measurement theory from the social sciences, for measuring concepts related to the capabilities, behaviors, and impacts of GenAI. This framework has two important implications for designing and evaluating evaluations: First, it can broaden the expertise involved in evaluating GenAI systems by enabling stakeholders with different perspectives to participate in conceptual debates. Second, it brings rigor to both conceptual and operational debates by offering a set of lenses for interrogating the validity of measurement instruments and their resulting measurements.

The International Liquid Mirror Telescope (ILMT) is a 4-m aperture, zenith-pointing telescope with a field-of-view of 22', situated in the foothills of the Himalayas. The telescope operates in continuous survey mode, making it a useful instrument for time-domain astronomy, particularly for detecting transients, variable stars, active galactic nuclei variability, and asteroids. This paper presents the PyLMT transient detection pipeline to detect such transient/varying sources in the ILMT images. The pipeline utilises the image subtraction technique to compare a pair of images from the same field, identifying such sources in subtracted images with the help of convolutional neural networks (CNN) based real/bogus classifiers. The test accuracies determined for the real/bogus classifiers ranged from 94% to 98%. The resulting precision of the pipeline calculated over candidate alerts in the ILMT frames is 0.91. It also houses a CNN-aided transient candidate classifier that classifies the transient/variable candidates based on host morphology. The test accuracy of the candidate classifier is 98.6%. It has the provision to identify catalogued asteroids and other solar system objects using public databases. The median execution time of the pipeline is approximately 29 minutes per image of 17 minutes exposure. Relevant CNNs have been trained on data acquired with the ILMT during the cycle of October-November 2022. Subsequent tests on those images have confirmed the detection of numerous catalogued asteroids, variable stars, and other uncatalogued sources. The pipeline has been operational and has detected 12 extragalactic transients, including 2 new discoveries in the November 2023-May 2024 observation cycle., Comment: 21 pages, 26 figures, accepted for publication in MNRAS

In this paper, we consider the minimization of a $C^2-$smooth and strongly convex objective depending on a given parameter, which is usually found in many practical applications. We suppose that we desire to solve the problem with some inertial methods which cover a broader existing well-known inertial methods. Our main goal is to analyze the derivative of this algorithm as an infinite iterative process in the sense of ``automatic'' differentiation. This procedure is very common and has gain more attention recently. From a pure optimization perspective and under some mild premises, we show that any sequence generated by these inertial methods converge to the unique minimizer of the problem, which depends on the parameter. Moreover, we show a local linear convergence rate of the generated sequence. Concerning the differentiation of the scheme, we prove that the derivative of the sequence with respect to the parameter converges to the derivative of the limit of the sequence showing that any sequence is <>. Finally, we investigate the rate at which the convergence occurs. We show that, this is locally linear with an error term tending to zero.

This paper introduces a new method for testing the statistical significance of estimated parameters in predictive regressions. The approach features a new family of test statistics that are robust to the degree of persistence of the predictors. Importantly, the method accounts for serial correlation and conditional heteroskedasticity without requiring any corrections or adjustments. This is achieved through a mechanism embedded within the test statistics that effectively decouples serial dependence present in the data. The limiting null distributions of these test statistics are shown to follow a chi-square distribution, and their asymptotic power under local alternatives is derived. A comprehensive set of simulation experiments illustrates their finite sample size and power properties.

In resonance fluorescence excitation experiments, light emitted from solid-state quantum emitters is typically filtered to eliminate the laser photons, ensuring that only red-shifted Stokes photons are detected. Theoretical analyses of the fluorescence intensity correlation often model emitters as two-level systems, focusing on light emitted exclusively from the purely electronic transition (the Zero-Phonon Line), or rely on statistical approaches based on conditional probabilities that do not account for quantum coherences. Here, we propose a general model to characterize the correlation of either Zero-Phonon Line photons or Stokes-shifted photons. This model successfully reproduces the experimental correlation of Stokes-shifted photons emitted from two interacting molecules and predicts that this correlation is affected by quantum coherence. Besides, we analyze the role of quantum coherence in light emission from two uncorrelated emitters, which helps to clarify the discrepancy between theory and experiments regarding the value of the correlation of photons emitted from this system at zero delay time., Comment: Main text: 6 pages, 3 figures. Supplemental Material: 9 pages, 5 figures

In this paper, we explore a taxonomy of connectivity for space-like structures. It is inspired by isolating posets of connected pieces of a space and examining its embedding in the ambient space. The taxonomy includes in its scope all standard notions of connectivity in point-set and point-free contexts, such as connectivity in graphs and hypergraphs (as well as k-connectivity in graphs), connectivity and path-connectivity in topology, and connectivity of elements in a frame., Comment: 22 pages, 13 figures

A longstanding open question in sub-Riemannian geometry is the following: are sub-Riemannian length minimizers smooth? We give a negative answer to this question, exhibiting an example of a $C^2$ but not $C^3$ length-minimizer of a real-analytic (even polynomial) sub-Riemannian structure.

We consider a teacher-student model of supervised learning with a fully-trained 2-layer neural network whose width $k$ and input dimension $d$ are large and proportional. We compute the Bayes-optimal generalisation error of the network for any activation function in the regime where the number of training data $n$ scales quadratically with the input dimension, i.e., around the interpolation threshold where the number of trainable parameters $kd+k$ and of data points $n$ are comparable. Our analysis tackles generic weight distributions. Focusing on binary weights, we uncover a discontinuous phase transition separating a "universal" phase from a "specialisation" phase. In the first, the generalisation error is independent of the weight distribution and decays slowly with the sampling rate $n/d^2$, with the student learning only some non-linear combinations of the teacher weights. In the latter, the error is weight distribution-dependent and decays faster due to the alignment of the student towards the teacher network. We thus unveil the existence of a highly predictive solution near interpolation, which is however potentially hard to find., Comment: 8 pages + appendix, 3 figures

Vortex states in thin film disks host spin wave modes that are geometrically quantized according to their radial and azimuthal indices. Previous studies have shown that hybridization between these modes and the vortex core results in a sizeable frequency nonreciprocity between low-order clockwise and counterclockwise propagating azimuthal modes. Here, we present a computational study of these spin wave modes in submicron disks in which the spatial extension of the vortex core becomes comparable to the wavelength of certain modes. In such cases, we find that the frequency nonreciprocity can be large even for higher order radial and azimuthal indices, reaching several GHz and comparable to the mode frequencies themselves.

Probabilistic neurosymbolic learning seeks to integrate neural networks with symbolic programming. Many state-of-the-art systems rely on a reduction to the Probabilistic Weighted Model Counting Problem (PWMC), which requires computing a Boolean formula called the logical provenance.However, PWMC is \\#P-hard, and the number of clauses in the logical provenance formula can grow exponentially, creating a major bottleneck that significantly limits the applicability of PNL solutions in practice.We propose a new approach centered around an exact algorithm DPNL, that enables bypassing the computation of the logical provenance.The DPNL approach relies on the principles of an oracle and a recursive DPLL-like decomposition in order to guide and speed up logical inference.Furthermore, we show that this approach can be adapted for approximate reasoning with $\epsilon$ or $(\epsilon, \delta)$ guarantees, called ApproxDPNL.Experiments show significant performance gains.DPNL enables scaling exact inference further, resulting in more accurate models.Further, ApproxDPNL shows potential for advancing the scalability of neurosymbolic programming by incorporating approximations even further, while simultaneously ensuring guarantees for the reasoning process.

This manuscript develops a framework for the strong approximation of Sobolev maps with values in compact manifolds, emphasizing the interplay between local and global topological properties. Building on topological concepts adapted to VMO maps, such as homotopy and the degree of continuous maps, it introduces and analyzes extendability properties, focusing on the notions of $\ell$-extendability and its generalization, $(\ell, e)$-extendability. We rely on Fuglede maps, providing a robust setting for handling compositions with Sobolev maps. Several constructions -- including opening, thickening, adaptive smoothing, and shrinking -- are carefully integrated into a unified approach that combines homotopical techniques with precise quantitative estimates. Our main results establish that a Sobolev map $u \in W^{k, p}$ defined on a compact manifold of dimension $m > kp$ can be approximated by smooth maps if and only if $u$ is $(\lfloor kp \rfloor, e)$-extendable with $e = m$. When $e < m$, the approximation can still be carried out using maps that are smooth except on structured singular sets of rank $m - e - 1$., Comment: 214 pages

The ability to generate quantum emitters with reproducible properties in solid-state matrices is crucial for quantum technologies. Here, we show that a high density of close-to-identical single-photon emitters can be created in commercial hexagonal boron nitride using annealing under oxygen atmosphere. This simple procedure yields a uniform in-plane distribution of color centers consistently emitting around 539.4 nm, with a wavelength spread smaller than 1 nm. We present an extensive characterization of their photophysical properties, showing that the emitters are bright and stable, and exhibit narrow lines at low temperatures with minimal spectral diffusion. These characteristics make this family of quantum emitters highly appealing for applications to quantum information science.

This work considers an Inertial version of Bregman Proximal Gradient algorithm (IBPG) for minimizing the sum of two single-valued functions in finite dimension. We suppose that one of the functions is proper, closed, and convex but non-necessarily smooth whilst the second is a smooth enough function but not necessarily convex. For the latter, we ask the smooth adaptable property (smad) with respect to some kernel or entropy which allows to remove the very popular global Lipschitz continuity requirement on the gradient of the smooth part. We consider the IBPG under the framework of the triangle scaling property (TSP) which is a geometrical property for which one can provably ensure acceleration for a certain subset of kernel/entropy functions in the convex setting. Based on this property, we provide global convergence guarantees when the entropy is strongly convex under the framework of the Kurdyka-\L{}ojasiewicz (KL) property. Turning to the local convergence properties, we show that when the nonsmooth part is partly smooth relative to a smooth submanifold, IBPG has a finite activity identification property before entering a local linear convergence regime for which we establish a sharp estimate of the convergence rate. We report numerical simulations to illustrate our theoretical results on low complexity regularized phase retrieval.