Search

The Latino and Adult Student Success (LASS) Academy is a multi-year initiative administered by CAEL (the Council for Adult and Experiential Learning) focused on supporting Hispanic-Serving Institutions (HSIs) in their efforts to improve outcomes for adult and Latino adult learners. In its first phase, from 2018 through 2021, the LASS Academy supported fifteen postsecondary initiatives across the country. In the second phase, from 2022 through 2024, the LASS Academy continued its work with four Texas institutions: Austin Community College (ACC), South Texas College (STC), Texas A&M University-Corpus Christi (TAMU-CC), and the University of Texas Rio Grande Valley (UTRGV). The LASS Academy model facilitates institutional change through four key activities: identify key priorities and action steps; provide technical assistance; collect, analyze, and report data; and facilitate peer-to-peer learning. During the second LASS Academy, each of the four participating institutions made progress toward their goals, buttressed by change management support from CAEL and collaboration with other technical assistance partners. Throughout this work, the institutions gained important insights into ways they can improve their services for adult and Latino adult learners.

Constructing valid confidence sets is a crucial task in statistical inference, yet traditional methods often face challenges when dealing with complex models or limited observed sample sizes. These challenges are frequently encountered in modern applications, such as Likelihood-Free Inference (LFI). In these settings, confidence sets may fail to maintain a confidence level close to the nominal value. In this paper, we introduce two novel methods, TRUST and TRUST++, for calibrating confidence sets to achieve distribution-free conditional coverage. These methods rely entirely on simulated data from the statistical model to perform calibration. Leveraging insights from conformal prediction techniques adapted to the statistical inference context, our methods ensure both finite-sample local coverage and asymptotic conditional coverage as the number of simulations increases, even if n is small. They effectively handle nuisance parameters and provide computationally efficient uncertainty quantification for the estimated confidence sets. This allows users to assess whether additional simulations are necessary for robust inference. Through theoretical analysis and experiments on models with both tractable and intractable likelihoods, we demonstrate that our methods outperform existing approaches, particularly in small-sample regimes. This work bridges the gap between conformal prediction and statistical inference, offering practical tools for constructing valid confidence sets in complex models.

Federated learning (FL) is an appealing approach to training machine learning models without sharing raw data. However, standard FL algorithms are iterative and thus induce a significant communication cost. One-shot federated learning (OFL) trades the iterative exchange of models between clients and the server with a single round of communication, thereby saving substantially on communication costs. Not surprisingly, OFL exhibits a performance gap in terms of accuracy with respect to FL, especially under high data heterogeneity. We introduce FENS, a novel federated ensembling scheme that approaches the accuracy of FL with the communication efficiency of OFL. Learning in FENS proceeds in two phases: first, clients train models locally and send them to the server, similar to OFL; second, clients collaboratively train a lightweight prediction aggregator model using FL. We showcase the effectiveness of FENS through exhaustive experiments spanning several datasets and heterogeneity levels. In the particular case of heterogeneously distributed CIFAR-10 dataset, FENS achieves up to a 26.9% higher accuracy over state-of-the-art (SOTA) OFL, being only 3.1% lower than FL. At the same time, FENS incurs at most 4.3x more communication than OFL, whereas FL is at least 10.9x more communication-intensive than FENS., Comment: Accepted at NeurIPS 2024

Fiducial markers are a computer vision tool used for object pose estimation and detection. These markers are highly useful in fields such as industry, medicine and logistics. However, optimal lighting conditions are not always available,and other factors such as blur or sensor noise can affect image quality. Classical computer vision techniques that precisely locate and decode fiducial markers often fail under difficult illumination conditions (e.g. extreme variations of lighting within the same frame). Hence, we propose DeepArUco++, a deep learning-based framework that leverages the robustness of Convolutional Neural Networks to perform marker detection and decoding in challenging lighting conditions. The framework is based on a pipeline using different Neural Network models at each step, namely marker detection, corner refinement and marker decoding. Additionally, we propose a simple method for generating synthetic data for training the different models that compose the proposed pipeline, and we present a second, real-life dataset of ArUco markers in challenging lighting conditions used to evaluate our system. The developed method outperforms other state-of-the-art methods in such tasks and remains competitive even when testing on the datasets used to develop those methods. Code available in GitHub: https://github.com/AVAuco/deeparuco/

The joint use of event-based vision and Spiking Neural Networks (SNNs) is expected to have a large impact in robotics in the near future, in tasks such as, visual odometry and obstacle avoidance. While researchers have used real-world event datasets for optical flow prediction (mostly captured with Unmanned Aerial Vehicles (UAVs)), these datasets are limited in diversity, scalability, and are challenging to collect. Thus, synthetic datasets offer a scalable alternative by bridging the gap between reality and simulation. In this work, we address the lack of datasets by introducing eWiz, a comprehensive library for processing event-based data. It includes tools for data loading, augmentation, visualization, encoding, and generation of training data, along with loss functions and performance metrics. We further present a synthetic event-based datasets and data generation pipelines for optical flow prediction tasks. Built on top of eWiz, eCARLA-scenes makes use of the CARLA simulator to simulate self-driving car scenarios. The ultimate goal of this dataset is the depiction of diverse environments while laying a foundation for advancing event-based camera applications in autonomous field vehicle navigation, paving the way for using SNNs on neuromorphic hardware such as the Intel Loihi.

In this work, we present an innovative application of the probabilistic weak formulation of mean field games (MFG) for modeling liquidity pools in a constant product automated market maker (AMM) protocol in the context of decentralized finance. Our work extends one of the most conventional applications of MFG, which is the price impact model in an order book, by incorporating an AMM instead of a traditional order book. Through our approach, we achieve results that support the existence of solutions to the Mean Field Game and, additionally, the existence of approximate Nash equilibria for the proposed problem. These results not only offer a new perspective for representing liquidity pools in AMMs but also open promising opportunities for future research in this emerging field., Comment: 27 pages, 1 figure

Cross-lingual in-context learning (XICL) has emerged as a transformative paradigm for leveraging large language models (LLMs) to tackle multilingual tasks, especially for low-resource languages. However, existing approaches often rely on external retrievers or task-specific fine-tuning, limiting their scalability and generalizability. In this paper, we propose a novel self-supervised framework that harnesses the generative capabilities of LLMs to internally select and utilize task-relevant examples. Our method introduces two key objectives: a retrieval-generation alignment loss to optimize the quality of selected examples and a semantic coherence loss to ensure cross-lingual consistency. Through extensive experiments on multilingual benchmarks, our approach achieves state-of-the-art performance, significantly outperforming existing baselines. Further analysis highlights its robustness across diverse language families and its ability to generalize to unseen tasks. Human evaluations confirm the superior fluency, relevance, and semantic correctness of outputs generated by our method. This work provides a scalable, effective, and generalizable solution for cross-lingual in-context learning.

We report the validation of multiple planets transiting the nearby ($d = 12.8$ pc) K5V dwarf HD 101581 (GJ 435, TOI-6276, TIC 397362481). The system consists of at least two Earth-size planets whose orbits are near a mutual 4:3 mean-motion resonance, HD 101581 b ($R_{p} = 0.956_{-0.061}^{+0.063}~R_{\oplus}$, $P = 4.47$ days) and HD 101581 c ($R_{p} = 0.990_{-0.070}^{+0.070}~R_{\oplus}$, $P = 6.21$ days). Both planets were discovered in Sectors 63 and 64 TESS observations and statistically validated with supporting ground-based follow-up. We also identify a signal that probably originates from a third transiting planet, TOI-6276.03 ($R_{p} = 0.982_{-0.098}^{+0.114}~R_{\oplus}$, $P = 7.87$ days). These planets are remarkably uniform in size and their orbits are evenly spaced, representing a prime example of the "peas-in-a-pod" architecture seen in other compact multi-planet systems. At $V = 7.77$, HD 101581 is the brightest star known to host multiple transiting planets smaller than $1.5~R_{\oplus}$. HD 101581 is a promising system for atmospheric characterization and comparative planetology of small planets., Comment: 25 pages, 13 figures. Accepted for publication in AJ

Essential information about the formation and evolution of planetary systems can be found in their architectures -- in particular, in stellar obliquity ($\psi$) -- as they serve as a signature of their dynamical evolution. Here, we present ESPRESSO observations of the Rossiter-Mclaughlin (RM) effect of 8 warm gas giants, revealing that independent of the eccentricities, all of them have relatively aligned orbits. Our 5 warm Jupiters -- WASP-106 b, WASP-130 b, TOI-558 b, TOI-4515 b, and TOI-5027 b -- have sky-projected obliquities $|\lambda|\simeq0-10$ deg while the 2 less massive warm Saturns -- K2-139 b and K2-329 A b -- are slightly misaligned having $|\lambda|\simeq15-25$ deg. Furthermore, for K2-139 b, K2-329 A b, and TOI-4515 b, we also measure true 3D obliquities $\psi\simeq15-30$ deg. We also report a non-detection of the RM effect produced by TOI-2179 b. Through hierarchical Bayesian modeling of the true 3D obliquities of hot and warm Jupiters, we find that around single stars, warm Jupiters are statistically more aligned than hot Jupiters. Independent of eccentricities, 95\% of the warm Jupiters have $\psi\lesssim30$ deg with no misaligned planets, while hot Jupiters show an almost isotropic distribution of misaligned systems. This implies that around single stars, warm Jupiters form in primordially aligned protoplanetary disks and subsequently evolve in a more quiescent way than hot Jupiters. Finally, we find that Saturns may have slightly more misaligned orbits than warm Jupiters, but more obliquity measurements are necessary to be conclusive., Comment: 16 pages, 6 figures. Submitted to AAS journals. Comments are welcomed

Image retrieval methods rely on metric learning to train backbone feature extraction models that can extract discriminant queries and reference (gallery) feature representations for similarity matching. Although state-of-the-art accuracy has improved considerably with the advent of deep learning (DL) models trained on large datasets, image retrieval remains challenging in many real-world video analytics and surveillance applications, e.g., person re-identification. Using the Euclidean space for matching limits the performance in real-world applications due to the curse of dimensionality, overfitting, and sensitivity to noisy data. We argue that the feature dissimilarity space is more suitable for similarity matching, and propose a dichotomy transformation to project query and reference embeddings into a single embedding in the dissimilarity space. We also advocate for end-to-end training of a backbone and binary classification models for pair-wise matching. As opposed to comparing the distance between queries and reference embeddings, we show the benefits of classifying the single dissimilarity space embedding (as similar or dissimilar), especially when trained end-to-end. We propose a method to train the max-margin classifier together with the backbone feature extractor by applying constraints to the L2 norm of the classifier weights along with the hinge loss. Our extensive experiments on challenging image retrieval datasets and using diverse feature extraction backbones highlight the benefits of similarity matching in the dissimilarity space. In particular, when jointly training the feature extraction backbone and regularised classifier for matching, the dissimilarity space provides a higher level of accuracy., Comment: 7 pages

We present TapeAgents, an agent framework built around a granular, structured log tape of the agent session that also plays the role of the session's resumable state. In TapeAgents we leverage tapes to facilitate all stages of the LLM Agent development lifecycle. The agent reasons by processing the tape and the LLM output to produce new thought and action steps and append them to the tape. The environment then reacts to the agent's actions by likewise appending observation steps to the tape. By virtue of this tape-centred design, TapeAgents can provide AI practitioners with holistic end-to-end support. At the development stage, tapes facilitate session persistence, agent auditing, and step-by-step debugging. Post-deployment, one can reuse tapes for evaluation, fine-tuning, and prompt-tuning; crucially, one can adapt tapes from other agents or use revised historical tapes. In this report, we explain the TapeAgents design in detail. We demonstrate possible applications of TapeAgents with several concrete examples of building monolithic agents and multi-agent teams, of optimizing agent prompts and finetuning the agent's LLM. We present tooling prototypes and report a case study where we use TapeAgents to finetune a Llama-3.1-8B form-filling assistant to perform as well as GPT-4o while being orders of magnitude cheaper. Lastly, our comparative analysis shows that TapeAgents's advantages over prior frameworks stem from our novel design of the LLM agent as a resumable, modular state machine with a structured configuration, that generates granular, structured logs and that can transform these logs into training text -- a unique combination of features absent in previous work.

A graph $G$ with $p$ vertices and $q$ edges is said to be edge-graceful if its edges can be labeled from $1$ through $q$, in such a way that the labels induced on the vertices by adding over the labels of incident edges modulo $p$ are distinct. A known result under this topic is Lo's Theorem, which states that if a graph $G$ with $p$ vertices and $q$ edges is edge-graceful, then $p\Big|\Big(q^{2}+q-\dfrac{p(p-1)}{2}\Big)$. This paper presents novel results on the edge-gracefulness of the usual fan graphs. Using Lo's Theorem, the concepts of divisibility and Diophantine equations, and a computer program created, we determine all edge-graceful usual fan graphs $F_{1,n}$ with their corresponding edge-graceful labels., Comment: 19 pages with C-codes as appendix

Quantum finite automata can be used for pattern recognition. Present implementations on actual quantum devices face decoherence issues, which compromise the quality of long strings computation. In this work, we focus on the Measure Once 1-way Quantum Finite Automata (MO1QFA) model for addressing the MOD^p problem, investigating how quantum errors may affect the quality of the computation in this model when implemented in IBM-Q superconducting environment. To improve the performance of the implementation, we use pulse-level programming for calibrating a fast single-qubit gate designed specifically for the automaton implementation. The demonstrations conducted on the Jakarta quantum computer show that using custom pulses significantly reduces errors during extended word computations. While realizing improvements in error variations and predictability -- with a fourfold reduction in circuit latency -- the proposed solution demonstrates a substantial increase in the supported computation length of the automaton. When considering thresholds of 10% and 20% in absolute errors of acceptance probabilities, the solution has the potential to increase the maximum word length by 12 and 7+ times, respectively, compared to the default Qiskit gate.

This paper addresses the integration of \emph{epistemic entrenchment} into paraconsistent belief revision systems based on Logics of Formal Inconsistency (LFIs). While systems like AGMp and AGM$\circ$ adapt AGM principles to paraconsistency, they lack mechanisms to rank beliefs, primarily due to the absence of properties such as the \emph{replacement property} in the underlying logics. We introduce two novel logics, \textbf{Cbr} and \textbf{RCBr}, with the latter extending the former to fully address these limitations given that it is self-extensional. Using \textbf{RCBr}, we define contraction operations via \emph{epistemic entrenchment}, adhering to key rationality principles. Our framework leverages non-deterministic matrix semantics (Nmatrices), providing a robust foundation for paraconsistent reasoning. These contributions advance the theory of paraconsistent belief revision and pave the way for applications in domains such as multi-agent systems and inconsistent knowledge bases.

We present a joint analysis of TTVs and Doppler data for the transiting exoplanet system TOI-4504. TOI-4504 c is a warm Jupiter-mass planet that exhibits the largest known transit timing variations (TTVs), with a peak-to-node amplitude of $\sim$ 2 days, the largest value ever observed, and a super-period of $\sim$ 930 d. TOI-4504 b and c were identified in public TESS data, while the TTVs observed in TOI-4504 c, together with radial velocity (RV) data collected with FEROS, allowed us to uncover a third, non-transiting planet in this system, TOI-4504 d. We were able to detect transits of TOI-4504 b in the TESS data with a period of 2.4261$\pm 0.0001$ days and derive a radius of 2.69$\pm 0.19$ R$_{\oplus}$. The RV scatter of TOI-4504 was too large to constrain the mass of TOI-4504 b, but the RV signals of TOI-4504 c \& d were sufficiently large to measure their masses. The TTV+RV dynamical model we apply confirms TOI-4504 c as a warm Jupiter planet with an osculating period of 82.54$\pm 0.02$ d, mass of 3.77$\pm 0.18$ M$_{\rm J}$ and a radius of 0.99$\pm 0.05$ R$_{\rm J}$, while the non-transiting planet TOI-4504 d, has an orbital period of 40.56$\pm 0.04$ days and mass of 1.42$_{-0.06}^{+0.07}$ M$_{\rm J}$. We present the discovery of a system with three exoplanets: a hot sub-Neptune and two warm Jupiter planets. The gas giant pair is stable and likely locked in a first-order 2:1 mean-motion resonance (MMR). The TOI-4504 system is an important addition to MMR pairs, whose increasing occurrence supports a smooth migration into a resonant configuration during the protoplanetary disk phase., Comment: 20 pages, 12 figures, accepted in ApJL

We obtain the Hawking spectrum by exponentiating a series of Feynman diagrams describing a scalar field scattering through a collapse background. Our approach is rooted in semiclassical methods of scattering amplitudes which have recently been developed for application to gravitational-wave physics. The diagrams we encounter do not compute a standard amplitude, but rather an in-in generalisation of an amplitude which is closely connected to the Bogoliubov coefficients. We also compute the subdominant one-loop correction in our perturbative approach, analogous to the triangle correction to Schwarzschild scattering. This term can be interpreted as a finite-size correction sensitive to the radius of the black hole., Comment: 35 pages + appendices

Time-resolved X-ray absorption spectroscopy (tr-XAS) has been shown to be a versatile measurement technique for investigating non-equilibrium dynamics. Novel X-ray free electron laser (XFEL) facilities like the European XFEL offer increased repetition rates for stroboscopic XAS experiments through a burst operation mode, which enables measurements with up to 4.5 MHz. These higher repetition rates lead to higher data acquisition rates but can also introduce long-lived excitations that persist and thus build up during each burst. Here, we report on such long-lived effects in Ni and NiO thin film samples that were measured at the European XFEL. We disentangle the long-lived excitations from the initial pump-induced change and perform a detailed modelling-based analysis of how they modify transient X-ray spectra. As a result, we link the long-lived effects in Ni to a local temperature increase, as well as the effects in NiO to excited charge carrier trapping through polaron formation. In addition, we present possible correction methods, as well as discuss ways in which the effects of these long-lived excitations could be minimized for future time-resolved X-ray absorption spectroscopy measurements.

This work explores the topological properties of altermagnets, a novel class of collinear magnetic materials. We employ equivariant K-theory of magnetic groups and Hamiltonian models to formulate a robust $C^z_4 \mathbb{T}$ topological invariant to classify 2D and 3D altermagnetic systems. Our findings demonstrate that the spin Chern number serves as a robust topological index, corresponding to the half-quantized Chern number of the divided Brillouin zone. This indicator enables the prediction of a topologically protected 2D altermagnetic insulators and 3D Weyl altermagnetic semimetals, highlighting the relationship between altermagnetism and topological phases. Furthermore, our results provide a pathway to the exploration of topological applications in $d$-wave altermagnetic materials., Comment: 12 pages, 5 figures

Numerical N-body simulations are commonly used to explore stability regions around exoplanets, offering insights into the possible existence of satellites and ring systems. This study aims to utilize Machine Learning (ML) techniques to generate predictive maps of stable regions surrounding a hypothetical planet. The approach can also be extended to planet-satellite systems, planetary ring systems, and other similar configurations. A dataset was generated using 10^5 numerical simulations, each incorporating nine orbital features for the planet and a test particle in a star-planet-test particle system. The simulations were classified as stable or unstable based on stability criteria, requiring particles to remain stable over a timespan equivalent to 10,000 orbital periods of the planet. Various ML algorithms were tested and fine-tuned through hyperparameter optimization to determine the most effective predictive model. Tree-based algorithms showed comparable accuracy in performance. The best-performing model, using the Extreme Gradient Boosting (XGBoost) algorithm, achieved an accuracy of 98.48%, with 94% recall and precision for stable particles and 99% for unstable particles. ML algorithms significantly reduce the computational time required for three-body simulations, operating approximately 100,000 times faster than traditional numerical methods. Predictive models can generate entire stability maps in less than a second, compared to the days required by numerical simulations. The results from the trained ML models will be made accessible through a public web interface, enabling broader scientific applications., Comment: 12 pages, 13 figures, Accepted for publication in Astronomy & Astrophysics

The coldest metal-poor population made of T and Y dwarfs are archaeological tracers of our Galaxy because they are very old and have kept the pristine material. The optical properties of these objects are important to characterise their atmospheric properties. We aim at characterising further the optical properties of ultracool metal-poor population with deep far-red optical images and parallax determinations. We solve trigonometric parallaxes of five metal-poor T dwarf candidates using 2-year monitoring with Calar-Alto 3.5-m telescope. We obtain $z'$-band photometry for the other 12 metal-poor T dwarf candidates using the 10.4-m GTC, the 8.2-m VLT, and the DES, increasing the sample of T subdwarfs with optical photometry from 12 to 24. We report a 3-$\sigma$ limit for the Accident in five optical bands using the 10.4-m GTC. We confirm four T subdwarfs and the Accident as a Y subdwarf, and propose two more Y subdwarf candidates. We emphasise that the $z_{PS1}-W1$ colour combining with the $W1-W2$ colour could break the metallicity-temperature degeneracy for T and possibly for Y dwarfs. The $z_{PS1}-W1$ colour shifts redward when metallicity decreases for a certain temperature, which is not predicted by state-of-the-art ultracool models. The Accident has the reddest $z_{PS1}-W1$ colour among our sample. The $z_{PS1}-W1$ colour will be useful to search for other examples of this cold and old population in upcoming and existing deep optical and infrared large-area surveys., Comment: 17 pages, 7 figures, 2 appendices, submitted to A&A, comments are welcome

We report JWST NIRSpec/G395H observations of TOI-1685 b, a hot rocky super-Earth orbiting an M2.5V star, during a full orbit. We obtain transmission and emission spectra of the planet and characterize the properties of the phase curve, including its amplitude and offset. The transmission spectrum rules out clear H$_2$-dominated atmospheres, while secondary atmospheres (made of water, methane, or carbon dioxide) cannot be statistically distinguished from a flat line. The emission spectrum is featureless and consistent with a blackbody-like brightness temperature, helping rule out thick atmospheres with high mean molecular weight. Collecting all evidence, the properties of TOI-1685 b are consistent with a blackbody with no heat redistribution and a low albedo, with a dayside brightness temperature 0.98$\pm$0.07 times that of a perfect blackbody in the NIRSpec NRS2 wavelength range (3.823-5.172 um). Our results add to the growing number of seemingly airless M-star rocky planets, thus constraining the location of the "Cosmic Shoreline". Three independent data reductions have been carried out, all showing a high-amplitude correlated noise component in the white and spectroscopic light curves. The correlated noise properties are different between the NRS1 and NRS2 detectors - importantly the timescales of the strongest components (4.5 hours and 2.5 hours, respectively) - suggesting the noise is from instrumental rather than astrophysical origins. We encourage the community to look into the systematics of NIRSpec for long time-series observations., Comment: 26 pages, 19 figures, 8 tables. Submitted to AAS Journals

We present the detection and characterization of TOI-4994b (TIC 277128619b), a warm Saturn-sized planet discovered by the NASA Transiting Exoplanet Survey Satellite (TESS). TOI-4994b transits a G-type star (V = 12.6 mag) with a mass, radius, and effective temperature of $M_{\star} =1.005^{+0.064}_{-0.061} M_{\odot}$, $R_{\star} = 1.055^{+0.040}_{-0.037} R_{\odot}$, and $T_{\rm eff} = 5640 \pm 110$ K. We obtained follow-up ground-based photometry from the Las Cumbres Observatory (LCO) and the Antarctic Search for Transiting ExoPlanets (ASTEP) telescopes, and we confirmed the planetary nature of TOI-4994b with multiple radial velocity observations from the PFS, CHIRON, HARPS, FEROS, and CORALIE instruments. From a global fit to the photometry and radial velocities, we determine that TOI-4994b is in a 21.5-day, eccentric orbit ($e = 0.32 \pm 0.04$) and has a mass of $M_{P}= 0.280^{+0.037}_{-0.034} M_{J}$, a radius of $R_{P}= 0.762^{+0.030}_{-0.027}R_{J}$, and a Saturn-like bulk density of $\rho_{p} = 0.78^{+0.16}_{-0.14}$ $\rm g/cm^3$. We find that TOI-4994 is a potentially viable candidate for follow-up stellar obliquity measurements. TOI-4994b joins the small sample of warm Saturn analogs and thus sheds light on our understanding of these rare and unique worlds., Comment: 17 pages, 10 figures. Accepted to AJ

A relation between the Dirac bracket (DB) and Nambu bracket (NB) is presented. The Nambu bracket can be related with Dirac bracket if we can write the DB as a generalized Poisson structure. The NB associated with DB have all the standard properties of the original DB. When the dimension of the phase space is $s+2$ where $s$ is the number of second class constraints, the associated Nambu structure has $s+2$ entries and reduces to the Dirac bracket when $s$ of its entries are fixed to be the second class constraints. In general, when the dimension of phase space is $d=r+s$ a new Nambu structure that describes correctly the constrained dynamics can also be constructed but in thsi case addicional conditidionts are requiered. In that case the associated NB corresponds to a ``Dirac-Nambu'' bracket with $r$ entries., Comment: 19 pages

We present the MULTIMODAL UNIVERSE, a large-scale multimodal dataset of scientific astronomical data, compiled specifically to facilitate machine learning research. Overall, the MULTIMODAL UNIVERSE contains hundreds of millions of astronomical observations, constituting 100\,TB of multi-channel and hyper-spectral images, spectra, multivariate time series, as well as a wide variety of associated scientific measurements and "metadata". In addition, we include a range of benchmark tasks representative of standard practices for machine learning methods in astrophysics. This massive dataset will enable the development of large multi-modal models specifically targeted towards scientific applications. All codes used to compile the MULTIMODAL UNIVERSE and a description of how to access the data is available at https://github.com/MultimodalUniverse/MultimodalUniverse, Comment: Accepted at NeurIPS Datasets and Benchmarks track

Electronic healthcare records (EHR) contain a huge wealth of data that can support the prediction of clinical outcomes. EHR data is often stored and analysed using clinical codes (ICD10, SNOMED), however these can differ across registries and healthcare providers. Integrating data across systems involves mapping between different clinical ontologies requiring domain expertise, and at times resulting in data loss. To overcome this, code-agnostic models have been proposed. We assess the effectiveness of a code-agnostic representation approach on the task of long-term microvascular complication prediction for individuals living with Type 2 Diabetes. Our method encodes individual EHRs as text using fine-tuned, pretrained clinical language models. Leveraging large-scale EHR data from the UK, we employ a multi-label approach to simultaneously predict the risk of microvascular complications across 1-, 5-, and 10-year windows. We demonstrate that a code-agnostic approach outperforms a code-based model and illustrate that performance is better with longer prediction windows but is biased to the first occurring complication. Overall, we highlight that context length is vitally important for model performance. This study highlights the possibility of including data from across different clinical ontologies and is a starting point for generalisable clinical models., Comment: Findings paper presented at Machine Learning for Health (ML4H) symposium 2024, December 15-16, 2024, Vancouver, Canada, 9 pages

We report the discovery and characterization of three new transiting giant planets orbiting TOI-6628, TOI-3837 and TOI-5027, and one new warm sub-Saturn orbiting TOI-2328, whose transits events were detected in the lightcurves of the Transiting Exoplanet Survey Satellite \textbf{(TESS)} space mission. By combining TESS lightcurves with ground-based photometric and spectroscopic follow-up observations we confirm the planetary nature of the observed transits and radial velocity variations. TOI-6628~$b$ has a mass of 0.75$\pm$0.06~$M_\mathrm{J}$, a radius of 0.98$\pm$0.05~$R_J$ and is orbiting a metal-rich star with a period of 18.18424$\pm{0.00001}$ days and an eccentricity of 0.667$\pm0.016$, making it one of the most eccentric orbits of all known warm giants. TOI-3837~$b$ has a mass of 0.59$\pm$0.06~$M_\mathrm{J}$, a radius of 0.96$\pm$0.05~$R_J$ and orbits its host star every 11.88865$\pm$0.00003~days, with a moderate eccentricity of 0.198$^{+0.046}_{-0.058}$. With a mass of 2.01$\pm$0.13~$M_\mathrm{J}$ and a radius of 0.99$^{+0.07}_{-0.12}$ $R_J$, TOI-5027~$b$ orbits its host star in an eccentric orbit with $e$~=~0.395$^{+0.032}_{-0.029}$ every 10.24368$\pm{0.00001}$~days. TOI-2328~$b$ is a Saturn-like planet with a mass of 0.16$\pm$0.02~$M_\mathrm{J}$ and a radius of 0.89$\pm$0.04~$R_J$, orbiting its host star in a nearly circular orbit with $e$~=~0.057$^{+0.046}_{-0.029}$ at an orbital period of 17.10197$\pm{0.00001}$ days. All four planets have orbital periods above 10 days, and our planet interior structure models are consistsent a rocky-icy core with a H/He envelope, providing evidence supporting the core accretion model of planet formation for this kind of planets., Comment: 23 pages, 12 figures

We present a novel representation learning method for downstream tasks such as anomaly detection and unsupervised transient classification in high-energy datasets. This approach enabled the discovery of a new fast X-ray transient (FXT) in the Chandra archive, XRT 200515, a needle-in-the-haystack event and the first Chandra FXT of its kind. Recent serendipitous breakthroughs in X-ray astronomy, including FXTs from binary neutron star mergers and an extragalactic planetary transit candidate, highlight the need for systematic transient searches in X-ray archives. We introduce new event file representations, E-t Maps and E-t-dt Cubes, designed to capture both temporal and spectral information, effectively addressing the challenges posed by variable-length event file time series in machine learning applications. Our pipeline extracts low-dimensional, informative features from these representations using principal component analysis or sparse autoencoders, followed by clustering in the embedding space with DBSCAN. New transients are identified within transient-dominant clusters or through nearest-neighbor searches around known transients, producing a catalog of 3,539 candidates (3,427 flares and 112 dips). XRT 200515 exhibits unique temporal and spectral variability, including an intense, hard <10 s initial burst followed by spectral softening in an ~800 s oscillating tail. We interpret XRT 200515 as either the first giant magnetar flare observed at low X-ray energies or the first extragalactic Type I X-ray burst from a faint LMXB in the LMC. Our method extends to datasets from other observatories such as XMM-Newton, Swift-XRT, eROSITA, Einstein Probe, and upcoming missions like AXIS., Comment: 25 pages, submitted to Monthly Notices of the Royal Astronomical Society, presented at the 2023 Conference on Machine Learning in Astronomical Surveys (ML-IAP/CCA-2023)

Until about a decade ago, ring systems were only known to exist around giant planets. Each one of the four giant planets harbours its own distinctive and unique system of rings and inner satellites. The past decade has been marked by the unexpected discoveries of dense rings around small objects of the outer solar system: the Centaur object Chariklo (and possibly Chiron), the dwarf planet Haumea and the trans-Neptunian object Quaoar. In the latter case, an additional surprise came from the fact that Quaoar's rings orbit well beyond the Roche limit of the body. Here, we address the possible origins and evolution of these ring systems., Comment: 22 pages, 2 figures, 1 table, accepted for publication in Philosophical Transactions A

Emotions have a profound impact on our daily lives, influencing our thoughts, behaviors, and interactions, but also our physiological reactions. Recent advances in wearable technology have facilitated studying emotions through cardio-respiratory signals. Accelerometers offer a non-invasive, convenient, and cost-effective method for capturing heart- and pulmonary-induced vibrations on the chest wall, specifically Seismocardiography (SCG) and Accelerometry-Derived Respiration (ADR). Their affordability, wide availability, and ability to provide rich contextual data make accelerometers ideal for everyday use. While accelerometers have been used as part of broader modality fusions for Emotion Recognition (ER), their stand-alone potential via SCG and ADR remains unexplored. Bridging this gap could significantly help the embedding of ER into real-world applications. To address this gap, we introduce SCG as a novel modality for ER and evaluate its performance using the EmoWear dataset. First, we replicate the single-trial emotion classification pipeline from the DEAP dataset study, achieving similar results. Then we use our validated pipeline to train models that predict affective valence-arousal states using SCG and compare it against established cardiac signals, Electrocardiography (ECG) and Blood Volume Pulse (BVP). Results show that SCG is a viable modality for ER, achieving similar performance to ECG and BVP. By combining ADR with SCG, we achieved a working ER framework that only requires a single chest-worn accelerometer. These findings pave the way for integrating ER into real-world, enabling seamless affective computing in everyday life., Comment: 15 pages, 9 figures

Retrieval Augmented Generation (RAG) systems are a widespread application of Large Language Models (LLMs) in the industry. While many tools exist empowering developers to build their own systems, measuring their performance locally, with datasets reflective of the system's use cases, is a technological challenge. Solutions to this problem range from non-specific and cheap (most public datasets) to specific and costly (generating data from local documents). In this paper, we show that using public question and answer (Q&A) datasets to assess retrieval performance can lead to non-optimal systems design, and that common tools for RAG dataset generation can lead to unbalanced data. We propose solutions to these issues based on the characterization of RAG datasets through labels and through label-targeted data generation. Finally, we show that fine-tuned small LLMs can efficiently generate Q&A datasets. We believe that these observations are invaluable to the know-your-data step of RAG systems development., Comment: to be published in the 31st International Conference on Computational Linguistics (COLING 2025)

We give necessary and sufficient conditions on the curvature and the torsion of a regular curve of the space forms $\h^3$ and $\s^3$ to be contained in a totally umbilical surface. In case that the curve has constant torsion, we obtain the value of the curvature of the curve. Also numerical pictures of these curves are shown., Comment: 17 pages, 3 figures

Spacelike intrinsic rotational surfaces with constant mean curvature in the Lorentz-Minkowski space $\E_1^3$ have been recently investigated by Brander et al., extending the known Smyth's surfaces in Euclidean space. Assuming that the surface is intrinsic rotational with coordinates $(u,v)$ and conformal factor $\rho(u)^2$, we replace the constancy of the mean curvature with the property that the Weingarten endomorphism $A$ can be expressed as $\Phi_{-\alpha(v)}\left(\begin{array}{ll}\lambda_1(u)&0\\ 0&\lambda_2(u)\end{array}\right)\Phi_{\alpha(v)}$, where $\Phi_{\alpha(v)}$ is the (Euclidean or hyperbolic) rotation of angle $\alpha(v)$ at each tangent plane and $\lambda_i$ are the principal curvatures. Under these conditions, it is proved that the mean curvature is constant and $\alpha$ is a linear function. This result also covers the case that the surface is timelike. If the mean curvature is zero, we determine all spacelike and timelike intrinsic rotational surfaces with rotational angle $\alpha$. This family of surfaces includes the spacelike and timelike Enneper surfaces., Comment: 28 pages, 3 figures. Accepted to publish in Hacettepe Journal of Mathematics and Statistics

We introduce the Frenet theory of curves in dual space $\d^3$. After defining the curvature and the torsion of a curve, we classify all curves in dual plane with constant curvature. We also establish the fundamental theorem of existence in the theory of dual curves, proving that there is a dual curve with prescribed curvature and torsion. Finally we classify all dual curves with constant curvature and torsion., Comment: 14 pages, 3 figures

Gaia XP spectra for over two hundred million stars have great potential for mapping metallicity across the Milky Way. Several recent studies have analyzed this data set to derive parameters and characterize systematics in the fluxes. We aim to construct an alternative catalog of atmospheric parameters from Gaia XP spectra by fitting them with synthetic spectra based on model atmospheres, and provide corrections to the XP fluxes according to stellar colors, magnitudes, and extinction. We use GaiaXPy to obtain calibrated spectra and apply FERRE to match the corrected XP spectra with models and infer atmospheric parameters. We train a neural network using stars in APOGEE to predict flux corrections as a function of wavelength for each target. Based on the comparison with APOGEE parameters, we conclude that our estimated parameters have systematic errors and uncertainties in $T_{\mathrm{eff}}$, $\log g$, and [M/H] about $-38 \pm 167$ K, $0.05 \pm 0.40$ dex, and $-0.12 \pm 0.19$ dex, respectively, for stars in the range $4000 \le T_{\mathrm{eff}} \le 7000$ K. The corrected XP spectra show better agreement with both models and Hubble Space Telescope CALSPEC data. Our correction increases the precision of the relative spectrophotometry of the XP data from $3.2\% - 3.7\%$ to $1.2\% - 2.4\%$. Finally, we have built a catalog of atmospheric parameters for stars within $4000 \le T_{\mathrm{eff}} \le 7000$ K, comprising $68,394,431$ sources, along with a subset of $124,188$ stars with $\mathrm{[M/H]} \le -2.5$. Our results confirm that the Gaia XP flux calibrated spectra show systematic patterns as a function of wavelength that are tightly related to colors, magnitudes, and extinction. Our optimization algorithm can give us accurate atmospheric parameters of stars with a clear and direct link to models of stellar atmospheres, and can be used to efficiently search for extremely metal-poor stars., Comment: 16 pages, 11 figures, submitted to A&A, catalogs and code in https://doi.org/10.5281/zenodo.14028589

Long-timescale processes pose significant challenges in atomistic simulations, particularly for phenomena such as diffusion and phase transitions. We present a deep reinforcement learning (DRL)-based computational framework, combined with a temporal difference (TD) learning method, to simulate long-timescale atomic processes of diffusive relaxation. We apply it to study the emergence of chemical short-range order (SRO) in medium- and high-entropy alloys (MEAs/HEAs), which plays a crucial role in unlocking unique material properties, and find that the proposed method effectively maps the relationship between time, temperature, and SRO change. By accelerating both the sampling of lower-energy states and the simulation of transition kinetics, we identify the thermodynamic limit and the role of kinetic trapping in the SRO. Furthermore, learning the mean first passage time to a given, target SRO relaxation allows capturing realistic timescales in diffusive atomistic rearrangements. This method offers valuable guidelines for optimizing material processing and extends atomistic simulations to previously inaccessible timescales, facilitating the study of slow, thermally activated processes essential for understanding and engineering material properties.

The nature of sub-Neptunes remains unknown due to degeneracies in interior structure solutions. However, a statistical set of small planets with measured masses and radii has been compiled. It can be used to test the prediction of large water reservoirs on sub-Neptunes by planet formation theory. We want to find out whether this water reservoir is included in photoevaporative winds and how much of it can partition into the rocky and metallic interior. We couple the result of a planetary formation model to evolution models which assume perfect mixing of water with H/He in the envelope or complete segregation. For the mixed envelopes, we also include fractionation during photoevaporative mass-loss. Further, the effect of equilibrium dissolution of water into an assumed magma ocean and into the metallic core is studied for the first time in coupled formation-evolution models. Out of the tested scenarios, the mass-radius relation of exoplanets is best matched under the mixed assumption without water sequestration to the interior. We quantify the radius valley location and scaling with stellar mass. Fractionation is not found to significantly alter the composition of the planets for our initial conditions due to initially massive envelopes on all planets. In contrast, water sequestration has a profound effect on the radius evolution and compositional budget of the planets. The model predicts the preservation of large quantities of water even if the gaseous envelope is lost. Planets with corresponding bulk densities are not observed in comparably large numbers. By combining formation and evolution model, we probe a parameter space favored by core accretion theory. We conclude that the dissolution of different volatiles into the planetary interior and solidification of the magma ocean are natural next steps for comprehensive treatment of atmosphere-interior interaction. (abridged), Comment: 25 pages, 17 figures, submitted to A&A

Mitigating climate change requires transforming agriculture to minimize environ mental impact and build climate resilience. Regenerative agricultural practices enhance soil organic carbon (SOC) levels, thus improving soil health and sequestering carbon. A challenge to increasing regenerative agriculture practices is cheaply measuring SOC over time and understanding how SOC is affected by regenerative agricultural practices and other environmental factors and farm management practices. To address this challenge, we introduce an AI-driven Soil Organic Carbon Copilot that automates the ingestion of complex multi-resolution, multi-modal data to provide large-scale insights into soil health and regenerative practices. Our data includes extreme weather event data (e.g., drought and wildfire incidents), farm management data (e.g., cropland information and tillage predictions), and SOC predictions. We find that integrating public data and specialized models enables large-scale, localized analysis for sustainable agriculture. In comparisons of agricultural practices across California counties, we find evidence that diverse agricultural activity may mitigate the negative effects of tillage; and that while extreme weather conditions heavily affect SOC, composting may mitigate SOC loss. Finally, implementing role-specific personas empowers agronomists, farm consultants, policymakers, and other stakeholders to implement evidence-based strategies that promote sustainable agriculture and build climate resilience.

Sub-Neptunes - volatile-rich exoplanets smaller than Neptune - are intrinsically the most common type of planet known. However, the formation and nature of these objects, as well as the distinctions between sub-classes (if any), remain unclear. Two powerful tools to tease out the secrets of these worlds are measurements of (i) atmospheric composition and structure revealed by transit and/or eclipse spectroscopy, and (ii) mass, radius, and density revealed by transit photometry and Doppler spectroscopy. Here we present OrCAS, a survey to better elucidate the origins, compositions, and atmospheres of sub-Neptunes. This radial velocity survey uses a repeatable, quantifiable metric to select targets suitable for subsequent transmission spectroscopy and address key science themes about the atmospheric & internal compositions and architectures of these systems. Our survey targets 26 systems with transiting sub-Neptune planet candidates, with the overarching goal of increasing the sample of such planets suitable for subsequent atmospheric characterization. This paper lays out our survey's science goals, defines our target prioritization metric, and performs light-curve fits and statistical validation using existing TESS photometry and ground-based follow-up observations. Our survey serves to continue expanding the sample of small exoplanets with well-measured properties orbiting nearby bright stars, ensuring fruitful studies of these systems for many years to come., Comment: 20 pages, 4 figures, 4 tables, 26 sub-Neptunes, 31 TOIs. Accepted to AJ

Dipole-bound states in anions exist when a polar neutral core binds an electron in a diffuse orbital through charge--dipole interaction. Electronically excited polar neutral cores can also bind an electron in a diffuse orbital to form Core-Excited Dipole-Bound States (CE-DBSs), which are difficult to observe because they usually lie above the electron detachment threshold, leading to very short lifetimes and, thus, unstructured transitions. We report here the photodetachment spectroscopy of cryogenically cooled acetylacetonate anion (C$_5$H$_7$O$_2^-$) recorded by detecting the neutral radical produced upon photodetachment and the infrared spectroscopy in He-nanodroplets. Two DBSs were identified in this anion. One of them lies close to the electron detachment threshold ($\sim$2.74 eV) and is associated with the ground state of the radical (D0-DBS). Surprisingly, the other DBS appears as resonant transitions at 3.69 eV and is assigned to the CE-DBS associated with the first excited state of the radical (D1-DBS). It is proposed that the resonant transitions of the D1-DBS are observed $\sim$1 eV above the detachment threshold because its lifetime is determined by the internal conversion to the D0-DBS, after which the fast electron detachment takes place.

Autonomous driving needs good roads, but 85% of Brazilian roads have damages that deep learning models may not regard as most semantic segmentation datasets for autonomous driving are high-resolution images of well-maintained urban roads. A representative dataset for emerging countries consists of low-resolution images of poorly maintained roads and includes labels of damage classes; in this scenario, three challenges arise: objects with few pixels, objects with undefined shapes, and highly underrepresented classes. To tackle these challenges, this work proposes the Performance Increment Strategy for Semantic Segmentation (PISSS) as a methodology of 14 training experiments to boost performance. With PISSS, we reached state-of-the-art results of 79.8 and 68.8 mIoU on the Road Traversing Knowledge (RTK) and Technik Autonomer Systeme 500 (TAS500) test sets, respectively. Furthermore, we also offer an analysis of DeepLabV3+ pitfalls for small object segmentation.

Given a graph G=(V,E), a connected Grundy coloring is a proper vertex coloring that can be obtained by a first-fit heuristic on a connected vertex sequence. A first-fit coloring heuristic is one that attributes to each vertex in a sequence the lowest-index color not used for its preceding neighbors. A connected vertex sequence is one in which each element, except for the first one, is connected to at least one element preceding it. The connected Grundy coloring problem consists of obtaining a connected Grundy coloring maximizing the number of colors. In this paper, we propose two integer programming (IP) formulations and a local-search enhanced biased random-key genetic algorithm (BRKGA) for the connected Grundy coloring problem. The first formulation follows the standard way of partitioning the vertices into color classes while the second one relies on the idea of representatives in an attempt to break symmetries. The BRKGA encompasses a local search procedure using a newly proposed neighborhood. A theoretical neighborhood analysis is also presented. Extensive computational experiments indicate that the problem is computationally demanding for the proposed IP formulations. Nonetheless, the formulation by representatives outperforms the standard one for the considered benchmark instances. Additionally, our BRKGA can find high-quality solutions in low computational times for considerably large instances, showing improved performance when enhanced with local search and a reset mechanism. Moreover we show that our BRKGA can be easily extended to successfully tackle the Grundy coloring problem, i.e., the one without the connectivity requirements.

Cross-domain generalization is an open problem in WiFi-based sensing due to variations in environments, devices, and subjects, causing domain shifts in channel state information. To address this, we propose Domain-Adversarial Test-Time Adaptation (DATTA), a novel framework combining domain-adversarial training (DAT), test-time adaptation (TTA), and weight resetting to facilitate adaptation to unseen target domains and to prevent catastrophic forgetting. DATTA is integrated into a lightweight, flexible architecture optimized for speed. We conduct a comprehensive evaluation of DATTA, including an ablation study on all key components using publicly available data, and verify its suitability for real-time applications such as human activity recognition. When combining a SotA video-based variant of TTA with WiFi-based DAT and comparing it to DATTA, our method achieves an 8.1% higher F1-Score. The PyTorch implementation of DATTA is publicly available at: https://github.com/StrohmayerJ/DATTA.

Let $Y \to E \stackrel{p}{\to} B$ be a fibration and let $f: E \to E$ be a fiber map over $B$. In this work, we study the geometric and algebraic Reidemeister classes of the iterates of $f$ and introduce a Nielsen-type periodic number over $B$, denoted by $N_B P_n(f)$. When $B$ is a point, then $N_B P_n(f)$ coincides with the classical Nielsen periodic number.

Our model for the lifespan of an enterprise is the geometric distribution. We do not formulate a model for enterprise foundation, but assume that foundations and lifespans are independent. We aim to fit the model to information about foundation and closure of German enterprises in the AFiD panel. The lifespan for an enterprise that has been founded before the first wave of the panel is either left truncated, when the enterprise is contained in the panel, or missing, when it already closed down before the first wave. Marginalizing the likelihood to that part of the enterprise history after the first wave contributes to the aim of a closed-form estimate and standard error. Invariance under the foundation distribution is achived by conditioning on observability of the enterprises. The conditional marginal likelihood can be written as a function of a martingale. The later arises when calculating the compensator, with respect some filtration, of a process that counts the closures. The estimator itself can then also be written as a martingale transform and consistency as well as asymptotic normality are easily proven. The life expectancy of German enterprises, estimated from the demographic information about 1.4 million enterprises for the years 2018 and 2019, are ten years. The width of the confidence interval are two months. Closure after the last wave is taken into account as right censored., Comment: 42 pages, 2 figures, 4 tables

We study transitive partially hyperbolic diffeomorphisms in dimension 3 preserving a center foliation on which they act quasi-isometrically. We show that the diffeomorphism is up to finite lift and iterate, either a skew-product or a discretised Anosov flow., Comment: 13 pages, 2 figures

We study certain generic systems of real polynomial equations associated with triangulations of convex polytopes and investigate their number of real solutions. Our main focus is set on pairs of plane algebraic curves which form a so-called Wronski system. The computational tasks arising in the analysis of such Wronski pairs lead us to the frontiers of current computer algebra algorithms and their implementations, both via Gr\"obner bases and numerical algebraic geometry., Comment: 16 pages

Exascale computers offer transformative capabilities to combine data-driven and learning-based approaches with traditional simulation applications to accelerate scientific discovery and insight. However, these software combinations and integrations are difficult to achieve due to the challenges of coordinating and deploying heterogeneous software components on diverse and massive platforms. We present the ExaWorks project, which addresses many of these challenges. We developed a workflow Software Development Toolkit (SDK), a curated collection of workflow technologies that can be composed and interoperated through a common interface, engineered following current best practices, and specifically designed to work on HPC platforms. ExaWorks also developed PSI/J, a job management abstraction API, to simplify the construction of portable software components and applications that can be used over various HPC schedulers. The PSI/J API is a minimal interface for submitting and monitoring jobs and their execution state across multiple and commonly used HPC schedulers. We also describe several leading and innovative workflow examples of ExaWorks tools used on DOE leadership platforms. Furthermore, we discuss how our project is working with the workflow community, large computing facilities, and HPC platform vendors to address the requirements of workflows sustainably at the exascale.

Oriented object detection predicts orientation in addition to object location and bounding box. Precisely predicting orientation remains challenging due to angular periodicity, which introduces boundary discontinuity issues and symmetry ambiguities. Inspired by classical works on edge and corner detection, this paper proposes to represent orientation in oriented bounding boxes as a structure tensor. This representation combines the strengths of Gaussian-based methods and angle-coder solutions, providing a simple yet efficient approach that is robust to angular periodicity issues without additional hyperparameters. Extensive evaluations across five datasets demonstrate that the proposed structure tensor representation outperforms previous methods in both fully-supervised and weakly supervised tasks, achieving high precision in angular prediction with minimal computational overhead. Thus, this work establishes structure tensors as a robust and modular alternative for encoding orientation in oriented object detection. We make our code publicly available, allowing for seamless integration into existing object detectors.