A controversial test for Large Language Models concerns the ability to discern possible from impossible language. While some evidence attests to the models' sensitivity to what crosses the limits of grammatically impossible language, this evidence has been contested on the grounds of the soundness of the testing material. We use model-internal representations to tap directly into the way Large Language Models represent the 'grammatical-ungrammatical' distinction. In a novel benchmark, we elicit probabilities from 4 models and compute minimal-pair surprisal differences, juxtaposing probabilities assigned to grammatical sentences to probabilities assigned to (i) lower frequency grammatical sentences, (ii) ungrammatical sentences, (iii) semantically odd sentences, and (iv) pragmatically odd sentences. The prediction is that if string-probabilities can function as proxies for the limits of grammar, the ungrammatical condition will stand out among the conditions that involve linguistic violations, showing a spike in the surprisal rates. Our results do not reveal a unique surprisal signature for ungrammatical prompts, as the semantically and pragmatically odd conditions consistently show higher surprisal. We thus demonstrate that probabilities do not constitute reliable proxies for model-internal representations of syntactic knowledge. Consequently, claims about models being able to distinguish possible from impossible language need verification through a different methodology.

We refine and extend a recent construction of sets of black hole microstates with semiclassical interiors that span a Hilbert space of dimension $e^S$, where $S$ is the black hole entropy. We elaborate on the definition and properties of microstates in statistical and black hole mechanics. The gravitational description of microstates employs matter shells in the interior of the black hole, and we argue that in the limit where the shells are very heavy, the construction acquires universal validity. To this end, we show it for very wide classes of black holes: we first extend the construction to rotating and charged black holes, including extremal and near-extremal solutions, with or without supersymmetry, and we sketch how the construction of microstates can be embedded in String Theory. We then describe how the approach can include general quantum corrections, near or far from extremality. For supersymmetric black holes, the microstates we construct differ from other recent constructions in that the interior excitations are not confined within the near-extremal throat.

Test-time adaptation enables a trained model to adjust to a new domain during inference, making it particularly valuable in clinical settings where such on-the-fly adaptation is required. However, existing techniques depend on large target domain datasets, which are often impractical and unavailable in medical scenarios that demand per-patient, real-time inference. Moreover, current methods commonly focus on two-dimensional images, failing to leverage the volumetric richness of medical imaging data. Bridging this gap, we propose a Patch-Based Multi-View Co-Training method for Single Image Test-Time adaptation. Our method enforces feature and prediction consistency through uncertainty-guided self-training, enabling effective volumetric segmentation in the target domain with only a single test-time image. Validated on three publicly available breast magnetic resonance imaging datasets for tumor segmentation, our method achieves performance close to the upper bound supervised benchmark while also outperforming all existing state-of-the-art methods, on average by a Dice Similarity Coefficient of 3.75%. We publicly share our accessible codebase, readily integrable with the popular nnUNet framework, at this https URL.

Gravitational waves from black-hole merging events have revealed a population of extra-galactic BHs residing in short-period binaries with masses that are higher than expected based on most stellar evolution models - and also higher than known stellar-origin black holes in our Galaxy. It has been proposed that those high-mass BHs are the remnants of massive metal-poor stars. Gaia astrometry is expected to uncover many Galactic wide-binary systems containing dormant BHs, which may not have been detected before. The study of this population will provide new information on the BH-mass distribution in binaries and shed light on their formation mechanisms and progenitors. As part of the validation efforts in preparation for the fourth Gaia data release (DR4), we analysed the preliminary astrometric binary solutions, obtained by the Gaia Non-Single Star pipeline, to verify their significance and to minimise false-detection rates in high-mass-function orbital solutions. The astrometric binary solution of one source, Gaia BH3, implies the presence of a 32.70 \pm 0.82 M\odot BH in a binary system with a period of 11.6 yr. Gaia radial velocities independently validate the astrometric orbit. Broad-band photometric and spectroscopic data show that the visible component is an old, very metal-poor giant of the Galactic halo, at a distance of 590 pc. The BH in the Gaia BH3 system is more massive than any other Galactic stellar-origin BH known thus far. The low metallicity of the star companion supports the scenario that metal-poor massive stars are progenitors of the high-mass BHs detected by gravitational-wave telescopes. The Galactic orbit of the system and its metallicity indicate that it might belong to the Sequoia halo substructure. Alternatively, and more plausibly, it could belong to the ED-2 stream, which likely originated from a globular cluster that had been disrupted by the Milky Way.

We implement Crossing Statistics to reconstruct in a model-agnostic manner the expansion history of the universe and properties of dark energy, using DESI Data Release 1 (DR1) BAO data in combination with one of three different supernova compilations (PantheonPlus, Union3, and DES-SN5YR) and Planck CMB observations. Our results hint towards an evolving and emergent dark energy behaviour, with negligible presence of dark energy at $z\gtrsim 1$, at varying significance depending on the data sets combined. In all these reconstructions, the cosmological constant lies outside the $95\%$ confidence intervals for some redshift ranges. This dark energy behaviour, reconstructed using Crossing Statistics, is in agreement with results from the conventional $w_0$--$w_a$ dark energy equation of state parametrization reported in the DESI Key cosmology paper. Our results add an extensive class of model-agnostic reconstructions with acceptable fits to the data, including models where cosmic acceleration slows down at low redshifts. We also report constraints on $H_0r_d$ from our model-agnostic analysis, independent of the pre-recombination physics.

Africa faces significant challenges in healthcare delivery due to limited infrastructure and access to advanced medical technologies. This study explores the use of federated learning to overcome these barriers, focusing on perinatal health. We trained a fetal plane classifier using perinatal data from five African countries: Algeria, Ghana, Egypt, Malawi, and Uganda, along with data from Spanish hospitals. To incorporate the lack of computational resources in the analysis, we considered a heterogeneous set of devices, including a Raspberry Pi and several laptops, for model training. We demonstrate comparative performance between a centralized and a federated model, despite the compute limitations, and a significant improvement in model generalizability when compared to models trained only locally. These results show the potential for a future implementation at a large scale of a federated learning platform to bridge the accessibility gap and improve model generalizability with very little requirements.

Spatially resolved stellar kinematics has become a key ingredient in time-delay cosmography to break the mass-sheet degeneracy in the mass profile and, in turn, provide a precise constraint on the Hubble constant and other cosmological parameters. In this paper, we present the first measurements of 2D resolved stellar kinematics for the lens galaxy in the quadruply lensed quasar system \lensname, using integral field spectroscopy from the JWST's Near-Infrared Spectrograph (NIRSpec), marking the first such measurement conducted with the JWST. In extracting robust kinematic measurements from this first-of-its-kind dataset, we made methodological improvements both in the data reduction and kinematic extraction. In our kinematic extraction procedure, we performed joint modeling of the lens galaxy, the quasar, and its host galaxy's contributions in the spectra to deblend the lens galaxy component and robustly constrain its stellar kinematics. Our improved methodological frameworks are released as software pipelines for future use: \textsc{squirrel} for extracting stellar kinematics, and \textsc{RegalJumper} for JWST-NIRSpec data reduction, incorporating additional artifact cleaning beyond the standard JWST pipeline. We compared our measured stellar kinematics from the JWST NIRSpec with previously obtained ground-based measurements from the Keck Cosmic Web Imager integral field unit and find that the two datasets are statistically consistent at a $\sim$1.1$\sigma$ confidence level. Our measured kinematics will be used in a future study to improve the precision of the Hubble constant measurement.

Most papers caution against using predictive models for disease stratification based on unselected radiomic features, as these features are affected by contouring variability. Instead, they advocate for the use of the Intraclass Correlation Coefficient (ICC) as a measure of stability for feature selection. However, the direct effect of segmentation variability on the predictive models is rarely studied. This study investigates the impact of segmentation variability on feature stability and predictive performance in radiomics-based prediction of Triple-Negative Breast Cancer (TNBC) subtype using Magnetic Resonance Imaging. A total of 244 images from the Duke dataset were used, with segmentation variability introduced through modifications of manual segmentations. For each mask, explainable radiomic features were selected using the Shapley Additive exPlanations method and used to train logistic regression models. Feature stability across segmentations was assessed via ICC, Pearson's correlation, and reliability scores quantifying the relationship between feature stability and segmentation variability. Results indicate that segmentation accuracy does not significantly impact predictive performance. While incorporating peritumoral information may reduce feature reproducibility, it does not diminish feature predictive capability. Moreover, feature selection in predictive models is not inherently tied to feature stability with respect to segmentation, suggesting that an overreliance on ICC or reliability scores for feature selection might exclude valuable predictive features.

In this work, we introduce a quantitative methodology to define what is the main trunk and what are the significant branches of a minimum spanning tree (MST). We apply it to the pulsar tree, i.e. the MST of the pulsar population constructed upon a Euclidean distance over the pulsar's intrinsic properties. Our method makes use of the betweenness centrality estimator, as well as of non-parametric tests to establish the distinct character of the defined branches. Armed with these concepts, we study how the pulsar population has evolved throughout history, and analyze how to judge whether a new class of pulsars appears in new data, future surveys, or new incarnations of pulsar catalogs.

A class of viable $F(R)$ gravity models which can provide a unified description of inflation with the dark energy era is confronted with the latest observational data on the dark energy era. These models have the unique characteristic that the de Sitter scalaron mass in the Einstein frame counterpart theory is a monotonic function of the curvature, which renders them viable descriptions for both the inflationary and the late-time acceleration eras. We also compare these models with other well-known viable $F(R)$ gravity models and with the $\Lambda$-Cold-Dark-Matter model. As we show, the most phenomenologically successful models are those which deviate significantly from the $\Lambda$-Cold-Dark-Matter model. Also some of the models presented, provide a statistically favorable description of the dark energy eras, compared with the exponential $F(R)$ gravity model and of course compared with the $\Lambda$-Cold-Dark-Matter model. All the models we present in this article are confronted with the observational data from the Planck collaboration, the Pantheon plus data from Type Ia supernovae, the two rounds of observations of the Dark Energy Spectroscopic Instrument, data from baryon acoustic oscillations and the Hubble constant measurements by SH0ES group. As we show, two of the models are statistically favorable by the data.

The segmentation of the pubic symphysis and fetal head (PSFH) constitutes a pivotal step in monitoring labor progression and identifying potential delivery complications. Despite the advances in deep learning, the lack of annotated medical images hinders the training of segmentation. Traditional semi-supervised learning approaches primarily utilize a unified network model based on Convolutional Neural Networks (CNNs) and apply consistency regularization to mitigate the reliance on extensive annotated data. However, these methods often fall short in capturing the discriminative features of unlabeled data and in delineating the long-range dependencies inherent in the ambiguous boundaries of PSFH within ultrasound images. To address these limitations, we introduce a novel framework, the Dual-Student and Teacher Combining CNN and Transformer (DSTCT), which synergistically integrates the capabilities of CNNs and Transformers. Our framework comprises a Vision Transformer (ViT) as the teacher and two student mod ls one ViT and one CNN. This dual-student setup enables mutual supervision through the generation of both hard and soft pseudo-labels, with the consistency in their predictions being refined by minimizing the classifier determinacy discrepancy. The teacher model further reinforces learning within this architecture through the imposition of consistency regularization constraints. To augment the generalization abilities of our approach, we employ a blend of data and model perturbation techniques. Comprehensive evaluations on the benchmark dataset of the PSFH Segmentation Grand Challenge at MICCAI 2023 demonstrate our DSTCT framework outperformed ten contemporary semi-supervised segmentation methods. Code available at this https URL.

Dynamic contrast-enhanced (DCE) MRI is essential for breast cancer diagnosis and treatment. However, its reliance on contrast agents introduces safety concerns, contraindications, increased cost, and workflow complexity. To this end, we present pre-contrast conditioned denoising diffusion probabilistic models to synthesize DCE-MRI, introducing, evaluating, and comparing a total of 22 generative model variants in both single-breast and full breast settings. Towards enhancing lesion fidelity, we introduce both tumor-aware loss functions and explicit tumor segmentation mask conditioning. Using a public multicenter dataset and comparing to respective pre-contrast baselines, we observe that subtraction image-based models consistently outperform post-contrast-based models across five complementary evaluation metrics. Apart from assessing the entire image, we also separately evaluate the region of interest, where both tumor-aware losses and segmentation mask inputs improve evaluation metrics. The latter notably enhance qualitative results capturing contrast uptake, albeit assuming access to tumor localization inputs that are not guaranteed to be available in screening settings. A reader study involving 2 radiologists and 4 MRI technologists confirms the high realism of the synthetic images, indicating an emerging clinical potential of generative contrast-enhancement. We share our codebase at this https URL.

Synthetic contrast enhancement offers fast image acquisition and eliminates the need for intravenous injection of contrast agent. This is particularly beneficial for breast imaging, where long acquisition times and high cost are significantly limiting the applicability of magnetic resonance imaging (MRI) as a widespread screening modality. Recent studies have demonstrated the feasibility of synthetic contrast generation. However, current state-of-the-art (SOTA) methods lack sufficient measures for consistent temporal evolution. Neural cellular automata (NCA) offer a robust and lightweight architecture to model evolving patterns between neighboring cells or pixels. In this work we introduce TeNCA (Temporal Neural Cellular Automata), which extends and further refines NCAs to effectively model temporally sparse, non-uniformly sampled imaging data. To achieve this, we advance the training strategy by enabling adaptive loss computation and define the iterative nature of the method to resemble a physical progression in time. This conditions the model to learn a physiologically plausible evolution of contrast enhancement. We rigorously train and test TeNCA on a diverse breast MRI dataset and demonstrate its effectiveness, surpassing the performance of existing methods in generation of images that align with ground truth post-contrast sequences.

We present HST2EUCLID, a novel Python code to generate Euclid realistic mock images in the $H_{\rm E}$, $J_{\rm E}$, $Y_{\rm E}$, and $I_{\rm E}$ photometric bands based on panchromatic Hubble Space Telescope observations. The software was used to create a simulated database of Euclid images for the 27 galaxy clusters observed during the Cluster Lensing And Supernova survey with Hubble (CLASH) and the Hubble Frontier Fields (HFF) program. Since the mock images were generated from real observations, they incorporate, by construction, all the complexity of the observed galaxy clusters. The simulated Euclid data of the galaxy cluster MACS J0416.1$-$2403 were then used to explore the possibility of developing strong lensing models based on the Euclid data. In this context, complementary photometric or spectroscopic follow-up campaigns are required to measure the redshifts of multiple images and cluster member galaxies. By Euclidising six parallel blank fields obtained during the HFF program, we provide an estimate of the number of galaxies detectable in Euclid images per ${\rm deg}^2$ per magnitude bin (number counts) and the distribution of the galaxy sizes. Finally, we present a preview of the Chandra Deep Field South that will be observed during the Euclid Deep Survey and two examples of galaxy-scale strong lensing systems residing in regions of the sky covered by the Euclid Wide Survey. The methodology developed in this work lends itself to several additional applications, as simulated Euclid fields based on HST (or JWST) imaging with extensive spectroscopic information can be used to validate the feasibility of legacy science cases or to train deep learning techniques in advance, thus preparing for a timely exploitation of the Euclid Survey data.

We present the MARD-Y3 catalog of between 1086 and 2171 galaxy clusters (52\% and 65\% new) produced using multi-component matched filter (MCMF) followup in 5000\,deg$^2$ of DES-Y3 optical data of the $\sim$20000 overlapping 2RXS X-ray sources. Optical counterparts are identified as peaks in galaxy richness as a function of redshift along the line of sight toward each 2RXS source within a search region informed by an X-ray prior. All peaks are assigned a probability \fcont\ of being a random superposition. The clusters lie at $0.020.5$. Residual contamination is 2.6\% and 9.6\% for the cuts adopted here. For each cluster we present the optical center, redshift, rest frame X-ray luminosity, $M_{500}$ mass, coincidence with NWAY infrared sources and estimators of dynamical state. About 2\% of MARD-Y3 clusters have multiple possible counterparts, the photo-z's are high quality with $\sigma_{\Delta z/(1+z)}=0.0046$, and $\sim$1\% of clusters exhibit evidence of X-ray luminosity boosting from emission by cluster AGN. Comparison with other catalogs (MCXC, RM, SPT-SZ, Planck) is performed to test consistency of richness, luminosity and mass estimates. We measure the MARD-Y3 X-ray luminosity function and compare it to the expectation from a fiducial cosmology and externally calibrated luminosity- and richness-mass relations. Agreement is good, providing evidence that MARD-Y3 has low contamination and can be understood as a simple two step selection-- X-ray and then optical-- of an underlying cluster population described by the halo mass function.

Gaia Data Release 3 (DR3) provides a wealth of new data products for the astronomical community to exploit, including astrophysical parameters for a half billion stars. In this work we demonstrate the high quality of these data products and illustrate their use in different astrophysical contexts. We query the astrophysical parameter tables along with other tables in Gaia DR3 to derive the samples of the stars of interest. We validate our results by using the Gaia catalogue itself and by comparison with external data. We have produced six homogeneous samples of stars with high quality astrophysical parameters across the HR diagram for the community to exploit. We first focus on three samples that span a large parameter space: young massive disk stars (~3M), FGKM spectral type stars (~3M), and UCDs (~20K). We provide these sources along with additional information (either a flag or complementary parameters) as tables that are made available in the Gaia archive. We furthermore identify 15740 bone fide carbon stars, 5863 solar-analogues, and provide the first homogeneous set of stellar parameters of the Spectro Photometric Standard Stars. We use a subset of the OBA sample to illustrate its usefulness to analyse the Milky Way rotation curve. We then use the properties of the FGKM stars to analyse known exoplanet systems. We also analyse the ages of some unseen UCD-companions to the FGKM stars. We additionally predict the colours of the Sun in various passbands (Gaia, 2MASS, WISE) using the solar-analogue sample.

We consider higher-derivative quantum gravity where renormalization group improved effective action beyond one-loop approximation is derived. Using this effective action, the quantum-corrected FRW equations are analyzed. De Sitter universe solution is found. It is demonstrated that such de Sitter inflationary universe is instable. The slow-roll inflationary parameters are calculated. The contribution of renormalization group improved Gauss-Bonnet term to quantum-corrected FRW equations as well as to instability of de Sitter universe is estimated. It is demonstrated that in this case the spectral index and tensor-to-scalar ratio are consistent with Planck data.

In this study, we present ScoreFormer, a novel graph transformer model designed to accurately predict molecular docking scores, thereby optimizing high-throughput virtual screening (HTVS) in drug discovery. The architecture integrates Principal Neighborhood Aggregation (PNA) and Learnable Random Walk Positional Encodings (LRWPE), enhancing the model's ability to understand complex molecular structures and their relationship with their respective docking scores. This approach significantly surpasses traditional HTVS methods and recent Graph Neural Network (GNN) models in both recovery and efficiency due to a wider coverage of the chemical space and enhanced performance. Our results demonstrate that ScoreFormer achieves competitive performance in docking score prediction and offers a substantial 1.65-fold reduction in inference time compared to existing models. We evaluated ScoreFormer across multiple datasets under various conditions, confirming its robustness and reliability in identifying potential drug candidates rapidly.