Search

This paper discusses our recent generalized optimal algebraic multigrid (AMG) convergence theory applied to the steady-state Stokes equations discretized using Taylor-Hood elements ($\pmb{ \mathbb{P}}_2/\mathbb{P}_{1}$). The generalized theory is founded on matrix-induced orthogonality of the left and right eigenvectors of a generalized eigenvalue problem involving the system matrix and relaxation operator. This framework establishes a rigorous lower bound on the spectral radius of the two-grid error-propagation operator, enabling precise predictions of the convergence rate for symmetric indefinite problems, such as those arising from saddle-point systems. We apply this theory to the recently developed monolithic smooth aggregation AMG (SA-AMG) solver for Stokes, constructed using evolution-based strength of connection, standard aggregation, and smoothed prolongation. The performance of these solvers is evaluated using additive and multiplicative Vanka relaxation strategies. Additive Vanka relaxation constructs patches algebraically on each level, resulting in a nonsymmetric relaxation operator due to the partition of unity being applied on one side of the block-diagonal matrix. Although symmetry can be restored by eliminating the partition of unity, this compromises convergence. Alternatively, multiplicative Vanka relaxation updates velocity and pressure sequentially within each patch, propagating updates multiplicatively across the domain and effectively addressing velocity-pressure coupling, ensuring a symmetric relaxation. We demonstrate that the generalized optimal AMG theory consistently provides accurate lower bounds on the convergence rate for SA-AMG applied to Stokes equations. These findings suggest potential avenues for further enhancement in AMG solver design for saddle-point systems., Comment: 10 pages, 3 figures, student competition paper, 22nd Copper Mountain Conference On Multigrid Methods, 2025

The currently leading artificial neural network (ANN) models of the visual ventral stream -- which are derived from a combination of performance optimization and robustification methods -- have demonstrated a remarkable degree of behavioral alignment with humans on visual categorization tasks. Extending upon previous work, we show that not only can these models guide image perturbations that change the induced human category percepts, but they also can enhance human ability to accurately report the original ground truth. Furthermore, we find that the same models can also be used out-of-the-box to predict the proportion of correct human responses to individual images, providing a simple, human-aligned estimator of the relative difficulty of each image. Motivated by these observations, we propose to augment visual learning in humans in a way that improves human categorization accuracy at test time. Our learning augmentation approach consists of (i) selecting images based on their model-estimated recognition difficulty, and (ii) using image perturbations that aid recognition for novice learners. We find that combining these model-based strategies gives rise to test-time categorization accuracy gains of 33-72% relative to control subjects without these interventions, despite using the same number of training feedback trials. Surprisingly, beyond the accuracy gain, the training time for the augmented learning group was also shorter by 20-23%. We demonstrate the efficacy of our approach in a fine-grained categorization task with natural images, as well as tasks in two clinically relevant image domains -- histology and dermoscopy -- where visual learning is notoriously challenging. To the best of our knowledge, this is the first application of ANNs to increase visual learning performance in humans by enhancing category-specific features.

Studies of the functional role of the primate ventral visual stream have traditionally focused on object categorization, often ignoring -- despite much prior evidence -- its role in estimating "spatial" latents such as object position and pose. Most leading ventral stream models are derived by optimizing networks for object categorization, which seems to imply that the ventral stream is also derived under such an objective. Here, we explore an alternative hypothesis: Might the ventral stream be optimized for estimating spatial latents? And a closely related question: How different -- if at all -- are representations learned from spatial latent estimation compared to categorization? To ask these questions, we leveraged synthetic image datasets generated by a 3D graphic engine and trained convolutional neural networks (CNNs) to estimate different combinations of spatial and category latents. We found that models trained to estimate just a few spatial latents achieve neural alignment scores comparable to those trained on hundreds of categories, and the spatial latent performance of models strongly correlates with their neural alignment. Spatial latent and category-trained models have very similar -- but not identical -- internal representations, especially in their early and middle layers. We provide evidence that this convergence is partly driven by non-target latent variability in the training data, which facilitates the implicit learning of representations of those non-target latents. Taken together, these results suggest that many training objectives, such as spatial latents, can lead to similar models aligned neurally with the ventral stream. Thus, one should not assume that the ventral stream is optimized for object categorization only. As a field, we need to continue to sharpen our measures of comparing models to brains to better understand the functional roles of the ventral stream., Comment: 29 pages, 20 figures, ICLR 2025

Integrated silicon photonic devices, which manipulate light to transmit and process information on a silicon-on-insulator chip, are highly sensitive to structural variations. Minor deviations during nanofabrication-the precise process of building structures at the nanometer scale-such as over- or under-etching, corner rounding, and unintended defects, can significantly impact performance. To address these challenges, we introduce SEMU-Net, a comprehensive set of methods that automatically segments scanning electron microscope images (SEM) and uses them to train two deep neural network models based on U-Net and its variants. The predictor model anticipates fabrication-induced variations, while the corrector model adjusts the design to address these issues, ensuring that the final fabricated structures closely align with the intended specifications. Experimental results show that the segmentation U-Net reaches an average IoU score of 99.30%, while the corrector attention U-Net in a tandem architecture achieves an average IoU score of 98.67%., Comment: Accepted to WACV 2025

We study the maximum ratio of the Euclidean norm to the comass norm of p-covectors in Euclidean n-space and improve the known upper bound found in the standard references by Whitney and Federer. We go on to prove stable systolic inequalities when the fundamental cohomology class of the manifold is a cup product of forms of lower degree., Comment: 11 pages. To appear in Differential Geometry and Its Applications

We explore the structural and electronic properties of the bilayer nickelate La3Ni2O7 on LaAlO3(001) and SrTiO3(001) by using density functional theory including a Coulomb repulsion term. For La$_3$Ni$_2$O$_{7}$/LaAlO$_3$(001), we find that compressive strain and electron doping across the interface result in the unconventional occupation of the antibonding Ni $3d_{z^2}$ states. In sharp contrast, no charge transfer is observed for La$_3$Ni$_2$O$_{7}$/SrTiO$_3$(001). Surprisingly, tensile strain drives a metallization of the bonding Ni $3d_{z^2}$ states, rendering a Fermi surface topology resembling the superconducting bulk La$_3$Ni$_2$O$_{7}$ under high pressure. Concomitantly, significant octahedral rotations are retained. We discuss the fundamental differences between hydrostatic pressure and epitaxial strain regarding the crystal geometry and Ni $3d$ orbital occupation, and establish that strain provides a much stronger control over the Ni $e_g$ orbital polarization. The results suggest epitaxial La$_3$Ni$_2$O$_{7}$ under tensile strain as interesting system to provide novel insights into the physics of bilayer nickelates and possibly induce superconductivity without external pressure., Comment: 7 pages, 3 figures

In this paper, we consider the conditional generation problem by guiding off-the-shelf unconditional diffusion models with differentiable loss functions in a plug-and-play fashion. While previous research has primarily focused on balancing the unconditional diffusion model and the guided loss through a tuned weight hyperparameter, we propose a novel framework that distinctly decouples these two components. Specifically, we introduce two variables ${x}$ and ${z}$, to represent the generated samples governed by the unconditional generation model and the guidance function, respectively. This decoupling reformulates conditional generation into two manageable subproblems, unified by the constraint ${x} = {z}$. Leveraging this setup, we develop a new algorithm based on the Alternating Direction Method of Multipliers (ADMM) to adaptively balance these components. Additionally, we establish the equivalence between the diffusion reverse step and the proximal operator of ADMM and provide a detailed convergence analysis of our algorithm under certain mild assumptions. Our experiments demonstrate that our proposed method ADMMDiff consistently generates high-quality samples while ensuring strong adherence to the conditioning criteria. It outperforms existing methods across a range of conditional generation tasks, including image generation with various guidance and controllable motion synthesis.

With the advent of Large Language Models (LLMs) and Multimodal (Visio-lingual) LLMs, a flurry of research has emerged, analyzing the performance of such models across a diverse array of tasks. While most studies focus on evaluating the capabilities of state-of-the-art (SoTA) MLLM models through task accuracy (e.g., Visual Question Answering, grounding) across various datasets, our work explores the related but complementary aspect of consistency - the ability of an MLLM model to produce semantically similar or identical responses to semantically similar queries. We note that consistency is a fundamental prerequisite (necessary but not sufficient condition) for robustness and trust in MLLMs. Humans, in particular, are known to be highly consistent (even if not always accurate) in their responses, and consistency is inherently expected from AI systems. Armed with this perspective, we propose the MM-R$^3$ benchmark, which analyses the performance in terms of consistency and accuracy in SoTA MLLMs with three tasks: Question Rephrasing, Image Restyling, and Context Reasoning. Our analysis reveals that consistency does not always align with accuracy, indicating that models with higher accuracy are not necessarily more consistent, and vice versa. Furthermore, we propose a simple yet effective mitigation strategy in the form of an adapter module trained to minimize inconsistency across prompts. With our proposed strategy, we are able to achieve absolute improvements of 5.7% and 12.5%, on average on widely used MLLMs such as BLIP-2 and LLaVa 1.5M in terms of consistency over their existing counterparts.

This work develops a computational framework for proving existence, uniqueness, isolation, and stability results for real analytic fixed points of $m$-th order Feigenbaum-Cvitanovi\'{c} renormalization operators. Our approach builds on the earlier work of Lanford, Eckman, Wittwer, Koch, Burbanks, Osbaldestin, and Thurlby \cite{iii1982computer,eckmann1987complete,MR0727816, burbanks2021rigorous2,burbanks2021rigorous1}, however the main point of departure between ours and previous studies is that we discretize the domain of the renormalization operators using Chebyshev rather than Taylor series. The advantage of Chebyshev series is that they are naturally adapted to spaces of real analytic functions, in the sense that they converge on ellipses containing real intervals rather than on disks in $\mathbb{C}$. The main disadvantage of working with Chebyshev series in this context is that the essential operations of rescaling and composition are less straight forward for Chebysehv than for Taylor series. These difficulties are overcome using a combination of a-priori information about decay rates in the Banach space with a-posteriori estimates on Chebyshev interpolation errors for analytic functions. Our arguments are implemented in the Julia programming language and exploit extended precision floating point interval arithmetic. In addition to proving the existence of multiple renormalization fixed points of order $m = 3, \ldots, 10$, and computing validated bounds on the values of their the universal constants, we also reprove the existence of the classical $m=2$ Feigenbaum renormalization fixed point and compute its universal constants to close to 500 correct decimal digits.

This study reports a comprehensive environmental scan of the generative AI (GenAI) infrastructure in the national network for clinical and translational science across 36 institutions supported by the Clinical and Translational Science Award (CTSA) Program led by the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH) at the United States. With the rapid advancement of GenAI technologies, including large language models (LLMs), healthcare institutions face unprecedented opportunities and challenges. This research explores the current status of GenAI integration, focusing on stakeholder roles, governance structures, and ethical considerations by administering a survey among leaders of health institutions (i.e., representing academic medical centers and health systems) to assess the institutional readiness and approach towards GenAI adoption. Key findings indicate a diverse range of institutional strategies, with most organizations in the experimental phase of GenAI deployment. The study highlights significant variations in governance models, with a strong preference for centralized decision-making but notable gaps in workforce training and ethical oversight. Moreover, the results underscore the need for a more coordinated approach to GenAI governance, emphasizing collaboration among senior leaders, clinicians, information technology staff, and researchers. Our analysis also reveals concerns regarding GenAI bias, data security, and stakeholder trust, which must be addressed to ensure the ethical and effective implementation of GenAI technologies. This study offers valuable insights into the challenges and opportunities of GenAI integration in healthcare, providing a roadmap for institutions aiming to leverage GenAI for improved quality of care and operational efficiency.

The BICEP3 and BICEP Array polarimeters are small-aperture refracting telescopes located at the South Pole designed to measure primordial gravitational wave signatures in the Cosmic Microwave Background (CMB) polarization, predicted by inflation. Constraining the inflationary signal requires not only excellent sensitivity, but also careful control of instrumental systematics. Both instruments use antenna-coupled orthogonally polarized detector pairs, and the polarized sky signal is reconstructed by taking the difference in each detector pair. As a result, the differential response between detectors within a pair becomes an important systematic effect we must control. Additionally, mapping the intensity and polarization response in regions away from the main beam can inform how sidelobe levels affect CMB measurements. Extensive calibration measurements are taken in situ every austral summer for control of instrumental systematics and instrument characterisation. In this work, we detail the set of beam calibration measurements that we conduct on the BICEP receivers, from deep measurements of main beam response to polarized beam response and sidelobe mapping. We discuss the impact of these measurements for instrumental systematics studies and design choices for future CMB receivers., Comment: 13 pages, 7 figures, 1 table, Proceedings paper SPIE 2024

Reliable predictions of the etch rate profile are desirable in semiconductor manufacturing to prevent etch rate target misses and yield rate excursions. Conventional methods for analyzing etch rate require extensive metrology, which adds considerable costs to manufacturing. We demonstrate a data-driven method to predict the etch rate profiles of a capacitively-coupled plasma RIE etcher from the tool's sensor data. The model employs a hybrid autoencoder-multiquadric interpolation-based approach, with the autoencoder being used to encode the features of the wafers' etch rate profiles into a latent space representation. The tool's sensor data is then used to construct interpolation maps for the latent space variables using multiquadric radial basis functions, which are then used to generate synthetic wafer etch rate profiles using the decoder. The accuracy of the model is determined using experimental data, and the errors are analyzed in interpolation and extrapolation., Comment: 6 pages, 9 figures

Let $U_{n,d}$ be the uniform matroid of rank $d$ on $n$ elements. Denote by $g_{U_{n,d}}(t)$ the Speyer's $g$-polynomial of $U_{n,d}$. The Tur\'{a}n inequality and higher order Tur\'{a}n inequality are related to the Laguerre-P\'{o}lya ($\mathcal{L}$-$\mathcal{P}$) class of real entire functions, and the $\mathcal{L}$-$\mathcal{P}$ class has close relation with the Riemann hypothesis. The Tur\'{a}n type inequalities have received much attention. Infinite log-concavity is also a deep generalization of Tur\'{a}n inequality with different direction. In this paper, we mainly obtain the infinite log-concavity and the higher order Tur\'{a}n inequality of the sequence $\{g_{U_{n,d}}(t)\}_{d=1}^{n-1}$ for any $t>0$. In order to prove these results, we show that the generating function of $g_{U_{n,d}}(t)$, denoted $h_n(x;t)$, has only real zeros for $t>0$. Consequently, for $t>0$, we also obtain the $\gamma$-positivity of the polynomial $h_n(x;t)$, the asymptotical normality of $g_{U_{n,d}}(t)$, and the Laguerre inequalities for $g_{U_{n,d}}(t)$ and $h_n(x;t)$., Comment: 16 pages. arXiv admin note: text overlap with arXiv:2408.15506

Preparing students for both college and careers has been an explicit national policy objective for more than a decade. Recently, the "College for All" discourse prominent in the early 2000s has faded as more resources and attention have been focused on career-connected learning. New York City Public Schools (NYCPS) has been at the forefront of this shift, establishing a goal that all students have a "pathway to and preparation for a rewarding career and long-term economic security." As part of this push, NYCPS has expanded apprenticeship and internship opportunities and codified the expectation that every high school student receives individualized career and college advising and planning support. Despite the increased emphasis on college and career preparation, high school advising remains understudied, including in the context of formal career and technical education (CTE) programs. Together with partners from the CTE Research Network, the Research Alliance is engaged in a multi-state project that is investigating the role of advising in students' postsecondary transitions. The current report examines New York City high school students' perceptions of and experiences with college and career planning, the extent to which planning support differs across students and schools, and the link between planning support and short-term student outcomes. The study draws on the Senior Exit Survey administered in 2018, which was issued to only about half of NYC's high schools. Since that time, NYCPS has mandated the documentation of individualized advising plans for all high school students, meaning it is quite possible that students' experiences with advising have changed in recent years. Still, the authors believe their findings highlight challenges that will need to be surmounted to ensure all students are prepared for college and careers.

There is currently a heavy emphasis on career-connected learning for high school students in New York City. ExpandED Schools' science, technology, engineering and mathematics (STEM) Options (ES Options) program predates the newest of these initiatives. Launched in 2019, ES Options combines a credit-bearing STEM apprenticeship in the spring with a teaching internship in summer, designed to offer students hands-on learning experiences that increase their interest in STEM while exposing them to potential careers. Aspects of this model, including the provision of high school credit for a hybrid learning and work experience, have been adopted more broadly by New York City Public Schools and nonprofit organizations across the City. Insights into ES Options' successes, challenges, and outcomes are therefore valuable to inform the continued development and improvement of local programs as well as broader efforts in the field. In the previous report, "Recruiting and Retaining High School Students in Out-of-School Work-Based Learning Lessons from New York City," the Research Alliance described ES Options' implementation in 2022, highlighting notable challenges with program recruitment and persistence and outlining promising strategies to address these challenges. In this new report, patterns of program participation for a more recent cohort of students (2023) are examined, and analyses to investigate key student outcomes are extended. Building from the findings of those analyses, ExpandED piloted two new program models with the 2024 cohort. The first offers apprenticeships at schools during the school day as part of students' regular schedules. The second folds the apprenticeship into a six-week paid summer intensive, avoiding a key point of program attrition. Initial feedback about the in-school and summer intensive models' changes appears positive, and worth monitoring moving forward.

This report analyzes the potential for large language models (LLMs) to expedite accurate replication of published message effects studies. We tested LLM-powered participants (personas) by replicating 133 experimental findings from 14 papers containing 45 recent studies in the Journal of Marketing (January 2023-May 2024). We used a new software tool, Viewpoints AI (https://viewpoints.ai/), that takes study designs, stimuli, and measures as input, automatically generates prompts for LLMs to act as a specified sample of unique personas, and collects their responses to produce a final output in the form of a complete dataset and statistical analysis. The underlying LLM used was Anthropic's Claude Sonnet 3.5. We generated 19,447 AI personas to replicate these studies with the exact same sample attributes, study designs, stimuli, and measures reported in the original human research. Our LLM replications successfully reproduced 76% of the original main effects (84 out of 111), demonstrating strong potential for AI-assisted replication of studies in which people respond to media stimuli. When including interaction effects, the overall replication rate was 68% (90 out of 133). The use of LLMs to replicate and accelerate marketing research on media effects is discussed with respect to the replication crisis in social science, potential solutions to generalizability problems in sampling subjects and experimental conditions, and the ability to rapidly test consumer responses to various media stimuli. We also address the limitations of this approach, particularly in replicating complex interaction effects in media response studies, and suggest areas for future research and improvement in AI-assisted experimental replication of media effects., Comment: 24 pages, 3 figures, 2 tables

Feature selection is vital for identifying relevant variables in classification and regression models, especially in single-cell RNA sequencing (scRNA-seq) data analysis. Traditional methods like LASSO often struggle with the nonlinearities and multicollinearities in scRNA-seq data due to complex gene expression and extensive gene interactions. Quantum annealing, a form of quantum computing, offers a promising solution. In this study, we apply quantum annealing-empowered quadratic unconstrained binary optimization (QUBO) for feature selection in scRNA-seq data. Using data from a human cell differentiation system, we show that QUBO identifies genes with nonlinear expression patterns related to differentiation time, many of which play roles in the differentiation process. In contrast, LASSO tends to select genes with more linear expression changes. Our findings suggest that the QUBO method, powered by quantum annealing, can reveal complex gene expression patterns that traditional methods might overlook, enhancing scRNA-seq data analysis and interpretation.

CrMnFeCoNi, also called the Cantor alloy, is a well-known high-entropy alloy whose magnetic properties have recently become a focus of attention. We present a detailed muon spin relaxation study of the influence of chemical composition and sample processing protocols on the magnetic phase transitions and spin dynamics of several different Cantor alloy samples. Specific samples studied include a pristine equiatomic sample, samples with deficient and excess Mn content, and equiatomic samples magnetized in a field of 9 T or plastically deformed in pressures up to 0.5 GPa. The results confirm the sensitive dependence of the transition temperature on composition and demonstrate that post-synthesis pressure treatments cause the transition to become significantly less homogeneous throughout the sample volume. In addition, we observe critical spin dynamics in the vicinity of the transition in all samples, reminiscent of canonical spin glasses and magnetic materials with ideal continuous phase transitions. Application of an external magnetic field suppresses the critical dynamics in the Mn-deficient sample, while the equiatomic and Mn-rich samples show more robust critical dynamics. The spin-flip thermal activation energy in the paramagnetic phase increases with Mn content, ranging from $3.1(3) \times 10^{-21}$ J for 0% Mn to $1.2(2) \times 10^{-20}$ J for 30% Mn content. These results shed light on critical magnetic behavior in environments of extreme chemical disorder and demonstrate the tunability of spin dynamics in the Cantor alloy via chemical composition and sample processing.

Discriminating between Parkinson's Disease (PD) and Progressive Supranuclear Palsy (PSP) is difficult due to overlapping symptoms, especially early on. Saccades (rapid conjugate eye movements between fixation points) are affected by both diseases but conventional saccade analyses exhibit group level differences only. We hypothesized analyzing entire saccade raw time series waveforms would permit superior individual level discrimination between PD, PSP, and healthy controls (HC). 13,309 saccadic eye movements from 127 participants were analyzed using a novel, calibration-free waveform analysis and hierarchical machine learning framework. Individual saccades were classified based on which trained model could reconstruct each waveform with minimum error, indicating the most likely condition. A hierarchical classifier then predicted overall status (recently diagnosed and medication-naive 'de novo' PD, 'established' PD on antiparkinsonian medication, PSP, and healthy controls) by combining each participant's saccade results. This approach substantially outperformed conventional metrics, achieving high AUROCs distinguishing de novo PD from PSP (0.92-0.97), de novo PD from HC (0.72-0.89), and PSP from HC (0.90-0.95), while the conventional model showed limited performance (AUROC range: 0.45-0.75). This calibration-free waveform analysis sets a new standard for precise saccadic classification of PD, PSP, and HC, increasing potential for clinical adoption, remote monitoring, and screening., Comment: 2 table, 7 Figure, 27 pages

Large language model (LLM) chatbots are susceptible to biases and hallucinations, but current evaluations of mental wellness technologies lack comprehensive case studies to evaluate their practical applications. Here, we address this gap by introducing the MHealth-EVAL framework, a new role-play based interactive evaluation method designed specifically for evaluating the appropriateness, trustworthiness, and safety of mental wellness chatbots. We also introduce Psyfy, a new chatbot leveraging LLMs to facilitate transdiagnostic Cognitive Behavioral Therapy (CBT). We demonstrate the MHealth-EVAL framework's utility through a comparative study of two versions of Psyfy against standard baseline chatbots. Our results showed that Psyfy chatbots outperformed the baseline chatbots in delivering appropriate responses, engaging users, and avoiding untrustworthy responses. However, both Psyfy and the baseline chatbots exhibited some limitations, such as providing predominantly US-centric resources. While Psyfy chatbots were able to identify most unsafe situations and avoid giving unsafe responses, they sometimes struggled to recognize subtle harmful intentions when prompted in role play scenarios. Our study demonstrates a practical application of the MHealth-EVAL framework and showcases Psyfy's utility in harnessing LLMs to enhance user engagement and provide flexible and appropriate responses aligned with an evidence-based CBT approach.

The use of trajectory data with abundant spatial-temporal information is pivotal in Intelligent Transport Systems (ITS) and various traffic system tasks. Location-Based Services (LBS) capitalize on this trajectory data to offer users personalized services tailored to their location information. However, this trajectory data contains sensitive information about users' movement patterns and habits, necessitating confidentiality and protection from unknown collectors. To address this challenge, privacy-preserving methods like K-anonymity and Differential Privacy have been proposed to safeguard private information in the dataset. Despite their effectiveness, these methods can impact the original features by introducing perturbations or generating unrealistic trajectory data, leading to suboptimal performance in downstream tasks. To overcome these limitations, we propose a Federated Variational AutoEncoder (FedVAE) approach, which effectively generates a new trajectory dataset while preserving the confidentiality of private information and retaining the structure of the original features. In addition, FedVAE leverages Variational AutoEncoder (VAE) to maintain the original feature space and generate new trajectory data, and incorporates Federated Learning (FL) during the training stage, ensuring that users' data remains locally stored to protect their personal information. The results demonstrate its superior performance compared to other existing methods, affirming FedVAE as a promising solution for enhancing data privacy and utility in location-based applications., Comment: 2023 IEEE 98th Vehicular Technology Conference

Transformer-based language models have shown state-of-the-art performance on a variety of natural language understanding tasks. To achieve this performance, these models are first pre-trained on general corpus and then fine-tuned on downstream tasks. Previous work studied the effect of pruning the training set of the downstream tasks on the performance of the model on its evaluation set. In this work, we propose an automatic dataset pruning method for the training set of fine-tuning tasks. Our method is based on the model's success rate in correctly classifying each training data point. Unlike previous work which relies on user feedback to determine subset size, our method automatically extracts training subsets that are adapted for each pair of model and fine-tuning task. Our method provides multiple subsets for use in dataset pruning that navigate the trade-off between subset size and evaluation accuracy. Our largest subset, which we also refer to as the winning ticket subset, is on average $3 \times$ smaller than the original training set of the fine-tuning task. Our experiments on 5 downstream tasks and 2 language models show that, on average, fine-tuning on the winning ticket subsets results in a $0.1 \%$ increase in the evaluation performance of the model., Comment: 28 pages, 17 figures. Accepted at the Third Conference on Lifelong Learning Agents (CoLLAs 2024)

The existence of nontrivial cup products or Massey products in the cohomology of a manifold leads to inequalities of systolic type, but in general such inequalities are not optimal (tight). Gromov proved an {optimal} systolic inequality for complex projective space. We provide a natural extension of Gromov's inequality to manifolds whose fundamental cohomology class is a cup product of 2-dimensional classes., Comment: 8 pages, published in Journal of Geometry

Abdominal Aortic Aneurysms (AAA) are often repaired through an Endovascular approach known as EVAR. The success and duration of these challenging procedures are primarily attributable to the accuracy and reliability of navigating corresponding interventional devices. This study investigates the performance of conventional non-steerable and steerable catheters in endovascular aneurysm repair (EVAR) procedures, focusing on two primary metrics: reachable workspace and gate cannulation success. We developed two abdominal aortic aneurysm (AAA) phantoms using patient CT images for our experiments. Under X-ray fluoroscopy guidance, the reachable workspace was quantified, and gate cannulation success rates, cannulation time, and fluoroscopy times were recorded for both non-steerable and steerable catheters and were compared. We were unable to observe statistically significant differences between the two catheter types in overall cannulation success rates or fluoroscopy time. However, in challenging anatomical scenarios (particularly a more challenging gate location), the steerable catheter showed statistically significant advantages in success rates and cannulation times. While there were no statistical differences in reachable workspace between non-steerable and steerable catheters when considering the whole aneurysm, segmented analysis showed that the steerable catheter performed better in the central region, and non-steerable catheters performed better in the peripheral region. This study provides a systematic method for quantifying the performance of endovascular devices. The findings suggest that while steerable catheters may offer advantages in complex anatomical conditions, non-steerable catheters are preferable in peripheral areas of the aneurysm. These insights can inform catheter selection in EVAR, potentially influencing device design and clinical practice.

Fenestrated endovascular aneurysm repair remains a technically challenging procedure in the presence of complex anatomy, as it increases the difficulty of target vessel cannulation and prolongs procedure time and fluoroscopy radiation exposure. This paper aims to design, develop, and assess a novel steerable catheter, the SideEye, and compare its performance with conventional catheters in a thoracoabdominal aortic aneurysm phantom model. Users were asked to perform target vessel cannulation under fluoroscopic guidance using the SideEye and conventional non-steerable and steerable catheters. The experiment was divided into two cases based on the stent graft orientation (aligned and misaligned). Total procedure times, individual target vessel cannulation times, and exposure times were analyzed and compared in each case. In the misaligned case, the average cannulation times of all target vessels were $703 \pm 274$ s using the non-steerable catheter, $517 \pm 309$ s using the steerable catheter, and $199 \pm 91.0$ s using the SideEye. The average exposure times were $12 \pm 4.6$ min using the non-steerable catheter, $8.6 \pm 4.1$ min using the steerable catheter, and $3.0 \pm 1.1$ min using the SideEye. Target vessel cannulation using the SideEye significantly reduced procedure time and overall exposure time, compared to conventional devices.

We present the first high-resolution, high-frequency radio continuum survey that fully maps an extragalactic deep field: the 10GHz survey of the Great Observatories Origins Deep Survey-North (GOODS-N) field. This is a Large Program of the Karl G. Jansky Very Large Array that allocated 380 hours of observations using the X-band ($8-12$GHz) receivers, leading to a 10GHz mosaic of the GOODS-field with an average rms noise $\sigma_{\rm n}=671\,\rm nJy\,beam^{-1}$ and angular resolution $\theta_{1/2}=0.22$arcsec across 297$\rm arcmin^2$. To maximize the brightness sensitivity we also produce a low-resolution mosaic with $\theta_{1/2}=1.0$arcsec and $\sigma_{\rm n}=968\,\rm nJy\,beam^{-1}$, from which we derive our master catalog containing 256 radio sources detected with peak signal-to-noise ratio $\geq 5$. Radio source size and flux density estimates from the high-resolution mosaic are provided in the master catalog as well. The total fraction of spurious sources in the catalog is 0.75%. Monte Carlo simulations are performed to derive completeness corrections of the catalog. We find that the 10GHz radio source counts in the GOODS-N field agree, in general, with predictions from numerical simulations/models and expectations from 1.4 and 3GHz radio counts., Comment: Accepted for publication in ApJ

The Boros-Moll sequences $\{d_\ell(m)\}_{\ell=0}^m$ arise in the study of evaluation of a quartic integral. After the infinite log-concavity conjecture of the sequence $\{d_\ell(m)\}_{\ell=0}^m$ was proposed by Boros and Moll, a lot of interesting inequalities on $d_\ell(m)$ were obtained, although the conjecture is still open. Since $d_\ell(m)$ has two parameters, it is natural to consider the properties for the sequences $\{d_\ell(m)\}_{m\ge \ell}$, which are called the \emph{transposed Boros-Moll sequences} here. In this paper, we mainly prove the extended reverse ultra log-concavity of the transposed Boros-Moll sequences $\{d_\ell(m)\}_{m\ge \ell}$, and hence give an upper bound for the ratio ${d_\ell^2(m)}/{(d_\ell(m-1)d_\ell(m+1))}$. A lower bound for this ratio is also established which implies a result stronger than the log-concavity of the sequences $\{d_\ell(m)\}_{m\ge \ell}$. As a consequence, we also show that the transposed Boros-Moll sequences possess a stronger log-concave property than the Boros-Moll sequences do. At last, we propose some conjectures on the Boros-Moll sequences and their transposes., Comment: 15 pages

The search for the polarized imprint of primordial gravitational waves in the cosmic microwave background (CMB) as direct evidence of cosmic inflation requires exquisite sensitivity and control over systematics. The next-generation CMB-S4 project intends to improve upon current-generation experiments by deploying a significantly greater number of highly-sensitive detectors, combined with refined instrument components based on designs from field-proven instruments. The Precursor Small Aperture Telescope (PreSAT) is envisioned as an early step to this next generation, which will test prototype CMB-S4 components and technologies within an existing BICEP Array receiver, with the aim of enabling full-stack laboratory testing and early risk retirement, along with direct correlation of laboratory component-level performance measurements with deployed system performance. The instrument will utilize new 95/155GHz dichroic dual-linear-polarization prototype detectors developed for CMB-S4, cooled to 100mK via the installation of an adiabatic demagnetization refrigerator, along with a prototype readout chain and prototype optics manufactured with wide-band anti-reflection coatings. The experience gained by integrating, deploying, and calibrating PreSAT will also help inform planning for CMB-S4 small aperture telescope commissioning, calibration, and operations well in advance of the fabrication of CMB-S4 production hardware., Comment: 12 pages, 4 figures, submitted to Proc. SPIE

Designing 3D indoor layouts is a crucial task with significant applications in virtual reality, interior design, and automated space planning. Existing methods for 3D layout design either rely on diffusion models, which utilize spatial relationship priors, or heavily leverage the inferential capabilities of proprietary Large Language Models (LLMs), which require extensive prompt engineering and in-context exemplars via black-box trials. These methods often face limitations in generalization and dynamic scene editing. In this paper, we introduce LLplace, a novel 3D indoor scene layout designer based on lightweight fine-tuned open-source LLM Llama3. LLplace circumvents the need for spatial relationship priors and in-context exemplars, enabling efficient and credible room layout generation based solely on user inputs specifying the room type and desired objects. We curated a new dialogue dataset based on the 3D-Front dataset, expanding the original data volume and incorporating dialogue data for adding and removing objects. This dataset can enhance the LLM's spatial understanding. Furthermore, through dialogue, LLplace activates the LLM's capability to understand 3D layouts and perform dynamic scene editing, enabling the addition and removal of objects. Our approach demonstrates that LLplace can effectively generate and edit 3D indoor layouts interactively and outperform existing methods in delivering high-quality 3D design solutions. Code and dataset will be released.

Planets orbiting binary systems are relatively unexplored compared to those around single stars. Detections of circumbinary planets and planetary systems offer a first detailed view into our understanding of circumbinary planet formation and dynamical evolution. The BEBOP (Binaries Escorted by Orbiting Planets) radial velocity survey plays a special role in this adventure as it focuses on eclipsing single-lined binaries with an FGK dwarf primary and M dwarf secondary allowing for the highest-radial velocity precision using the HARPS and SOPHIE spectrographs. We obtained 4512 high-resolution spectra for the 179 targets in the BEBOP survey which we used to derive the stellar atmospheric parameters using both equivalent widths and spectral synthesis. We furthermore derive stellar masses, radii, and ages for all targets. With this work, we present the first homogeneous catalogue of precise stellar parameters for these eclipsing single-lined binaries., Comment: accepted for publication in MNRAS

Despite the ubiquity of ice, a molecular-resolution image of ice crystallized from liquid water or the resulting defect structure has never been obtained. Here, we report the stabilization and angstrom-resolution electron imaging of ice Ih crystallized from liquid water by developing cryogenic liquid-cell transmission electron microscopy (CRYOLIC-TEM). We combine lattice mapping with molecular dynamics simulations to reveal that ice formation is highly tolerant to nanoscale defects such as misoriented subdomains and trapped gas bubbles, which are stabilized by molecular-scale structural motifs. Importantly, bubble surfaces adopt low-energy nanofacets and create negligible strain fields in the surrounding crystal. These bubbles can dynamically nucleate, grow, migrate, dissolve, and coalesce under electron irradiation and be monitored in situ near a steady state. This work opens the door to understanding water crystallization behaviors at an unprecedented spatial resolution., Comment: 26 pages, 5 figures, and 1 table. Supplementary Information: 35 pages, 37 figures, and 9 tables

Located at the bottom of the main sequence, ultracool dwarf stars are widespread in the solar neighbourhood. Nevertheless, their extremely low luminosity has left their planetary population largely unexplored, and only one of them, TRAPPIST-1, has so far been found to host a transiting planetary system. In this context, we present the SPECULOOS project's detection of an Earth-sized planet in a 17 h orbit around an ultracool dwarf of M6.5 spectral type located 16.8 pc away. The planet's high irradiation (16 times that of Earth) combined with the infrared luminosity and Jupiter-like size of its host star make it one of the most promising rocky exoplanet targets for detailed emission spectroscopy characterization with JWST. Indeed, our sensitivity study shows that just ten secondary eclipse observations with the Mid-InfraRed Instrument/Low-Resolution Spectrometer on board JWST should provide strong constraints on its atmospheric composition and/or surface mineralogy.