Search

As software pervades more and more areas of our professional and personal lives, there is an ever-increasing need to maintain software, and for programmers to be able to efficiently write and understand program code. In the first study of its kind, we analyze fixation-related potentials (FRPs) to explore the online processing of program code patterns that are ambiguous to programmers, but not the computer (so-called atoms of confusion), and their underlying neurocognitive mechanisms in an ecologically valid setting. Relative to unambiguous counterparts in program code, atoms of confusion elicit a late frontal positivity with a duration of about 400 to 700 ms after first looking at the atom of confusion. As the frontal positivity shows high resemblance with an event-related potential (ERP) component found during natural language processing that is elicited by unexpected but plausible words in sentence context, we take these data to suggest that the brain engages similar neurocognitive mechanisms in response to unexpected and informative inputs in program code and in natural language. In both domains, these inputs lead to an update of a comprehender's situation model that is essential for information extraction from a quickly unfolding input.

Letters of recommendation are a common tool used in graduate admissions. Most admissions systems require three letters for each applicant, burdening both letter writers and admissions committees with a heavy work load that may not be time well-spent. Most applicants do not have three research advisors who can comment meaningfully on research readiness, adding a large number of letters that are not useful. Ideally, letters of recommendation will showcase the students' promise for a research career, but in practice, the letters often do not fulfill this purpose. As a group of early and mid-career faculty who write dozens of letters every year for promising undergraduates, we are concerned and overburdened by the inefficiencies of the current system. In this open letter to the AAS Graduate Admissions Task Force, we offer an alternative to the current use of letters of recommendation: a portfolio submitted by the student, which highlights e.g., a paper, plot, or presentation that represents their past work and readiness for grad school, uploaded to a centralized system used by astronomy and astrophysics PhD programs. While we argue that we could eliminate letters in this new paradigm, it may instead be advisable to limit the number of letters of recommendation to one per applicant., Comment: Accepted for publication to the Bulletin of the AAS

Optical coating, an integral part of many optical systems, is prone to damage from environmental exposure and laser irradiation. This underscores the need for reliable and sensitive coating diagnostics. We introduce second harmonic generation (SHG) as a method for the sensitive detection of defects and inhomogeneities within optical thin films. We demonstrate the use of SHG on Si3N4 layers tested for their laser-induced damage threshold. The SHG mapping was compared with commonly used diagnostic techniques, including Nomarski microscopy, white light interferometry, and electron microscopy. Owing to its sensitivity to variations in local symmetry, SHG is able to discern minute alterations in material composition, mechanical stress, and interface structures. Therefore, SHG identified modifications in the tested layers, highly extending the damage recognized by standard methods. Furthermore, we demonstrate SHG's ability to enhance specific features by modulation of incidence angles and polarization modes. Based on these findings, we propose SHG as an ideal diagnostic tool for the early identification of laser-induced modifications in centrosymmetric thin films.

Non-verbal communication often comprises of semantically rich gestures that help convey the meaning of an utterance. Producing such semantic co-speech gestures has been a major challenge for the existing neural systems that can generate rhythmic beat gestures, but struggle to produce semantically meaningful gestures. Therefore, we present RAG-Gesture, a diffusion-based gesture generation approach that leverages Retrieval Augmented Generation (RAG) to produce natural-looking and semantically rich gestures. Our neuro-explicit gesture generation approach is designed to produce semantic gestures grounded in interpretable linguistic knowledge. We achieve this by using explicit domain knowledge to retrieve exemplar motions from a database of co-speech gestures. Once retrieved, we then inject these semantic exemplar gestures into our diffusion-based gesture generation pipeline using DDIM inversion and retrieval guidance at the inference time without any need of training. Further, we propose a control paradigm for guidance, that allows the users to modulate the amount of influence each retrieval insertion has over the generated sequence. Our comparative evaluations demonstrate the validity of our approach against recent gesture generation approaches. The reader is urged to explore the results on our project page., Comment: Preprint. Project page: https://vcai.mpi-inf.mpg.de/projects/RAG-Gesture/

The Steiner Forest problem is an important generalization of the Steiner Tree problem. We are given an undirected graph with nonnegative edge costs and a collection of pairs of vertices. The task is to compute a cheapest forest with the property that the elements of each pair belong to the same connected component of the forest. The current best approximation factor for Steiner Forest is 2, which is achieved by the classical primal-dual algorithm; improving on this factor is a big open problem in the area. Motivated by this open problem, we study an LP relaxation for Steiner Forest that generalizes the well-studied Bidirected Cut Relaxation for Steiner Tree. We prove that this relaxation has several promising properties. Among them, it is possible to round any half-integral LP solution to a Steiner Forest instance while increasing the cost by at most a factor 16/9. To prove this result we introduce a novel recursive densest-subgraph contraction algorithm.

In multimedia broadcasting, no-reference image quality assessment (NR-IQA) is used to indicate the user-perceived quality of experience (QoE) and to support intelligent data transmission while optimizing user experience. This paper proposes an improved no-reference light field image quality assessment (NR-LFIQA) metric for future immersive media broadcasting services. First, we extend the concept of depthwise separable convolution (DSC) to the spatial domain of light field image (LFI) and introduce "light field depthwise separable convolution (LF-DSC)", which can extract the LFI's spatial features efficiently. Second, we further theoretically extend the LF-DSC to the angular space of LFI and introduce the novel concept of "light field anglewise separable convolution (LF-ASC)", which is capable of extracting both the spatial and angular features for comprehensive quality assessment with low complexity. Third, we define the spatial and angular feature estimations as auxiliary tasks in aiding the primary NR-LFIQA task by providing spatial and angular quality features as hints. To the best of our knowledge, this work is the first exploration of deep auxiliary learning with spatial-angular hints on NR-LFIQA. Experiments were conducted in mainstream LFI datasets such as Win5-LID and SMART with comparisons to the mainstream full reference IQA metrics as well as the state-of-the-art NR-LFIQA methods. The experimental results show that the proposed metric yields overall 42.86% and 45.95% smaller prediction errors than the second-best benchmarking metric in Win5-LID and SMART, respectively. In some challenging cases with particular distortion types, the proposed metric can reduce the errors significantly by more than 60%.

Retrieval Augmented Generation (RAG) enables Large Language Models (LLMs) to generalize to new information by decoupling reasoning capabilities from static knowledge bases. Traditional RAG enhancements have explored vertical scaling -- assigning subtasks to specialized modules -- and horizontal scaling -- replicating tasks across multiple agents -- to improve performance. However, real-world applications impose diverse Service Level Agreements (SLAs) and Quality of Service (QoS) requirements, involving trade-offs among objectives such as reducing cost, ensuring answer quality, and adhering to specific operational constraints. In this work, we present a systems-oriented approach to multi-agent RAG tailored for real-world Question Answering (QA) applications. By integrating task-specific non-functional requirements -- such as answer quality, cost, and latency -- into the system, we enable dynamic reconfiguration to meet diverse SLAs. Our method maps these Service Level Objectives (SLOs) to system-level parameters, allowing the generation of optimal results within specified resource constraints. We conduct a case study in the QA domain, demonstrating how dynamic re-orchestration of a multi-agent RAG system can effectively manage the trade-off between answer quality and cost. By adjusting the system based on query intent and operational conditions, we systematically balance performance and resource utilization. This approach allows the system to meet SLOs for various query types, showcasing its practicality for real-world applications.

This paper investigates the geometric structure of a quasigeostrophic approximation to a recently introduced reduced-gravity thermal rotating shallow-water model that accounts for stratification. Specifically, it considers a low-frequency approximation of a model for flow above the ocean thermocline, governed by primitive equations with buoyancy variations in both horizontal and vertical directions. Like the thermal model, the stratified variant generates circulation patterns reminiscent of submesoscale instabilities visible in satellite images. An improvement is its ability to model mixed-layer restratification due to baroclinic instability. The primary contribution of this paper is to demonstrate that the model is derived from an Euler--Poincar\'e variational principle, culminating in a Kelvin--Noether theorem, previously established solely for the primitive-equation parent model. The model's Lie--Poisson Hamiltonian structure, earlier obtained through direct calculation, is shown to result from a Legendre transform with the associated geometry elucidated by identifying the relevant momentum map. Another significant contribution of this paper is the identification of the Casimirs of the Lie--Poisson system, including a newly found weaker Casimir family forming the kernel of the Lie--Poisson bracket, which results in potential vorticity evolution independent of buoyancy details as it advects under the flow. These conservation laws related to particle relabeling symmetry are explicitly linked to Noether quantities from the Euler--Poincar\'e principle when variations are not constrained to vanish at integration endpoints. The dual Euler--Poincar\'e/Lie--Poisson formalism provides a unified framework for describing quasigeostrophic reduced-gravity stratified thermal flow, mirroring the approach used in the primitive-equation setting., Comment: Submitted to AIP Journal of Mathematical Physics. Comments welcome!

We study the $k$-center problem in the context of individual fairness. Let $P$ be a set of $n$ points in a metric space and $r_x$ be the distance between $x \in P$ and its $\lceil n/k \rceil$-th nearest neighbor. The problem asks to optimize the $k$-center objective under the constraint that, for every point $x$, there is a center within distance $r_x$. We give bicriteria $(\beta,\gamma)$-approximation algorithms that compute clusterings such that every point $x \in P$ has a center within distance $\beta r_x$ and the clustering cost is at most $\gamma$ times the optimal cost. Our main contributions are a deterministic $O(n^2+ kn \log n)$ time $(2,2)$-approximation algorithm and a randomized $O(nk\log(n/\delta)+k^2/\varepsilon)$ time $(10,2+\varepsilon)$-approximation algorithm, where $\delta$ denotes the failure probability. For the latter, we develop a randomized sampling procedure to compute constant factor approximations for the values $r_x$ for all $x\in P$ in subquadratic time; we believe this procedure to be of independent interest within the context of individual fairness.

We report on an educational pilot program for low-cost physics experimentation run in Ecuador, South Africa, and the United States. The program was developed after having needs-based discussions with African educators, researchers, and leaders. It was determined that the need and desire for low-cost, skills-building, and active-learning tools is very high. From this, we developed a 3D-printable, Raspberry Pi-based multispectral camera (15 to 25 spectral channels in the visible and near-IR) for as little as $100. The program allows students to learn 3D modeling, 3D printing, feedback, control, image analysis, Python programming, systems integration and artificial intelligence as well as spectroscopy. After completing their cameras, the students in the program studied plant health, plant stress, post-harvest fruit ripeness, and polarization and spectral analysis of nanostructured insect wings, the latter of which won the ``best-applied research" award at a conference poster session and will be highlighted in this paper. Importantly, these cameras can be an integral part of any developing country's agricultural, recycling, medical, and pharmaceutical infrastructure. Thus, we believe this experiment can play an important role at the intersection of student training and developing countries' capacity building.

Tangible User Interfaces (TUI) for human--computer interaction (HCI) provide the user with physical representations of digital information with the aim to overcome the limitations of screen-based interfaces. Although many compelling demonstrations of TUIs exist in the literature, there is a lack of research on TUIs intended for daily two-handed tasks and processes, such as cooking. In response to this gap, we propose SPICE (Smart Projection Interface for Cooking Enhancement). SPICE investigates TUIs in a kitchen setting, aiming to transform the recipe following experience from simply text-based to tangibly interactive. SPICE includes a tracking system, an agent-based software, and vision large language models to create and interpret a kitchen environment where recipe information is projected directly onto the cooking surface. We conducted a comparative usability study of SPICE and text-based recipe following with 30 participants, assessing the task difficulty, total duration, and efficiency, as well as user confidence and taste perception. The results indicate that SPICE allowed participants to perform the recipe with less stops and in shorter time while also improving self-reported efficiency, confidence, and taste. Despite this, participants self-reported no change in overall difficulty, which is a direction for future research. Overall, the SPICE project demonstrates the potential of using TUIs to improve everyday activities, paving the way for future research in HCI and new computing interfaces., Comment: Article submitted to IUI 2025

Synthetic data generation with Large Language Models is a promising paradigm for augmenting natural data over a nearly infinite range of tasks. Given this variety, direct comparisons among synthetic data generation algorithms are scarce, making it difficult to understand where improvement comes from and what bottlenecks exist. We propose to evaluate algorithms via the makeup of synthetic data generated by each algorithm in terms of data quality, diversity, and complexity. We choose these three characteristics for their significance in open-ended processes and the impact each has on the capabilities of downstream models. We find quality to be essential for in-distribution model generalization, diversity to be essential for out-of-distribution generalization, and complexity to be beneficial for both. Further, we emphasize the existence of Quality-Diversity trade-offs in training data and the downstream effects on model performance. We then examine the effect of various components in the synthetic data pipeline on each data characteristic. This examination allows us to taxonomize and compare synthetic data generation algorithms through the components they utilize and the resulting effects on data QDC composition. This analysis extends into a discussion on the importance of balancing QDC in synthetic data for efficient reinforcement learning and self-improvement algorithms. Analogous to the QD trade-offs in training data, often there exist trade-offs between model output quality and output diversity which impact the composition of synthetic data. We observe that many models are currently evaluated and optimized only for output quality, thereby limiting output diversity and the potential for self-improvement. We argue that balancing these trade-offs is essential to the development of future self-improvement algorithms and highlight a number of works making progress in this direction.

Synthetic data is gaining increasing popularity for face recognition technologies, mainly due to the privacy concerns and challenges associated with obtaining real data, including diverse scenarios, quality, and demographic groups, among others. It also offers some advantages over real data, such as the large amount of data that can be generated or the ability to customize it to adapt to specific problem-solving needs. To effectively use such data, face recognition models should also be specifically designed to exploit synthetic data to its fullest potential. In order to promote the proposal of novel Generative AI methods and synthetic data, and investigate the application of synthetic data to better train face recognition systems, we introduce the 2nd FRCSyn-onGoing challenge, based on the 2nd Face Recognition Challenge in the Era of Synthetic Data (FRCSyn), originally launched at CVPR 2024. This is an ongoing challenge that provides researchers with an accessible platform to benchmark i) the proposal of novel Generative AI methods and synthetic data, and ii) novel face recognition systems that are specifically proposed to take advantage of synthetic data. We focus on exploring the use of synthetic data both individually and in combination with real data to solve current challenges in face recognition such as demographic bias, domain adaptation, and performance constraints in demanding situations, such as age disparities between training and testing, changes in the pose, or occlusions. Very interesting findings are obtained in this second edition, including a direct comparison with the first one, in which synthetic databases were restricted to DCFace and GANDiffFace.

In this paper, we complete the classification of 4 x 4 solutions of the Yang-Baxter equation. Regular solutions were recently classified and in this paper we find the remaining non-regular solutions. We present several new solutions, then consider regular and non-regular Lax operators and study their relation to the quantum Yang-Baxter equation. We show that for regular solutions there is a correspondence, which is lost in the non-regular case. In particular, we find non-regular Lax operators whose R-matrix from the fundamental commutation relations is regular but does not satisfy the Yang-Baxter equation. These R-matrices satisfy a modified Yang-Baxter equation instead.

Regression model have a substantial impact on interpretation of treatments, genetic characteristics and other covariates in survival analysis. In many datasets, the description of censoring and survival curve reveals the presence of cure fraction on data, which leads to alternative modelling. The most common approach to introduce covariates under a parametric estimation are the cure rate models and their variations, although the use of defective distributions have introduced a more parsimonious and integrated approach. Defective distributions is given by a density function whose integration is not one after changing the domain of one the parameters. In this work, we introduce two new defective regression models for long-term survival data in the Marshall-Olkin family: the Marshall-Olkin Gompertz and the Marshall-Olkin inverse Gaussian. The estimation process is conducted using the maximum likelihood estimation and Bayesian inference. We evaluate the asymptotic properties of the classical approach in Monte Carlo studies as well as the behavior of Bayes estimates with vague information. The application of both models under classical and Bayesian inferences is provided in an experiment of time until death from colon cancer with a dichotomous covariate. The Marshall-Olkin Gompertz regression presented the best adjustment and we present some global diagnostic and residual analysis for this proposal.

In cancer research, the term epigenetics was used in the 1970s in its modern sense encompassing non-genetic events modifying the chromatin state, mainly to oppose the emerging oncogene paradigm. However, starting from the establishment of this prominent concept, the importance of these epigenetic phenomena in cancer rarely led to questioning the causal role of genetic alterations. Only in the last 10 years, the accumulation of problematic data, better experimental technologies, and some ambitious models pushed the idea that epigenetics could be at least as important as genetics in early oncogenesis. Until this year, a direct demonstration of epigenetic oncogenesis was still lacking. Now Parreno, Cavalli and colleagues, using a refined experimental model in the fruit fly Drosophila melanogaster, enforced the initiation of tumours solely by imposing a transient loss of Polycomb repression, leading to a purely epigenetic oncogenesis phenomenon. Despite a few caveats that we discuss, this pioneering work represents a major breakpoint in cancer research that leads us to consider the theoretical and conceptual implications on oncogenesis and to search for links between this artificial experimental model and naturally occurring processes, while revisiting cancer theories that were previously proposed as alternatives to the oncogene-centered paradigm.

Given the rising proliferation and diversity of AI writing assistance tools, especially those powered by large language models (LLMs), both writers and readers may have concerns about the impact of these tools on the authenticity of writing work. We examine whether and how writers want to preserve their authentic voice when co-writing with AI tools and whether personalization of AI writing support could help achieve this goal. We conducted semi-structured interviews with 19 professional writers, during which they co-wrote with both personalized and non-personalized AI writing-support tools. We supplemented writers' perspectives with opinions from 30 avid readers about the written work co-produced with AI collected through an online survey. Our findings illuminate conceptions of authenticity in human-AI co-creation, which focus more on the process and experience of constructing creators' authentic selves. While writers reacted positively to personalized AI writing tools, they believed the form of personalization needs to target writers' growth and go beyond the phase of text production. Overall, readers' responses showed less concern about human-AI co-writing. Readers could not distinguish AI-assisted work, personalized or not, from writers' solo-written work and showed positive attitudes toward writers experimenting with new technology for creative writing.

We use the Monte Carlo For AGN (active galactic nucleus) Channel Testing and Simulation (McFACTS, https://www.github.com/mcfacts/mcfacts) code to study the effect of AGN disk and nuclear star cluster parameters on predicted mass distributions for LIGO-Virgo-KAGRA (LVK) compact binaries forming in AGN disks. The assumptions we vary include the black hole (BH) initial mass function, disk model, disk size, disk lifetime, and the prograde-to-retrograde fraction of newly formed black hole binaries. Broadly we find that dense, moderately short-lived AGN disks are preferred for producing a $(q,\chi_{\rm eff})$ anti-correlation like those identified from existing gravitational wave (GW) observations. Additionally, a BH initial mass function (MF $\propto M^{-2}$) is preferred over a more top-heavy MF ($M^{-1}$). The preferred fraction of prograde-to-retrograde is $>90\%$, to produce results consistent with observations., Comment: Submitted to ApJ, 18 pages, 9 figures, 2 tables

Remote sensing (RS) involves the acquisition of data about objects or areas from a distance, primarily to monitor environmental changes, manage resources, and support planning and disaster response. A significant challenge in RS segmentation is the scarcity of high-quality labeled images due to the diversity and complexity of RS image, which makes pixel-level annotation difficult and hinders the development of effective supervised segmentation algorithms. To solve this problem, we propose Adaptively Augmented Consistency Learning (AACL), a semi-supervised segmentation framework designed to enhances RS segmentation accuracy under condictions of limited labeled data. AACL extracts additional information embedded in unlabeled images through the use of Uniform Strength Augmentation (USAug) and Adaptive Cut-Mix (AdaCM). Evaluations across various RS datasets demonstrate that AACL achieves competitive performance in semi-supervised segmentation, showing up to a 20% improvement in specific categories and 2% increase in overall performance compared to state-of-the-art frameworks.

The influence of growth regimes of the silicon capping layer on the optical properties of heterostructures with submonolayer InAs quantum dots embedded in a silicon matrix has been studied. The photoluminescence signal at 1650 nm from submonolayer quantum dots at low temperatures up to 120 K was obtained. It was established that the use of a two-stage method of silicon overgrowing InAs nanoislands makes it possible to increase the photoluminescence intensity by improving the crystalline quality of heterostructures. Analysis of the temperature dependence allowed us to calculate an activation energy for electrons confined in the quantum dots potential well at the level of the thermal energy at room temperature., Comment: 22 pages, in Russian language, 4 figures

This paper discusses the effect of measurement errors in the estimation of the carbon dioxide (CO$_2$) airborne fraction. We are the first to present regression-based estimates and standard errors that are robust to measurement errors for the extended model, the preferred specification to estimate the CO$_2$ airborne fraction. To achieve this goal, we add to the literature in three ways: $i)$ We generalise the Deming regression to handle multiple variables. $ii)$ We introduce a bootstrap approach to construct confidence intervals for Deming regression in both univariate and multivariate scenarios. $iii)$ Propose to estimate the airborne fraction using instrumental variables (IV), taking advantage of the variation of additional measurements, to obtain consistent estimates that are robust to measurement errors. IV estimates for the airborne fraction are 44.8%($\pm$ 1.4%; 1$\sigma$) for the simple specification, and 47.3%($\pm$ 1.1%; 1$\sigma$) for the extended specification. We show that these estimates are not statistically different from the ordinary least squares (OLS) estimates, while being robust to measurement errors without relying on additional assumptions. In contrast, OLS estimates are shown to fall outside the confidence interval of the Deming regression estimates., Comment: 13 pages, 2 figures for associated supplementary material, see https://github.com/everval/Robust-CO2-Estimation-Supplementary

The widespread use of large language models (LLMs) has dramatically improved many applications of Natural Language Processing (NLP), including Information Retrieval (IR). However, domains that are not driven by commercial interest often lag behind in benefiting from AI-powered solutions. One such area is religious and heritage corpora. Alongside similar domains, Islamic literature holds significant cultural value and is regularly utilized by scholars and the general public. Navigating this extensive amount of text is challenging, and there is currently no unified resource that allows for easy searching of this data using advanced AI tools. This work focuses on the development of a multilingual non-profit IR system for the Islamic domain. This process brings a few major challenges, such as preparing multilingual domain-specific corpora when data is limited in certain languages, deploying a model on resource-constrained devices, and enabling fast search on a limited budget. By employing methods like continued pre-training for domain adaptation and language reduction to decrease model size, a lightweight multilingual retrieval model was prepared, demonstrating superior performance compared to larger models pre-trained on general domain data. Furthermore, evaluating the proposed architecture that utilizes Rust Language capabilities shows the possibility of implementing efficient semantic search in a low-resource setting.

Optoacoustic tomography image reconstruction has been a problem of interest in recent years. By exploiting the exceptional generative power of the recently proposed diffusion models we consider a scheme which is based on a conditional diffusion process. Using a simple initial image reconstruction method such as Delay and Sum, we consider a specially designed autoencoder architecture which generates a latent representation which is used as conditional information in the generative diffusion process. Numerical results show the merits of our proposal in terms of quality metrics such as PSNR and SSIM, showing that the conditional information generated in terms of the initial reconstructed image is able to bias the generative process of the diffusion model in order to enhance the image, correct artifacts and even recover some finer details that the initial reconstruction method is not able to obtain., Comment: Published in 2024 IEEE Biennial Congress of Argentina. arXiv admin note: substantial text overlap with arXiv:2404.10239

We present $24$ cosmological dark matter (DM)--only zoom-in simulations of a Milky Way (MW) analog with initial conditions appropriate for scenarios where non-cold DM is a fraction of the total DM abundance (f-NCDM models), as the second installment of the COZMIC suite. We initialize our simulations using transfer functions, $T_{\mathrm{f-NCDM}}(k)\equiv\sqrt{P_{\mathrm{f-NCDM}}(k)/P_{\mathrm{CDM}}(k)}$ (where $P(k)$ is the linear matter power spectrum), with an initial suppression similar to thermal-relic warm DM (WDM) followed by a constant-amplitude plateau. We simulate suppression wave numbers $[22.8,~ 32.1,~ 41.8,~ 52.0,~ 57.1,~ 95.3]~\mathrm{Mpc}^{-1}$, corresponding to thermal-relic WDM masses $m_{\mathrm{WDM}}\in [3,~ 4,~ 5,~ 6,~ 6.5,~ 10]~\mathrm{keV}$, and plateau amplitudes $\delta\in [0.2,~ 0.4,~ 0.6,~ 0.8]$. We model the subhalo mass function in terms of the suppression wave number and $\delta$. Integrating these models into a forward model of the MW satellite galaxy population yields new limits on f-NCDM scenarios, with suppression wave numbers greater than $46$ and $ 40~\mathrm{Mpc}^{-1}$ for $\delta=0.2$, $0.4$, respectively, at $95\%$ confidence. The current data do not constrain $\delta>0.4$. We map these limits to scenarios where a fraction $f_{\mathrm{WDM}}$ of DM behaves as a thermal relic, which yields the following bounds on cosmologies with a mixture of WDM and CDM: $m_{\mathrm{WDM}}>3.6,~ 4.1,~ 4.6,~ 4.9,~ 5.4~\mathrm{keV}$ for $f_{\mathrm{WDM}}=0.5,~ 0.6,~ 0.7,~ 0.8,~ 0.9$, respectively, at $95\%$ confidence. The current data do not constrain WDM fractions $f_{\mathrm{WDM}}<0.5$. Our results affirm that low-mass halo abundances are sensitive to partial suppression in $P(k)$, indicating the possibility of using galactic substructure to reconstruct $P(k)$ on small scales., Comment: 19 pages, 14 figures, 1 table; comments welcome

Large-scale cell microscopy screens are used in drug discovery and molecular biology research to study the effects of millions of chemical and genetic perturbations on cells. To use these images in downstream analysis, we need models that can map each image into a feature space that represents diverse biological phenotypes consistently, in the sense that perturbations with similar biological effects have similar representations. In this work, we present the largest foundation model for cell microscopy data to date, a new 1.9 billion-parameter ViT-G/8 MAE trained on over 8 billion microscopy image crops. Compared to a previous published ViT-L/8 MAE, our new model achieves a 60% improvement in linear separability of genetic perturbations and obtains the best overall performance on whole-genome biological relationship recall and replicate consistency benchmarks. Beyond scaling, we developed two key methods that improve performance: (1) training on a curated and diverse dataset; and, (2) using biologically motivated linear probing tasks to search across each transformer block for the best candidate representation of whole-genome screens. We find that many self-supervised vision transformers, pretrained on either natural or microscopy images, yield significantly more biologically meaningful representations of microscopy images in their intermediate blocks than in their typically used final blocks. More broadly, our approach and results provide insights toward a general strategy for successfully building foundation models for large-scale biological data., Comment: NeurIPS 2024 Foundation Models for Science Workshop (38th Conference on Neural Information Processing Systems). 18 pages, 7 figures

When conducting a survey, many choices regarding survey design features have to be made. These choices affect the response rate of a survey. This paper analyzes the individual effects of these survey design features on the response rate. For this purpose, data from a systematic review of crime surveys conducted in Germany between 2001--2021 were used. First, a meta-analysis of proportions is used to estimate the summary response rate. Second, a meta-regression was fitted, modeling the relationship between the observed response rates and survey-design features, such as the study year, target population, coverage area, data collection mode, and institute. The developed model informs about the influence of certain survey design features and can predict the expected response rate when (re-) designing a survey. This study highlights that a thoughtful survey design and professional survey administration can result in high response rates.

We present 850 $\mu$m polarization observations of the IC 348 star-forming region in the Perseus molecular cloud as part of the B-fields In STar-forming Region Observation (BISTRO) survey. We study the magnetic properties of two cores (HH 211 MMS and IC 348 MMS) and a filamentary structure of IC 348. We find that the overall field tends to be more perpendicular than parallel to the filamentary structure of the region. The polarization fraction decreases with intensity, and we estimate the trend by power-law and the mean of the Rice distribution fittings. The power indices for the cores are much smaller than 1, indicative of possible grain growth to micron size in the cores. We also measure the magnetic field strengths of the two cores and the filamentary area separately by applying the Davis-Chandrasekhar-Fermi method and its alternative version for compressed medium. The estimated mass-to-flux ratios are 0.45-2.20 and 0.63-2.76 for HH 211 MMS and IC 348 MMS, respectively, while the ratios for the filament is 0.33-1.50. This result may suggest that the transition from subcritical to supercritical conditions occurs at the core scale ($\sim$ 0.05 pc) in the region. In addition, we study the energy balance of the cores and find that the relative strength of turbulence to the magnetic field tends to be stronger for IC 348 MMS than HH 211 MMS. The result could potentially explain the different configurations inside the two cores: a single protostellar system in HH 211 MMS and multiple protostars in IC 348 MMS., Comment: Accepted for publication in ApJ. 21 pages, 12 figures

In recent years, there has been a surge of interest in studying different ways to reformulate nonconvex optimization problems, especially those that involve binary variables. This interest surge is due to advancements in computing technologies, such as quantum and Ising devices, as well as improvements in quantum and classical optimization solvers that take advantage of particular formulations of nonconvex problems to tackle their solutions. Our research characterizes the equivalence between equality-constrained nonconvex optimization problems and their Lagrangian relaxation, enabling the aforementioned new technologies to solve these problems. In addition to filling a crucial gap in the literature, our results are readily applicable to many important situations in practice. To obtain these results, we bridge between specific optimization problem characteristics and broader, classical results on Lagrangian duality for general nonconvex problems. Further, our approach takes a comprehensive approach to the question of equivalence between problem formulations. We consider this question not only from the perspective of the problem's objective but also from the viewpoint of its solution. This perspective, often overlooked in existing literature, is particularly relevant for problems featuring continuous and binary variables.

We consider extreme type-II superconductors modeled by the Ginzburg--Landau energy with a pinning term $a_\varepsilon(x)$, which we assume to be a bounded measurable function such that $b\leq a_\varepsilon(x)\leq 1$ for some constant $b>0$. A crucial feature of this type of superconductors is the occurrence of vortices, which appear above the so-called first critical field $H_{c_1}$. In this paper we estimate this value and characterize the behavior of the Meissner solution, the unique vortexless configuration that globally minimizes the energy below $H_{c_1}$. In addition, we show that beyond this value, for applied fields whose strength is slightly below the so-called superheating field $H_{sh}$, there exists a unique Meissner-type solution that locally minimizes the energy., Comment: 36 pages

The Active Galactic Nuclei (AGN) channel for the formation of binary black hole (BBH) mergers has been previously studied as a potential formation channel for the merging compact binaries observed by the LIGO/Virgo/KAGRA (LVK) scientific collaboration. The first two papers in this series explored the McFACTS code for the evolution of black hole orbits in AGN accretion disks for individual galaxy models and described the characteristics of predicted BBH populations in realizations of those models (such as the correlation between mass ratio and aligned spin). In this work, we explore the impact of the properties of AGN host galaxies and assume an AGN lifetime and cosmological model for the density of AGN in a universe like our own. By sampling from an inferred population of AGN, we marginalize over galaxy mass to predict a population of BBH mergers observable by modern ground-based gravitational wave observatories. We find that for reasonable assumptions, AGN disk environments may account for massive BBH mergers such as GW190521 and GW190929_012149. We find that the majority of observable BBH mergers from our simulation are expected to originate in galaxies with a super-massive black hole between $10^{7}M_{\odot}$ and $10^{9.4}M_{\odot}$. We also find that if hierarchical mergers from AGN disks account for a substantial part of the LVK population, our current models require an AGN lifetime of 0.5 to 2.5 Myr., Comment: Submitting to ApJ

QTRAJ is a computer code that simulates the propagation of quarkonium in the quark-gluon plasma (QGP) based on the quantum-trajectory algorithm. This algorithm solves a master equation in which the quarkonium is treated as an open quantum system (OQS). A major advantage of this approach is that it turns a 3D spatial evolution for a density matrix into a 1D Schr\"odinger equation for a wavefunction with a non-hermitian Hamiltonian, drastically reducing the computational cost. So far, the interaction implemented in the master equation was obtained within the framework of potential non-relativistic QCD (pNRQCD), and restricted to the regime $rT \ll 1$, where $r$ is the size of the color dipole and $T$ is the temperature. In the environment produced in heavy-ion collisions (HIC's) this limit is accurate for $\Upsilon(1S)$, but the applicability to other quarkonium states is dubious. In the present study we generalize the above approach, extending it to the regime $rT\!\sim\! 1$ in the one-gluon exchange approximation, with proper Hard Thermal Loop (HTL) resummation of medium effects. This is done by implementing new jump operators connecting different color states of the $Q\bar Q$ pair and expanding them in plane waves, giving rise to a variation of the algorithm present in QTRAJ 1.0. Here we provide an overview of this approach comparing the $rT \ll 1$ and $rT\sim 1$ cases, and we discuss prospects for phenomenological application to excited states of bottomonium., Comment: Proceedings of the Conference QNP 2024

Despite significant advancements in natural language generation, controlling language models to produce texts with desired attributes remains a formidable challenge. In this work, we introduce RSA-Control, a training-free controllable text generation framework grounded in pragmatics. RSA-Control directs the generation process by recursively reasoning between imaginary speakers and listeners, enhancing the likelihood that target attributes are correctly interpreted by listeners amidst distractors. Additionally, we introduce a self-adjustable rationality parameter, which allows for automatic adjustment of control strength based on context. Our experiments, conducted with two task types and two types of language models, demonstrate that RSA-Control achieves strong attribute control while maintaining language fluency and content consistency. Our code is available at https://github.com/Ewanwong/RSA-Control., Comment: Accepted to EMNLP 2024 (main conference)

Active galactic nuclei (AGN) are a promising source of the binary black hole (BBH) mergers observed in gravitational waves with LIGO-Virgo-Kagra (LVK). Constraining the AGN channel allows us to limit AGN parameter space (disk density, size, average lifetime) and nuclear star cluster (NSC) parameter space. Constraints on AGN and NSCs have implications for $\Lambda$CDM models of AGN feedback and models of AGN-driven SMBH merger and growth. Here we present several qualitative studies of the AGN channel using new public, open-source, fast, reproducible code \texttt{McFACTS}\footnote{https://github.com/mcfacts}:Monte Carlo for AGN channel Testing \& Simulation. We demonstrate several important features for testing the AGN channel, including: i) growth to large mass IMBH is helped by the presence of migration traps or disk boundaries, ii) flat BH initial mass functions highlight hierarchical merger features in the mass spectrum, iii) high rates of dynamical encounters strongly inhibit BBH formation and merger at migration traps, iv) the ($q,\chi_{\rm eff}$) anti-correlation is a strong test of the bias to prograde mergers in the AGN channel, v) spheroid encounters can drive a fraction of mergers with high in-plane spin components ($\chi_{\rm p}$), vi) a high rate of extreme mass ratio inspirals (EMRIs) are driven by an initial population of embedded retrograde BH, vii) Both LVK and LISA are powerful probes of models of AGN disks and their embedded populations., Comment: 21 pages, 12 figures, 1 table, ApJ (submitted)

Informal sensemaking surrounding U.S. election processes has been fraught in recent years, due to the inherent uncertainty of elections, the complexity of election processes in the U.S., and to disinformation. Based on insights from qualitative analysis of election rumors spreading online in 2020 and 2022, we introduce the concept of manipulated sensemaking to describe how disinformation functions by disrupting online audiences ability to make sense of novel, uncertain, or ambiguous information. We describe how at the core of this disruption is the ability for disinformation to shape broad, underlying stories called deep stories which determine the frames we use to make sense of this novel information. Additionally, we explain how sensemakings orientation around plausible explanations over accurate explanations makes it vulnerable to manipulation. Lastly, we demonstrate how disinformed deep stories shape sensemaking not just for a single event, but for many events in the future., Comment: 10 pages, CHI 2024 Sensemaking workshop

The direct pair-production of the superpartner of the $\tau$-lepton, the $\widetilde{\tau}$, is one of the most interesting channels to search for SUSY in: the $\widetilde{\tau}$ is likely to be the lightest of the scalar leptons, and is one of the most experimentally chalanging ones. The current model-independent $\widetilde{\tau}$ limits come from LEP, while limits obtained at the LHC do extend to higher masses, but are model-dependent. The future Higgs factories will be powerful facilities for SUSY searches, offering advantages with respect to previous electron-positron colliders as well as to hadron machines. In order to quantify the capabilities of these future $e^+e^-$ colliders, the "worst-case" scenario for $\widetilde{\tau}$ exclusion/discovery has been studied, taking into account the effect of the $\widetilde{\tau}$ mixing on $\widetilde{\tau}$ production cross-section and detection efficiency. To evaluate the latter, the ILD concept, originally developed for the International Linear Collider (ILC), and the ILC beam conditions at a centre-of-mass energy of 500 GeV have been used for detailed simulations. The obtained exclusion and discovery reaches extend to only a few GeV below the kinematic limit even in the worst-case scenario. The results of the detailed simulation study are then discussed in view of the experimental environment of other proposed Higgs factory projects., Comment: Contributed to The International Workshop on Future Linear Colliders (LCWS2024), 8-11 July 2024,Tokyo, Japan

Social network analysis allows researchers to discover insights from connections between people. While the process of building a social network is relatively straightforward for contemporary social media, deriving connections from historical archives remains a challenging task, with every data collection presenting its unique challenges. Our contribution focuses on building and analysing a social network from the correspondence archive of Sybren Valkema (1916-1996), a Dutch glass artist and educator. The archive contains both typewritten and handwritten documents in multiple languages, and includes letters from glass artists, art students, art collectors and other agents. We develop an automatic pipeline approach which includes separating handwritten and typed documents, performing text recognition specific to the document modality, extracting names of people from text using named entity recognition, de-duplicating the resulting names to create actor nodes, classifying the actors using entity linking, and, finally, connecting them together and analysing the resulting network. Every part of the pipeline is evaluated against a manual analysis performed by an art historian on a subset of the data collection in order to find out which pitfalls of the automatic approach need to be resolved in future work and, on the contrary, whether using the automatic approach allows to discover any additional insights. The results show strong performance in discovering sender-receiver connections as well as additional meaningful connections in text, with the main challenge being text recognition on scanned pages.

In this work, we introduce a methodology for alignment designed to enhance the ability of large language models (LLMs) to articulate their reasoning (self-explanation) even in the absence of annotated rationale explanations. Our alignment methodology comprises three key components: explanation quality assessment, self-instruction dataset generation, and model alignment. Additionally, we present a novel technique called Alignment with Anchor Preference Pairs, which improves the selection of preference pairs by categorizing model outputs into three groups: consistently correct, consistently incorrect, and variable. By applying tailored strategies to each category, we enhance the effectiveness of Direct Preference Optimization (DPO). Our experimental results demonstrate that this approach significantly improves explanation quality while maintaining accuracy compared to other fine-tuning strategies.

The FLASH effect, characterized by potential sparing of organs at risk (OAR) through ultra-high dose rate irradiation, has garnered significant attention for its capability to address indications previously untreatable at conventional dose rates (DR) with hypofractionated schemes. While considerable biological research is needed to understand the FLASH effect and determine the FLASH modifying factors (FMF) for individual OARs, exploratory treatment planning studies have also emerged. This study aims to show that spinal metastases are candidate treatment sites likely to benefit from this phenomenon and establish the requisite FMFs to achieve the protective FLASH effect. A conformal FLASH Proton SBRT plan was generated for a patient with spine metastasis in a research version of RayStation11B (RaySearch laboratories AB, Stockhom) on an IBA Proteus Plus system. Two oblique posterior beams were used in the plan. The prescribed dose to the CTV was set according to 3 different fractionation regimens: 5 fractions (fx) of 7 Gy, 8 fx of 5 Gy, and 10 fx of 4.2 Gy. Spot filtering and sorting techniques were applied to maximize the 5% pencil beam scanning DR in the spinal cord (SC). The FLASH effect was assumed to be observed within irradiated regions above 40 Gy/s and 4 Gy per fraction. The generated plans successfully ensure robust target coverage in each fraction. The volume of SC that does not comply with the clinical goal adheres to the FLASH effect conditions in each fraction. Depending on the aforementioned fractionation schemes used, a FMF of approximately 0.6 to 0.8 is necessary to enable such treatment in FLASH conditions. Our study demonstrates the potential of hypofractionated FLASH PT in treating spine metastasis while preserving SC integrity. This approach could enable more effective treatment of spinal metastases, potentially preventing SC compression and paralysis in these patients., Comment: 7 pages, 3 figures (+ supplementary material)

We consider a sequence $\mathbf{T} = (\mathcal{T}_n : n \in \mathbb{N}^+)$ of trees $\mathcal{T}_n$ where, for some $\Delta \in \mathbb{N}^+$ every $\mathcal{T}_n$ has height at most $\Delta$ and as $n \to \infty$ the minimal number of children of a nonleaf tends to infinity. We can view every tree as a (first-order) $\tau$-structure where $\tau$ is a signature with one binary relation symbol. For a fixed (arbitrary) finite and relational signature $\sigma \supseteq \tau$ we consider the set $\mathbf{W}_n$ of expansions of $\mathcal{T}_n$ to $\sigma$ and a probability distribution $\mathbb{P}_n$ on $\mathbf{W}_n$ which is determined by a (parametrized/lifted) Probabilistic Graphical Model (PGM) $\mathbb{G}$ which can use the information given by $\mathcal{T}_n$. The kind of PGM that we consider uses formulas of a many-valued logic that we call $PLA^*$ with truth values in the unit interval $[0, 1]$. We also use $PLA^*$ to express queries, or events, on $\mathbf{W}_n$. With this setup we prove that, under some assumptions on $\mathbf{T}$, $\mathbb{G}$, and a (possibly quite complex) formula $\varphi(x_1, \ldots, x_k)$ of $PLA^*$, as $n \to \infty$, if $a_1, \ldots, a_k$ are vertices of the tree $\mathcal{T}_n$ then the value of $\varphi(a_1, \ldots, a_k)$ will, with high probability, be almost the same as the value of $\psi(a_1, \ldots, a_k)$, where $\psi(x_1, \ldots, x_k)$ is a ``simple'' formula the value of which can always be computed quickly (without reference to $n$), and $\psi$ itself can be found by using only the information that defines $\mathbf{T}$, $\mathbb{G}$ and $\varphi$. A corollary of this, subject to the same conditions, is a probabilistic convergence law for $PLA^*$-formulas., Comment: arXiv admin note: text overlap with arXiv:2401.04802

Observations of the high-$z$ Universe from JWST have revealed a new population of bright, early galaxies. A robust statistical interpretation of this data requires fast forward models that account for uncertainties in galaxy evolution and incorporate observational systematic effects. We present a probabilistic framework for population synthesis of high-$z$ galaxies and inference of their properties. Our framework is based on the semi-analytic galaxy-formation model Galacticus. To infer the astrophysical parameters governing high-$z$ galaxy evolution, we analyze JWST data from the CEERS and NGDEEP surveys and calculate the likelihood of observing individual objects in apparent magnitude--redshift space, for $z\geq8.5$ galaxy candidates. We include observational selection effects due to limited survey volume and depth, as well as photometric redshift uncertainties. We recover the posterior probability distribution for parameters describing star formation and outflow rates. We place an upper limit on the star formation timescale of $500~\mathrm{Myr}$ at a disk velocity of $50~\mathrm{km\ s}^{-1}$, and we infer a characteristic velocity at which the outflow mass-loading factor is $\sim 1$ of $150^{+280}_{-60}~\mathrm{km\ s}^{-1}$, both at $95\%$ confidence. Marginalizing over our astrophysical model, we find that galaxies in CEERS and NGDEEP data occupy halos with virial masses $10^{10\pm 0.5}~M_{\mathrm{\odot}}$ at $8.5\leq z\leq 12$, at $95\%$ confidence. The star formation timescale preferred by our fit is relatively short compared to typical values at lower redshifts, consistent with previous findings. The modeling and analysis framework presented here can enable systematic tests of high-$z$ galaxies' dust content, initial mass functions, and star formation burstiness in the future., Comment: 14 pages, 10 figures

This paper presents the findings from the third edition of the Chat Translation Shared Task. As with previous editions, the task involved translating bilingual customer support conversations, specifically focusing on the impact of conversation context in translation quality and evaluation. We also include two new language pairs: English-Korean and English-Dutch, in addition to the set of language pairs from previous editions: English-German, English-French, and English-Brazilian Portuguese. We received 22 primary submissions and 32 contrastive submissions from eight teams, with each language pair having participation from at least three teams. We evaluated the systems comprehensively using both automatic metrics and human judgments via a direct assessment framework. The official rankings for each language pair were determined based on human evaluation scores, considering performance in both translation directions--agent and customer. Our analysis shows that while the systems excelled at translating individual turns, there is room for improvement in overall conversation-level translation quality., Comment: 12 pages, 5 figures, 13 tables

We measure the growth of cosmic density fluctuations on large scales and across the redshift range $0.3

Objective. As proton arc therapy (PAT) approaches clinical implementation, optimizing treatment plans for this innovative delivery modality remains challenging, especially in addressing arc delivery time. Existing algorithms for minimizing delivery time are either optimal but computationally demanding or fast but at the expense of sacrificing many degrees of freedom. In this study, we introduce a flexible method for pre-selecting energy layers (EL) in PAT treatment planning before the actual robust spot weight optimization. Our EL pre-selection method employs metaheuristics to minimize a bi-objective function, considering a dynamic delivery time proxy and tumor geometrical coverage penalized as a function of selected organs-at-risk crossing. It is capable of parallelizing multiple instances of the problem. We evaluate the method using three different treatment sites, providing a comprehensive dosimetric analysis benchmarked against dynamic proton arc plans generated with early energy layer selection and spot assignment (ELSA) and IMPT plans in RayStation TPS. The algorithm efficiently generates Pareto-optimal EL pre-selections in approximately 5 minutes. Subsequent PAT treatment plans derived from these selections and optimized within the TPS, demonstrate high-quality target coverage, achieving a high conformity index, and effective sparing of organs at risk. These plans meet clinical goals while achieving a 20 to 40% reduction in delivery time compared to ELSA plans. The proposed algorithm offers speed and efficiency, producing high-quality PAT plans by placing proton arc sectors to efficiently reduce delivery time while maintaining good target coverage and healthy tissues sparing., Comment: 16 pages, 10 figures, submitted to Physics in Medecine and Biology

The kinematic Sunyaev--Zel'dovich (kSZ) effect induces a non-zero density-density-temperature bispectrum, which we can use to reconstruct the large-scale velocity field from a combination of cosmic microwave background (CMB) and galaxy density measurements, in a procedure known as ``kSZ velocity reconstruction''. This method has been forecast to constrain large-scale modes with future galaxy and CMB surveys, improving their measurement beyond what is possible with the galaxy surveys alone. Such measurements will enable tighter constraints on large-scale signals such as primordial non-Gaussianity, deviations from homogeneity, and modified gravity. In this work, we demonstrate a statistically significant measurement of kSZ velocity reconstruction for the first time, by applying quadratic estimators to the combination of the ACT DR6 CMB+kSZ map and the DESI LRG galaxies (with photometric redshifts) in order to reconstruct the velocity field. We do so using a formalism appropriate for the 2-dimensional projected galaxy fields that we use, which naturally incorporates the curved-sky effects important on the largest scales. We find evidence for the signal by cross-correlating with an external estimate of the velocity field from the spectroscopic BOSS survey and rejecting the null (no-kSZ) hypothesis at $3.8\sigma$. Our work presents a first step towards the use of this observable for cosmological analyses., Comment: 16 pages (main)+5 pages (Appendix); 13 figures (main) + 8 figures (appendix)

This review examines current and potential applications of DTs in healthcare, focusing on the integration of multi-omics data using multilayer network approaches in cancer research. We discuss methodologies, tools and platforms commonly used for this integration, while highlighting case studies, challenges, and research gaps. Finally, we advocate the need for incorporating diverse data types to produce more effective DTs for personalization of cancer treatments and in silico clinical trials., Comment: 20 pages, 3 figures

This scientific article presents the implementation of an automated control system for detecting and classifying faults in tuna metal cans using artificial vision. The system utilizes a conveyor belt and a camera for visual recognition triggered by a photoelectric sensor. A robotic arm classifies the metal cans according to their condition. Industry 4.0 integration is achieved through an IoT system using Mosquitto, Node-RED, InfluxDB, and Grafana. The YOLOv5 model is employed to detect faults in the metal can lids and the positioning of the easy-open ring. Training with GPU on Google Colab enables OCR text detection on the labels. The results indicate efficient real-time problem identification, optimization of resources, and delivery of quality products. At the same time, the vision system contributes to autonomy in quality control tasks, freeing operators to perform other functions within the company., Comment: 6 pages, 12 figures

This document presents the study of the problem of location and trajectory that a robot must follow. It focuses on applying the Kalman filter to achieve location and trajectory estimation in an autonomous mobile differential robot. The experimental data was carried out through tests obtained with the help of two incremental encoders that are part of the construction of the differential robot. The data transmission is carried out from a PC where the control is carried out with the Matlab/Simulink software. The results are expressed in graphs showing the path followed by the robot using PI control, the estimator of the Kalman filter in a real system., Comment: 7 pages, 13 figures

We present $72$ cosmological dark matter (DM)--only N-body zoom-in simulations with initial conditions beyond cold, collisionless dark matter (CDM), as the first installment of the COZMIC suite. We simulate Milky Way (MW) analogs with linear matter power spectra, $P(k)$ for: i) thermal-relic warm dark matter (WDM) with masses $m_{\mathrm{WDM}}\in [3,4,5,6,6.5,10]~\mathrm{keV}$, ii) fuzzy dark matter (FDM) with masses $m_{\mathrm{FDM}}\in [25.9,69.4,113,151,185,490]\times 10^{-22}~\mathrm{eV}$, and iii) interacting dark matter (IDM) with a velocity-dependent elastic proton scattering cross section $\sigma=\sigma_0 v^n$ relative particle velocity scaling $n\in [2,4]$, and DM mass $m_{\mathrm{IDM}}\in[10^{-4},~ 10^{-2},~ 1]~\mathrm{GeV}$. Subhalo mass function (SHMF) suppression is significantly steeper in FDM versus WDM, while dark acoustic oscillations in $P(k)$ can reduce SHMF suppression for IDM. We fit SHMF models to our simulation results and derive new bounds on WDM and FDM from the MW satellite population, obtaining $m_{\mathrm{WDM}}>5.9~\mathrm{keV}$ and $m_{\mathrm{FDM}}>1.4\times 10^{-20}~\mathrm{eV}$ at $95\%$ confidence; these limits are $\approx 10\%$ weaker and $5\times$ stronger than previous constraints due to the updated transfer functions and SHMF models, respectively. We estimate IDM bounds for $n=2$ ($n=4$) and obtain $\sigma_0 < 1.0\times 10^{-27}$, $1.3\times 10^{-24}$, and $3.1\times 10^{-23}~\mathrm{cm}^2$ ($\sigma_0 < 9.9\times 10^{-27}$, $9.8\times 10^{-21}$, and $2.1\times 10^{-17}~\mathrm{cm}^2$) for $m_{\mathrm{IDM}}=10^{-4}$, $10^{-2}$, and $1$ GeV, respectively. Thus, future development of IDM SHMF models can improve IDM cross section bounds by up to a factor of $\sim 20$ with current data. COZMIC presents an important step toward accurate small-scale structure modeling in beyond-CDM cosmologies, critical to upcoming observational searches for DM physics., Comment: 36 pages, 19 figures, 2 tables; comments welcome

Dense features, customized for different business scenarios, are essential in short video classification. However, their complexity, specific adaptation requirements, and high computational costs make them resource-intensive and less accessible during online inference. Consequently, these dense features are categorized as `Privileged Dense Features'.Meanwhile, end-to-end multi-modal models have shown promising results in numerous computer vision tasks. In industrial applications, prioritizing end-to-end multi-modal features, can enhance efficiency but often leads to the loss of valuable information from historical privileged dense features. To integrate both features while maintaining efficiency and manageable resource costs, we present Confidence-aware Privileged Feature Distillation (CPFD), which empowers features of an end-to-end multi-modal model by adaptively distilling privileged features during training. Unlike existing privileged feature distillation (PFD) methods, which apply uniform weights to all instances during distillation, potentially causing unstable performance across different business scenarios and a notable performance gap between teacher model (Dense Feature enhanced multimodal-model DF-X-VLM) and student model (multimodal-model only X-VLM), our CPFD leverages confidence scores derived from the teacher model to adaptively mitigate the performance variance with the student model. We conducted extensive offline experiments on five diverse tasks demonstrating that CPFD improves the video classification F1 score by 6.76% compared with end-to-end multimodal-model (X-VLM) and by 2.31% with vanilla PFD on-average. And it reduces the performance gap by 84.6% and achieves results comparable to teacher model DF-X-VLM. The effectiveness of CPFD is further substantiated by online experiments, and our framework has been deployed in production systems for over a dozen models., Comment: Camera ready for CIKM 2024