Search

Developing algorithms to search through data efficiently is a challenging part of searching for signs of technology beyond our solar system. We have built a digital signal processing system and computer cluster on the backend of the Karl G. Jansky Very Large Array (VLA) in New Mexico in order to search for signals throughout the Galaxy consistent with our understanding of artificial radio emissions. In our first paper, we described the system design and software pipelines. In this paper, we describe a postprocessing pipeline to identify persistent sources of interference, filter out false positives, and search for signals not immediately identifiable as anthropogenic radio frequency interference during the VLA Sky Survey. As of 01 September 2024, the Commensal Open-source Multi-mode Interferometric Cluster had observed more than 950,000 unique pointings. This paper presents the strategy we employ when commensally observing during the VLA Sky Survey and a postprocessing strategy for the data collected during the survey. To test this postprocessing pipeline, we searched toward 511 stars from the $Gaia$ catalog with coherent beams. This represents about 30 minutes of observation during VLASS, where we typically observe about 2000 sources per hour in the coherent beamforming mode. We did not detect any unidentifiable signals, setting isotropic power limits ranging from 10$^{11}$ to 10$^{16}$W., Comment: 17 Pages, Accepted in AJ

Medium-timescale (minutes to hours) radio transients are a relatively unexplored population. The wide field-of-view and high instantaneous sensitivity of instruments such as MeerKAT provides an opportunity to probe this class of sources, using image-plane detection techniques. The previous letter in this series describes our project and associated TRON pipeline designed to mine archival MeerKAT data for transient and variable sources. In this letter, we report on a new transient, a radio flare, associated with Gaia DR3 6865945581361480448, a G type star, whose parallax places it at a distance of 1334 pc. Its duration and high degree of circular polarization suggests electron cyclotron maser instability as the mechanism, consistent with an RS CVn variable., Comment: 4 pages, 4 figures, submitted to MNRAS Letters

Medium-timescale (minutes to hours) radio transients are a relatively unexplored population. The wide field-of-view and high instantaneous sensitivity of instruments such as MeerKAT provides an opportunity to probe this class of sources, using image-plane detection techniques. We aim to systematically mine archival synthesis imaging data in order to search for medium-timescale transients and variables that are not detected by conventional long-track image synthesis techniques. We deploy a prototype blind transient and variable search pipeline named TRON. This processes calibrated visibility data, constructs high-time cadence images, performs a search for variability on multiple timescales, and extracts lightcurves for detected sources. As proof of concept, we apply it to three MeerKAT observations of globular clusters, known to host transient or variable sources. We detect a previously known eclipsing MSP suspected to be a `black widow' system, in the globular cluster Omega Centauri, with a light curve confirming the eclipsing nature of the emission. We detect a previously known `red back' eclipsing MSP in the globular cluster Terzan 5. Using observations of the globular cluster 47 Tucanae, we detect two known millisecond pulsars (MSPs), and one previously reported MSP candidate, with hints of eclipsing behaviour., Comment: 6 pages, 8 figures, submitted to MNRAS Letters

Correlative computational microscopy is accelerating the mapping of dynamic biological systems by integrating morphological and molecular measurements across spatial scales, from organelles to entire organisms. Visualization, measurement, and prediction of interactions among the components of biological systems can be accelerated by generalist computational imaging frameworks that relax the trade-offs imposed by multiplex dynamic imaging. This work reports a generalist framework for wave optical imaging of the architectural order (waveOrder) among biomolecules for encoding and decoding multiple specimen properties from a minimal set of acquired channels, with or without fluorescent labels. waveOrder expresses material properties in terms of elegant physically motivated basis vectors directly interpretable as phase, absorption, birefringence, diattenuation, and fluorophore density; and it expresses image data in terms of directly measurable Stokes parameters. We report a corresponding multi-channel reconstruction algorithm to recover specimen properties in multiple contrast modes. With this framework, we implement multiple 3D computational microscopy methods, including quantitative phase imaging, quantitative label-free imaging with phase and polarization, and fluorescence deconvolution imaging, across scales ranging from organelles to whole zebrafish. These advances are available via an extensible open-source computational imaging library, waveOrder, and a napari plugin, recOrder., Comment: Main text: 11 pages with 5 figures and one table; Ancillary files: 15 pages of supplementary text with 4 figures and one table; 5 videos. Changelog v1->v2: separated supplemental doc and updated video 2

Vision-Language Models (VLMs) have shown significant promise in Visual Question Answering (VQA) tasks by leveraging web-scale multimodal datasets. However, these models often struggle with continual learning due to catastrophic forgetting when adapting to new tasks. As an effective remedy to mitigate catastrophic forgetting, rehearsal strategy uses the data of past tasks upon learning new task. However, such strategy incurs the need of storing past data, which might not be feasible due to hardware constraints or privacy concerns. In this work, we propose the first data-free method that leverages the language generation capability of a VLM, instead of relying on external models, to produce pseudo-rehearsal data for addressing continual VQA. Our proposal, named as GaB, generates pseudo-rehearsal data by posing previous task questions on new task data. Yet, despite being effective, the distribution of generated questions skews towards the most frequently posed questions due to the limited and task-specific training data. To mitigate this issue, we introduce a pseudo-rehearsal balancing module that aligns the generated data towards the ground-truth data distribution using either the question meta-statistics or an unsupervised clustering method. We evaluate our proposed method on two recent benchmarks, \ie VQACL-VQAv2 and CLOVE-function benchmarks. GaB outperforms all the data-free baselines with substantial improvement in maintaining VQA performance across evolving tasks, while being on-par with methods with access to the past data.

Let $X\rightarrow S$ be a fibration of relative dimension at most two and let $(X,\Delta)$ be a klt pair for which $K_X+\Delta \equiv_S 0$. We show that there are only finitely many Mori chambers and Mori faces in the movable effective cone $\mathcal{M}^e(X/S)$ up to the action of relative pseudo-automorphisms of $X/S$ preserving $\Delta$., Comment: 35 pages

Given an $r$-colouring of the edges of a graph $G$, we say that it can be partitioned into $p$ monochromatic cycles when there exists a set of $p$ vertex-disjoint monochromatic cycles covering all the vertices of $G$. In the literature of this problem, an edge and a single vertex both count as a cycle. We show that for every $2$-colouring of the edges of a complete balanced bipartite graph, $K_{n,n}$, it can be partitioned into at most 4 monochromatic cycles. This type of question was first studied in 1970 for complete graphs and in 1983, by Gy\'arf\'as and Lehel, for $K_{n,n}$. In 2014, Pokrovskiy, has showed that any $2$-colouring of the edges of $K_{n,n}$ can be partitioned into at most $3$ paths. It turns out that finding monochromatic cycles instead of paths is a natural question that has also being asked for other graphs. In 2015, Schaudt and Stein have showed that at most 14 cycles are necessary., Comment: 17 pages, 21 figures

Metal carbides, nitrides, or carbonitrides of early transition metals, better known as MXenes, possess notable structural, electrical, and magnetic properties. Analyzing electronic structures by calculating structural stability, band structure, density of states, Bader charge transfer, and work functions utilizing first principle calculations, we revealed that titanium nitride Mxenes, namely Ti$_2$N and Ti$_3$N$_2$, have excess anionic electrons in their pseudo-atomic structure inside the crystal lattice, making them MXene electrides. Bulk Ti$_3$N$_2$ has competing antiferromagnetic (AFM) and ferromagnetic(FM) configurations with slightly more stable AFM configurations, while the Ti$_2$N MXene is nonmagnetic. Although Ti$_3$N$_2$ favors AFM configurations with hexagonal crystal systems having $6/mmm$ point group symmetry, Ti$_3$N$_2$ does not support altermagnetism. The monolayer of the Ti$_3$N$_2$ MXene is a ferromagnetic electride. These unique properties of having non-nuclear interstitial anionic electrons in the electronic structure of titanium nitride MXene have not yet been reported in the literature. Density functional theory calculations show TiN is neither an electride, MXene, or magnetic.

Standard Unsupervised Domain Adaptation (UDA) aims to transfer knowledge from a labeled source domain to an unlabeled target but usually requires simultaneous access to both source and target data. Moreover, UDA approaches commonly assume that source and target domains share the same labels space. Yet, these two assumptions are hardly satisfied in real-world scenarios. This paper considers the more challenging Source-Free Open-set Domain Adaptation (SF-OSDA) setting, where both assumptions are dropped. We propose a novel approach for SF-OSDA that exploits the granularity of target-private categories by segregating their samples into multiple unknown classes. Starting from an initial clustering-based assignment, our method progressively improves the segregation of target-private samples by refining their pseudo-labels with the guide of an uncertainty-based sample selection module. Additionally, we propose a novel contrastive loss, named NL-InfoNCELoss, that, integrating negative learning into self-supervised contrastive learning, enhances the model robustness to noisy pseudo-labels. Extensive experiments on benchmark datasets demonstrate the superiority of the proposed method over existing approaches, establishing new state-of-the-art performance. Notably, additional analyses show that our method is able to learn the underlying semantics of novel classes, opening the possibility to perform novel class discovery.

Causal Representation Learning (CRL) aims at identifying high-level causal factors and their relationships from high-dimensional observations, e.g., images. While most CRL works focus on learning causal representations in a single environment, in this work we instead propose a first step towards learning causal representations from temporal sequences of images that can be adapted in a new environment, or composed across multiple related environments. In particular, we introduce DECAF, a framework that detects which causal factors can be reused and which need to be adapted from previously learned causal representations. Our approach is based on the availability of intervention targets, that indicate which variables are perturbed at each time step. Experiments on three benchmark datasets show that integrating our framework with four state-of-the-art CRL approaches leads to accurate representations in a new environment with only a few samples., Comment: Accepted to the 3rd Conference on Causal Learning and Reasoning (CLeaR 2024)

This study provides a comprehensive benchmark framework for Source-Free Unsupervised Domain Adaptation (SF-UDA) in image classification, aiming to achieve a rigorous empirical understanding of the complex relationships between multiple key design factors in SF-UDA methods. The study empirically examines a diverse set of SF-UDA techniques, assessing their consistency across datasets, sensitivity to specific hyperparameters, and applicability across different families of backbone architectures. Moreover, it exhaustively evaluates pre-training datasets and strategies, particularly focusing on both supervised and self-supervised methods, as well as the impact of fine-tuning on the source domain. Our analysis also highlights gaps in existing benchmark practices, guiding SF-UDA research towards more effective and general approaches. It emphasizes the importance of backbone architecture and pre-training dataset selection on SF-UDA performance, serving as an essential reference and providing key insights. Lastly, we release the source code of our experimental framework. This facilitates the construction, training, and testing of SF-UDA methods, enabling systematic large-scale experimental analysis and supporting further research efforts in this field.

The flow of yield stress fluids in porous media presents interesting complexity due to the interplay between the non-linear rheology and the heterogeneity of the medium. A remarkable consequence is that the number of flow paths increases with the applied pressure difference and is responsible for a non-linear Darcy law. Previous studies have focused on the protocol where the pressure difference is imposed. Here we consider instead the case of imposed flow rate, $Q$. In contrast to Newtonian fluids, the two types of boundary conditions have an important influence on the flow field. Using a two-dimensional pore network model we observe a boundary layer of merging flow paths of size $\ell(Q) \sim Q^{-\mu/\delta}$ where $\mu = 0.42 \pm 0.02$ and $\delta \simeq 0.63 \pm 0.05$. Beyond this layer the density of the flow paths is homogeneous and grows as $Q^\mu$. Using a mapping to the directed polymer model we identify $\delta$ with the roughness exponent of the polymer. We also characterize the statistics of non-flowing surfaces in terms of avalanches pulled at one end.

The primary goal of the search for extraterrestrial intelligence (SETI) is to gain an understanding of the prevalence of technologically advanced beings (organic or inorganic) in the Galaxy. One way to approach this is to look for technosignatures: remotely detectable indicators of technology, such as temporal or spectral electromagnetic emissions consistent with an artificial source. With the new Commensal Open-Source Multimode Interferometer Cluster (COSMIC) digital backend on the Karl G. Jansky Very Large Array (VLA), we aim to conduct a search for technosignatures that is significantly larger, more sensitive, and more efficient than previously attempted. The COSMIC system is currently operational on the VLA, recording data, and designed with the flexibility to provide user-requested modes. This paper describes the hardware system design, the current software pipeline, and plans for future development., Comment: 30 pages, 17 figures. Submitted to AJ

We investigate the behavior of drainage displacements in heterogeneous porous media finding a transition from viscous fingering to foam-like region. A pore network model incorporating the formation of blobs is adopted to study this phenomenon. By imposing a pressure difference between the inlet and outlet, we observe that the displacement pattern undergoes a significant transition from a continuous front of growing viscous fingers to the emergence of foam, which develops and propagates until breakthrough. This transition occurs at a specific distance from the inlet, which we measure and analyze as a function of the viscosity ratio and the capillary number, demonstrating that it follows a non-trivial power-law decay with both the parameters. Moreover, we discuss the relationship between the evolution of the total flow rate and the local pressure drop, showing that the foam developed reduces global mobility. We observe that foam is formed from the fragmentation of viscous fingers beneath the front, and this instability mechanism is connected with fluctuations of the local flow rate, which we analyze both in the viscous fingering region and in the foam region., Comment: Main text: 14 pages, 11 Figures. Supplementary Material: 3 pages, 5 Figures

Fine-tuning and Domain Adaptation emerged as effective strategies for efficiently transferring deep learning models to new target tasks. However, target domain labels are not accessible in many real-world scenarios. This led to the development of Unsupervised Domain Adaptation (UDA) methods, which only employ unlabeled target samples. Furthermore, efficiency and privacy requirements may also prevent the use of source domain data during the adaptation stage. This challenging setting, known as Source-Free Unsupervised Domain Adaptation (SF-UDA), is gaining interest among researchers and practitioners due to its potential for real-world applications. In this paper, we provide the first in-depth analysis of the main design choices in SF-UDA through a large-scale empirical study across 500 models and 74 domain pairs. We pinpoint the normalization approach, pre-training strategy, and backbone architecture as the most critical factors. Based on our quantitative findings, we propose recipes to best tackle SF-UDA scenarios. Moreover, we show that SF-UDA is competitive also beyond standard benchmarks and backbone architectures, performing on par with UDA at a fraction of the data and computational cost. In the interest of reproducibility, we include the full experimental results and code as supplementary material.

Understanding the flow of yield stress fluids in porous media is a major challenge. In particular, experiments and extensive numerical simulations report a non-linear Darcy law as a function of the pressure gradient. In this letter, we consider a tree-like porous structure for which the problem of the flow can be resolved exactly thanks to a mapping with the directed polymer (DP) with disordered bond energies on the Cayley tree. Our results confirm the non-linear behavior of the flow and expresses its full pressure-dependence via the density of low-energy paths of DP restricted to vanishing overlap. These universal predictions are confirmed by extensive numerical simulations., Comment: 5 pages,3 figures + 1 Supplementary material

Puzzle solving is a combinatorial challenge due to the difficulty of matching adjacent pieces. Instead, we infer a mental image from all pieces, which a given piece can then be matched against avoiding the combinatorial explosion. Exploiting advancements in Generative Adversarial methods, we learn how to reconstruct the image given a set of unordered pieces, allowing the model to learn a joint embedding space to match an encoding of each piece to the cropped layer of the generator. Therefore we frame the problem as a R@1 retrieval task, and then solve the linear assignment using differentiable Hungarian attention, making the process end-to-end. In doing so our model is puzzle size agnostic, in contrast to prior deep learning methods which are single size. We evaluate on two new large-scale datasets, where our model is on par with deep learning methods, while generalizing to multiple puzzle sizes., Comment: Accepted at International Conference of Image Processing (ICIP22)