Search

This paper presents a method for generating large-scale datasets to improve class-agnostic video segmentation across robots with different form factors. Specifically, we consider the question of whether video segmentation models trained on generic segmentation data could be more effective for particular robot platforms if robot embodiment is factored into the data generation process. To answer this question, a pipeline is formulated for using 3D reconstructions (e.g. from HM3DSem) to generate segmented videos that are configurable based on a robot's embodiment (e.g. sensor type, sensor placement, and illumination source). A resulting massive RGB-D video panoptic segmentation dataset (MVPd) is introduced for extensive benchmarking with foundation and video segmentation models, as well as to support embodiment-focused research in video segmentation. Our experimental findings demonstrate that using MVPd for finetuning can lead to performance improvements when transferring foundation models to certain robot embodiments, such as specific camera placements. These experiments also show that using 3D modalities (depth images and camera pose) can lead to improvements in video segmentation accuracy and consistency. The project webpage is available at https://topipari.com/projects/MVPd, Comment: Accepted in IEEE Robotics and Automation Letters October 2024

A $d$-subgroup of a lattice-ordered group ($\ell$-group) $G$ is a subgroup that contains the principal polars generated by each of its elements. We call $G$ a Martinez $\ell$-group if every convex $\ell$-subgroup of $G$ is a $d$-subgroup (equivalently: $G(a)=a^{\perp \perp}$ for every $a \in G$); known examples include all hyperarchimedean $\ell$-groups and all existentially closed abelian $\ell$-groups. This paper gives new characterizations of the Martinez $\ell$-groups, shows that the abelian Martinez $\ell$-groups form a radical class, investigates a related type of archimedean $\ell$-group called a Yosida $\ell$-group, and uses an analogue of a construction from ring theory, and other methods, to produce new examples of Martinez $\ell$-groups with special properties.

We present the unplanned detection of a white dwarf companion to the star HD 218261 in mid-infrared (10-16 $\mu$m) observations with JWST/MIRI. This star was observed as a calibrator for coronagraphic observations of the exoplanet host HR 8799. HD 218261 B has only previously been detected by Gaia, and only in visible light. We confidently detect the companion in the mid-infrared, where it is less luminous than the primary by a factor of ~10$^4$. The visible and mid-infrared photometry are consistent with a white dwarf of $T_\text{eff}\approx10000$ K, $M\approx0.8 M_\odot$, though observation of its optical spectrum is required to precisely constrain its physical parameters. These results demonstrate that precise mid-infrared photometry of white dwarf companions to bright stars can be obtained with MIRI, opening up new possibilities for studying white dwarfs in close binaries., Comment: 4 pages, 2 figures, 2 tables. Published in MNRAS Letters

We study the real-time remote tracking of a two-state Markov process by an energy harvesting source. The source decides whether to transmit over an unreliable channel based on the state. We formulate this scenario as a Markov decision process (MDP) to determine the optimal transmission policy that minimizes the average Version Innovation Age (VIA) as a performance metric. We demonstrate that the optimal transmission policy is threshold-based, determined by the battery level, source state, and VIA value. We numerically verify the analytical structure of the optimal policy and compare the performance of our proposed policy against two baseline policies across various system parameters, establishing the superior performance of our approach.

Ensuring robust safety measures across a wide range of scenarios is crucial for user-facing systems. While Large Language Models (LLMs) can generate valuable data for safety measures, they often exhibit distributional biases, focusing on common scenarios and neglecting rare but critical cases. This can undermine the effectiveness of safety protocols developed using such data. To address this, we propose a novel framework that integrates active learning with clustering to guide LLM generation, enhancing their representativeness and robustness in safety scenarios. We demonstrate the effectiveness of our approach by constructing a dataset of 5.4K potential safety violations through an iterative process involving LLM generation and an active learner model's feedback. Our results show that the proposed framework produces a more representative set of safety scenarios without requiring prior knowledge of the underlying data distribution. Additionally, data acquired through our method improves the accuracy and F1 score of both the active learner model as well models outside the scope of active learning process, highlighting its broad applicability.

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

Optimal Bayesian decision making on what geoscientific data to acquire requires stating a prior model of uncertainty. Data acquisition is then optimized by reducing uncertainty on some property of interest maximally, and on average. In the context of exploration, very few, sometimes no data at all, is available prior to data acquisition planning. The prior model therefore needs to include human interpretations on the nature of spatial variability, or on analogue data deemed relevant for the area being explored. In mineral exploration, for example, humans may rely on conceptual models on the genesis of the mineralization to define multiple hypotheses, each representing a specific spatial variability of mineralization. More often than not, after the data is acquired, all of the stated hypotheses may be proven incorrect, i.e. falsified, hence prior hypotheses need to be revised, or additional hypotheses generated. Planning data acquisition under wrong geological priors is likely to be inefficient since the estimated uncertainty on the target property is incorrect, hence uncertainty may not be reduced at all. In this paper, we develop an intelligent agent based on partially observable Markov decision processes that plans optimally in the case of multiple geological or geoscientific hypotheses on the nature of spatial variability. Additionally, the artificial intelligence is equipped with a method that allows detecting, early on, whether the human stated hypotheses are incorrect, thereby saving considerable expense in data acquisition. Our approach is tested on a sediment-hosted copper deposit, and the algorithm presented has aided in the characterization of an ultra high-grade deposit in Zambia in 2023.

Based on the analysis of a large number of high-fidelity nonlinear gyrokinetic simulations, we propose a novel strategy to improve confinement in tokamak plasmas by combining up-down asymmetric flux surface shaping with the Low Momentum Diffusivity (LMD) regime. We show that the intrinsic momentum flux driven by up-down asymmetry creates strong flow shear in the LMD regime that can significantly reduce energy transport, increasing the critical gradient by up to $25\%$. In contrast to traditional methods for generating flow shear, such as neutral beam injection, this approach requires no external momentum source and is expected to scale well to large fusion devices. The experimental applicability of this strategy in spherical tokamaks is addressed via simulations by considering actual equilibria from MAST and a preliminary equilibrium from SMART., Comment: 7 pages, 5 figures

The atmospheric characterisation of GJ1214 b has so far remained uncertain due to the observed flatness of the transit spectra of this planet that is typically attributed to the presence of hazes or clouds in its atmosphere. Here we combine for the first time transit and eclipse observations obtained with JWST to benefit from both type of constraints and advance on the atmospheric characterisation of GJ1214 b. Our results reveal that photochemical hazes can be produced at high enough mass fluxes in the atmosphere of GJ1214 b to explain both type of observations. These hazes have a drastic impact on the atmospheric thermal structure, which has further ramifications on the emitted radiation of the planet, as well as, its Bond albedo. Clouds of KCL, NaCL and ZnS composition also form in this atmosphere but their opacity is too small to explain the observed flatness of the transit spectrum. We find that metallicities in the range 2000-3000x solar provide atmospheric structures that are closest to the observations for haze mass fluxes in the range of (1-3)x1E-11 g cm-2 s-1. Correspondingly the Bond albedo is within 10-20%. Moreover, sulfur photochemistry produces abundant OCS that has a detectable signature in the transit spectra and should be seaked for in future observations. Sulfur should also participate to the haze formation in this atmosphere, therefore optical properties of such compounds are needed.

The Fourier Transform (FT) is a fundamental tool that permeates modern science and technology. While chemistry undergraduates encounter the FT as early as second year, their courses often only mention it in passing because computers frequently perform it automatically behind the scenes. Although this automation enables students to focus on `the chemistry', students miss out on an opportunity to understand and use one of the most powerful tools in the scientific arsenal capable of revealing how time-dependent signals encode chemical structure. Although many educational resources introduce chemists to the FT, they often require familiarity with sophisticated mathematical and computational concepts. Here, we present a series of three self-contained, Python-based laboratory activities for undergraduates to understand the FT and apply it to analyze audio signals, an infrared (IR) spectroscopy interferogram, and a nuclear magnetic resonance (NMR) free induction decay (FID). In these activities, students observe how the FT reveals and quantifies the contribution of each frequency present in a temporal signal and how decay timescales dictate signal broadening. Our activities empower students with the tools to transform their own temporal datasets (e.g., FID) to a frequency spectrum. To ensure accessibility of the activities and lower the barrier to implementation, we utilize Google Colab's open-source, cloud-based platform to run Jupyter notebooks. We also offer a pre-laboratory activity that introduces students to the basics of Python and the Colab platform, and reviews the math and programming skills needed to complete the lab activities. These lab activities help students build a qualitative, quantitative, and practical understanding of the FT., Comment: 11 pages, 5 figures

This article examines the problem of area coverage for a network of mobile robots with imprecise agent localization. Each robot has uniform radial sensing ability, governed by first order kinodynamics. The convex-space is partitioned based on the Guaranteed Voronoi (GV) principle and each robot's area of responsibility corresponds to its GV-cell, bounded by hyperbolic arcs. The proposed control law is distributed, demands the positioning information about its GV-Delaunay neighbors and has an inherent collision avoidance property., Comment: In proceedings of the 24th Mediterranean Conference on Control and Automation, 2016. 6 pages, 10 figures, video available at https://sotiris.papatheodorou.xyz/papers/2016_MED_PST/2016_MED_PST.mp4

Classical Wolf-Rayet stars are the descendants of massive OB stars that have lost their hydrogen envelopes and are burning helium in their cores prior to exploding as type Ib/c supernovae. The mechanisms for losing their hydrogen envelopes are either through binary interactions or through strong stellar winds potentially coupled with episodic mass-loss. Amongst the bright classical WR stars, the binary system WR\,137 (HD\,192641; WC7d + O9e) is the subject of this paper. This binary is known to have a 13-year period and produces dust near periastron. Here we report on interferometry with the CHARA Array collected over a decade of time and providing the first visual orbit for the system. We combine these astrometric measurements with archival radial velocities to measure masses of the stars of $M_{\rm WR} = 9.5\pm3.4 M_\odot$ and $M_{\rm O} = 17.3\pm 1.9 M_\odot$ when we use the most recent \textit{Gaia} distance. These results are then compared to predicted dust distribution using these orbital elements, which match the observed imaging from \textit{JWST} as discussed recently by Lau et al. Furthermore, we compare the system to the BPASS models, finding that the WR star likely formed through stellar winds and not through binary interactions. However, the companion O star did likely accrete some material from the WR's mass-loss to provide the rotation seen today that drives its status as an Oe star., Comment: Accepted to ApJ

The Large Magellanic Cloud (LMC) is a Milky Way (MW) satellite that is massive enough to gravitationally attract the MW disc and inner halo, causing significant motion of the inner MW with respect to the outer halo. In this work, we probe this interaction by constructing a sample of 9,866 blue horizontal branch (BHB) stars with radial velocities from the DESI spectroscopic survey out to 120 kpc from the Galactic centre. This is the largest spectroscopic set of BHB stars in the literature to date, and it contains four times more stars with Galactocentric distances beyond 50 kpc than previous BHB catalogues. Using the DESI BHB sample combined with SDSS BHBs, we measure the bulk radial velocity of stars in the outer halo and observe that the velocity in the Southern Galactic hemisphere is different by 3.7$\sigma$ from the North. Modelling the projected velocity field shows that its dipole component is directed at a point 22 degrees away from the LMC along its orbit, which we interpret as the travel direction of the inner MW. The velocity field includes a monopole term that is -24 km/s, which we refer to as compression velocity. This velocity is significantly larger than predicted by the current models of the MW and LMC interaction. This work uses DESI data from its first two years of observations, but we expect that with upcoming DESI data releases, the sample of BHB stars will increase and our ability to measure the MW-LMC interaction will improve significantly., Comment: 22 pages, 19 figures. Submitted to MNRAS

Dust from core-collapse supernovae (CCSNe), specifically Type IIP SNe, has been suggested to be a significant source of the dust observed in high-redshift galaxies. CCSNe eject large amounts of newly formed heavy elements, which can condense into dust grains in the cooling ejecta. However, infrared (IR) observations of typical CCSNe generally measure dust masses that are too small to account for the dust production needed at high redshifts. Type IIn SNe, classified by their dense circumstellar medium (CSM), are also known to exhibit strong IR emission from warm dust, but the dust origin and heating mechanism have generally remained unconstrained because of limited observational capabilities in the mid-IR. Here, we present a JWST/MIRI Medium Resolution Spectrograph (MRS) spectrum of the Type IIn SN 2005ip nearly 17 years post-explosion. The Type IIn SN 2005ip is one of the longest-lasting and most well-studied SNe observed to date. Combined with a Spitzer mid-IR spectrum of SN 2005ip obtained in 2008, this data set provides a rare 15-year baseline, allowing for a unique investigation of the evolution of dust. The JWST spectrum shows a new high-mass dust component ($\gtrsim0.08$ M$_{\odot}$) that is not present in the earlier Spitzer spectrum. Our analysis shows dust likely formed over the past 15 years in the cold, dense shell (CDS), between the forward and reverse shocks. There is also a smaller mass of carbonaceous dust ($\gtrsim0.005$ M$_{\odot}$) in the ejecta. These observations provide new insights into the role of SN dust production, particularly within the CDS, and its potential contribution to the rapid dust enrichment of the early Universe.

We explore the transformative potential of SAM 2, a vision foundation model, in advancing gaze estimation and eye tracking technologies. By significantly reducing annotation time, lowering technical barriers through its ease of deployment, and enhancing segmentation accuracy, SAM 2 addresses critical challenges faced by researchers and practitioners. Utilizing its zero-shot segmentation capabilities with minimal user input-a single click per video-we tested SAM 2 on over 14 million eye images from diverse datasets, including virtual reality setups and the world's largest unified dataset recorded using wearable eye trackers. Remarkably, in pupil segmentation tasks, SAM 2 matches the performance of domain-specific models trained solely on eye images, achieving competitive mean Intersection over Union (mIoU) scores of up to 93% without fine-tuning. Additionally, we provide our code and segmentation masks for these widely used datasets to promote further research., Comment: Virmarie Maquiling and Sean Anthony Byrne contributed equally to this paper, 8 pages, 3 figures, CHI Case Study, pre-print

Initially designed to detect and characterize exoplanets, extreme adaptive optics systems (AO) open a new window on the solar system by resolving its small bodies. Nonetheless, despite the always increasing performances of AO systems, the correction is not perfect, degrading their image and producing a bright halo that can hide faint and close moons. Using a reference point spread function (PSF) is not always sufficient due to the random nature of the turbulence. In this work, we present our method to overcome this limitation. It blindly reconstructs the AO-PSF directly in the data of interest, without any prior on the instrument nor the asteroid's shape. This is done by first estimating the PSF core parameters under the assumption of a sharp-edge and flat object, allowing the image of the main body to be deconvolved. Then, the PSF faint extensions are reconstructed with a robust penalization optimization, discarding outliers on-the-fly such as cosmic rays, defective pixels and moons. This allows to properly model and remove the asteroid's halo. Finally, moons can be detected in the residuals, using the reconstructed PSF and the knowledge of the outliers learned with the robust method. We show that our method can be easily applied to different instruments (VLT/SPHERE, Keck/NIRC2), efficiently retrieving the features of AO-PSFs. Compared with state-of-the-art moon enhancement algorithms, moon signal is greatly improved and our robust detection method manages to discriminate faint moons from outliers., Comment: arXiv admin note: text overlap with arXiv:2407.21548

In the context of the GRAVITY+ upgrade, the adaptive optics (AO) systems of the GRAVITY interferometer are undergoing a major lifting. The current CILAS deformable mirrors (DM, 90 actuators) will be replaced by ALPAO kilo-DMs (43x43, 1432 actuators). On top of the already existing 9x9 Shack-Hartmann wavefront sensors (SH-WFS) for infrared (IR) natural guide star (NGS), new 40x40 SH-WFSs for visible (VIS) NGS will be deployed. Lasers will also be installed on the four units of the Very Large Telescope to provide a laser guide star (LGS) option with 30x30 SH-WFSs and with the choice to either use the 9x9 IR-WFSs or 2x2 VIS-WFSs for low order sensing. Thus, four modes will be available for the GRAVITY+ AO system (GPAO): IR-NGS, IR-LGS, VIS-NGS and VIS-LGS. To prepare the instrument commissioning and help the observers to plan their observations, a tool is needed to predict the performances of the different modes and for different observing conditions (NGS magnitude, science object magnitude, turbulence conditions...) We developed models based on a Mar{\'e}chal approximation to predict the Strehl ratio of the four GPAO modes in order to feed the already existing tool that simulates the GRAVITY performances. Waiting for commissioning data, our model was validated and calibrated using the TIPTOP toolbox, a Point Spread Function simulator based on the computation of Power Spectrum Densities. In this work, we present our models of the NGS modes of GPAO and their calibration with TIPTOP.

The High Energy Modular Array (HEMA) is one of three instruments that compose the STROBE-X mission concept. The HEMA is a large-area, high-throughput non-imaging pointed instrument based on the Large Area Detector developed as part of the LOFT mission concept. It is designed for spectral timing measurements of a broad range of sources and provides a transformative increase in sensitivity to X-rays in the energy range of 2--30 keV compared to previous instruments, with an effective area of 3.4 m$^{2}$ at 8.5 keV and an energy resolution of better than 300 eV at 6 keV in its nominal field of regard., Comment: 16 pages, 10 figures

This paper presents the NSX front-end ASIC, being developed to read charge signals from the HEMA and WFM X-ray detectors for the STROBE-X mission. The ASIC reads out signals from up to 64 anodes of linear Silicon Drift Detectors (SDDs). When unloaded, the ASIC channel has a charge resolution, expressed in Equivalent Noise Charge (ENC) of about 2.8 e-. Once connected to the SDD anode we anticipate, for the 80 keV energy range, a ENC of about 10.7 e- at a leakage current of 2 pA, which corresponds to a FWHM of about 145 eV at 6 keV once the Fano-limited statistics from charge generation in Si is included. The acquisition is event-triggered and, for events exceeding the threshold, the ASIC measures the peak amplitude and stores it in an analog memory for subsequent readout. The ASIC can also force the measurement of the sub-threshold channels neighboring the triggered channel, including the ones that belong to neighbor chips by using bi-directional differential inter-chip communication. Alternatively, the ASIC can measure the amplitudes of all channels at the time of the first detected peak. Additional features include a high-resolution option, channel power down and skip function, a low-noise pulse generator, a temperature sensor, and the monitoring of the channel analog output and trimmed threshold. The power consumption of the individual channel is ~590 $\mu$W and, when including all shared circuits, it averages to ~670 $\mu$W / channel., Comment: 16 pages, 12 figures, accepted for publication in JATIS

Adversarial example attacks have emerged as a critical threat to machine learning. Adversarial attacks in image classification abuse various, minor modifications to the image that confuse the image classification neural network -- while the image still remains recognizable to humans. One important domain where the attacks have been applied is in the automotive setting with traffic sign classification. Researchers have demonstrated that adding stickers, shining light, or adding shadows are all different means to make machine learning inference algorithms mis-classify the traffic signs. This can cause potentially dangerous situations as a stop sign is recognized as a speed limit sign causing vehicles to ignore it and potentially leading to accidents. To address these attacks, this work focuses on enhancing defenses against such adversarial attacks. This work shifts the advantage to the user by introducing the idea of leveraging historical images and majority voting. While the attacker modifies a traffic sign that is currently being processed by the victim's machine learning inference, the victim can gain advantage by examining past images of the same traffic sign. This work introduces the notion of ''time traveling'' and uses historical Street View images accessible to anybody to perform inference on different, past versions of the same traffic sign. In the evaluation, the proposed defense has 100% effectiveness against latest adversarial example attack on traffic sign classification algorithm.

We study the Laplacian of the undirected De Bruijn graph over an alphabet $A$ of order $k$. While the eigenvalues of this Laplacian were found in 1998 by Delorme and Tillich [1], an explicit description of its eigenvectors has remained elusive. In this work, we find these eigenvectors in closed form and show that they yield a natural and canonical basis for the cut- and cycle-spaces of De Bruijn graphs. Remarkably, we find that the cycle basis we construct is a basis for the cycle space of both the undirected and the directed De Bruijn graph. This is done by developing an analogue of the Fourier transform on the De Bruijn graph, which acts to diagonalize the Laplacian. Moreover, we show that the cycle-space of De Bruijn graphs, when considering all possible orders of $k$ simultaneously, contains a rich algebraic structure, that of a graded Hopf algebra.

In this article, we present a new linear independence criterion for values of the $p$-adic polygamma functions defined by J.~Diamond. As an application, we obtain the linear independence of some families of values of the $p$-adic Hurwitz zeta function $\zeta_p(s,x)$ at distinct shifts $x$. This improves and extends a previous result due to P.~Bel [5], as well as irrationality results established by F.~Beukers [7]. Our proof is based on a novel and explicit construction of Pad\'{e}-type approximants of the second kind of Diamond's $p$-adic polygamma functions. This construction is established by using a difference analogue of the Rodrigues formula for orthogonal polynomials., Comment: 45pages, 2figures

Performances of an adaptive optics (AO) system are directly linked with the quality of its alignment. During the instrument calibration, having open loop fast tools with a large capture range are necessary to quickly assess the system misalignment and to drive it towards a state allowing to close the AO loop. During operation, complex systems are prone to misalignments (mechanical flexions, rotation of optical elements, etc.) that potentially degrade the AO performances, creating a need for a monitoring tool to tackle their driftage. In this work, we first present an improved perturbative method to quickly assess large lateral errors in open loop. It uses the spatial correlation of the measured interaction matrix of a limited number of 2D spatial modes with a synthetic model. Then, we introduce a novel solution to finely measure and correct these lateral errors via the closed loop telemetry. Non-perturbative, this method consequently does not impact the science output of the instrument. It is based on the temporal correlation of 2D spatial frequencies in the deformable mirror commands. It is model-free (no need of an interaction matrix model) and sparse in the Fourier space, making it fast and easily scalable to complex systems such as future extremely large telescopes. Finally, we present some results obtained on the development bench of the GRAVITY+ extreme AO system (Cartesian grid, 1432 actuators). In addition, we show with on-sky results gathered with CHARA and GRAVITY/CIAO that the method is adaptable to non-conventional AO geometries (hexagonal grids, 60 actuators).

JWST spectroscopy has discovered a population of $z \gtrsim 3.5$ galaxies with broad Balmer emission lines, and narrow forbidden lines, that are consistent with hosting active galactic nuclei (AGN). Many of these systems, now known as ``little red dots" (LRDs), are compact and have unique colors that are very red in the optical/near-infrared and blue in the ultraviolet. The relative contribution of galaxy starlight and AGN to these systems remains uncertain, especially for the galaxies with unusual blue+red spectral energy distributions. In this work, we use Balmer decrements to measure the independent dust attenuation of the broad and narrow emission-line components of a sample of 29 broad-line AGN identified from three public JWST spectroscopy surveys: CEERS, JADES, and RUBIES. Stacking the narrow components from the spectra of 25 sources with broad H$\rm{\alpha}$ and no broad H$\rm{\beta}$ results in a median narrow H$\rm{\alpha}$/H$\rm{\beta}$ = $2.47^{+0.05}_{-0.05}$ (consistent with $A_{v} = 0$) and broad H$\rm{\alpha}$/H$\rm{\beta}$ $> 8.85$ ($A_{v} > 3.63$). The narrow and broad Balmer decrements imply little-to-no attenuation of the narrow emission lines, which are consistent with being powered by star formation and located on larger physical scales. Meanwhile, the lower limit in broad H$\rm{\alpha}$/H$\rm{\beta}$ decrement, with broad H$\rm{\beta}$ undetected in the stacked spectrum of 25 broad-H$\rm{\alpha}$ AGN, implies significant dust attenuation of the broad-line emitting region that is presumably associated with the central AGN. Our results indicate that these systems, on average, are consistent with heavily dust-attenuated AGN powering the red parts of their SED while their blue UV emission is powered by unattenuated star formation in the host galaxy., Comment: 4 figures, 3 tables

Failed supernovae (SNe), which are likely the main channel for forming stellar-mass black holes, are predicted to accompany mass ejections much weaker than typical core-collapse SNe. We conduct a grid of one-dimensional radiation hydrodynamical simulations to explore the emission of failed SNe from red supergiant progenitors, leveraging recent understanding of the weak explosion and the dense circumstellar matter (CSM) surrounding these stars. We find from these simulations and semi-analytical modeling that diffusion in the CSM prolongs the early emission powered by shock breakout/cooling. The early emission has peak luminosities of $\sim 10^7$-$10^8~L_\odot$ in optical and UV, and durations of days to weeks. The presence of dense CSM aids detection of the early bright peak from these events via near-future wide-field surveys such as Rubin Observatory, ULTRASAT and UVEX., Comment: 17 pages, 8 figures, accepted to ApJ. Model light curves available here: https://github.com/DTsuna/failedSNeLCs

Large language models (LLMs), like ChatGPT, have shown that even trained with noisy prior data, they can generalize effectively to new tasks through in-context learning (ICL) and pre-training techniques. Motivated by this, we explore whether a similar approach can be applied to scientific foundation models (SFMs). Our methodology is structured as follows: (i) we collect low-cost physics-informed neural network (PINN)-based approximated prior data in the form of solutions to partial differential equations (PDEs) constructed through an arbitrary linear combination of mathematical dictionaries; (ii) we utilize Transformer architectures with self and cross-attention mechanisms to predict PDE solutions without knowledge of the governing equations in a zero-shot setting; (iii) we provide experimental evidence on the one-dimensional convection-diffusion-reaction equation, which demonstrate that pre-training remains robust even with approximated prior data, with only marginal impacts on test accuracy. Notably, this finding opens the path to pre-training SFMs with realistic, low-cost data instead of (or in conjunction with) numerical high-cost data. These results support the conjecture that SFMs can improve in a manner similar to LLMs, where fully cleaning the vast set of sentences crawled from the Internet is nearly impossible.

This work generalizes the classical metriplectic formalism to model Hamiltonian systems with nonconservative dissipation. Classical metriplectic representations allow for the description of energy conservation and production of entropy via a suitable selection of an entropy function and a bilinear symmetric metric. By relaxing the Casimir invariance requirement of the entropy function, this paper shows that the generalized formalism induces the free energy analogous to thermodynamics. The monotonic change of free energy can serve as a more precise criterion than mechanical energy or entropy alone. This paper provides examples of the generalized metriplectic system in a 2-dimensional Hamiltonian system and $\mathrm{SO}(3)$. This paper also provides a bilevel convex optimization approach for the identification of the metriplectic system given measurements of the system.

A machine-learnable variational scheme using Gaussian radial basis functions (GRBFs) is presented and used to approximate linear problems on bounded and unbounded domains. In contrast to standard mesh-free methods, which use GRBFs to discretize strong-form differential equations, this work exploits the relationship between integrals of GRBFs, their derivatives, and polynomial moments to produce exact quadrature formulae which enable weak-form expressions. Combined with trainable GRBF means and covariances, this leads to a flexible, generalized Galerkin variational framework which is applied in the infinite-domain setting where the scheme is conforming, as well as the bounded-domain setting where it is not. Error rates for the proposed GRBF scheme are derived in each case, and examples are presented demonstrating utility of this approach as a surrogate modeling technique.

Purpose: To present a novel generalized MR image reconstruction based on pseudoinversion of the encoding matrix (Pinv-Recon) as a simple yet powerful method, and demonstrate its computational feasibility for diverse MR imaging applications. Methods: MR image encoding constitutes a linear mapping of the unknown image to the measured k-space data mediated via an encoding matrix ($ data = Encode \times image$). Pinv-Recon addresses MR image reconstruction as a linear inverse problem ($image = Encode^{-1} \times data$), explicitly calculating the Moore-Penrose pseudoinverse of the encoding matrix using truncated singular value decomposition (tSVD). Using a discretized, algebraic notation, we demonstrate constructing a generalized encoding matrix by stacking relevant encoding mechanisms (e.g., gradient encoding, coil sensitivity encoding, chemical shift inversion) and encoding distortions (e.g., off-center positioning, B$_0$ inhomogeneity, spatiotemporal gradient imperfections, transient relaxation effects). Iterative reconstructions using the explicit generalized encoding matrix, and the computation of the spatial-response-function (SRF) and noise amplification, were demonstrated. Results: We evaluated the computation times and memory requirements (time ~ (size of the encoding matrix)$^{1.4}$). Using the Shepp-Logan phantom, we demonstrated the versatility of the method for various intertwined MR image encoding and distortion mechanisms, achieving better MSE, PSNR and SSIM metrics than conventional methods. A diversity of datasets, including the ISMRM CG-SENSE challenge, were used to validate Pinv-Recon. Conclusion: Although pseudo-inversion of large encoding matrices was once deemed computationally intractable, recent advances make Pinv-Recon feasible. It has great promise for both research and clinical applications, and for educational use., Comment: 26 pages, 8 figures, 2 tables (+ Supplementary Material). Submitted to Magnetic Resonance in Medicine

In this work, we propose a new form of exponential quantum advantage in the context of sensing correlated noise. Specifically, we focus on the problem of estimating parameters associated with Lindblad dephasing dynamics, and show that entanglement can lead to an exponential enhancement in the sensitivity (as quantified via quantum Fisher information of the sensor state) for estimating a small parameter characterizing the deviation of system Lindbladians from a class of maximally correlated dephasing dynamics. This result stands in stark contrast with previously studied scenarios of sensing uncorrelated dephasing noise, where one can prove that entanglement does not lead to an advantage in the signal-to-noise ratio. Our work thus opens a novel pathway towards achieving entanglement-based sensing advantage, which may find applications in characterizing decoherence dynamics of near-term quantum devices. Further, our approach provides a potential quantum-enhanced probe of many-body correlated phases by measuring noise generated by a sensing target. We also discuss realization of our protocol using near-term quantum hardware., Comment: 7+2 pages, 1 figure

The local controllability of a rich class of affine nonlinear control systems with nonhomogeneous quadratic drift and constant control vector fields is analyzed. The interest in this particular class of systems stems from the ubiquity in science and engineering of some of its notable representatives, namely the Sprott system, the Lorenz system and the rigid body among others. A necessary and sufficient condition for strong accessibility reminiscent of the Kalman rank condition is derived, and it generalizes Crouch's condition for the rigid body. This condition is in general not sufficient to infer small-time local controllability. However, under some additional mild assumptions local controllability is established. In particular for the Sprott and Lorenz systems, sharp conditions for small-time local controllability are obtained in the single-input case., Comment: 18 pages, 2 figures

In recent years, there has been considerable interest in designing stochastic first-order algorithms to tackle finite-sum smooth minimax problems. To obtain the gradient estimates, one typically relies on the uniform sampling-with-replacement scheme or various sampling-without-replacement (also known as shuffling) schemes. While the former is easier to analyze, the latter often have better empirical performance. In this paper, we propose a novel single-loop stochastic gradient descent-ascent (GDA) algorithm that employs both shuffling schemes and variance reduction to solve nonconvex-strongly concave smooth minimax problems. We show that the proposed algorithm achieves $\epsilon$-stationarity in expectation in $\mathcal{O}(\kappa^2 \epsilon^{-2})$ iterations, where $\kappa$ is the condition number of the problem. This outperforms existing shuffling schemes and matches the complexity of the best-known sampling-with-replacement algorithms. Our proposed algorithm also achieves the same complexity as that of its deterministic counterpart, the two-timescale GDA algorithm. Our numerical experiments demonstrate the superior performance of the proposed algorithm.

This study introduces a novel frequency-based data-driven controller for adaptive optics, using power spectral density for optimization while ensuring stability criteria. It addresses disturbance rejection, command amplitude constraints and system transfer functions through convex optimization to obtain an optimal control in an infinite input response filter form. Evaluated within the SAXO+ project, it demonstrates efficacy under diverse atmospheric conditions and operational scenarios. The proposed controller is tested in both standard and disentangled adaptive optics schemes, showcasing its adaptability and performance. Experimental validation is conducted using the COMPASS simulation tool, affirming the controller's promise for enhancing adaptive optics systems in real-world applications.

This paper studies the nonsmooth optimization landscape of the $\ell_1$-norm rank-one symmetric matrix factorization problem using tools from second-order variational analysis. Specifically, as the main finding of this paper, we show that any second-order stationary point (and thus local minimizer) of the problem is actually globally optimal. Besides, some other results concerning the landscape of the problem, such as a complete characterization of the set of stationary points, are also developed, which should be interesting in their own rights. Furthermore, with the above theories, we revisit existing results on the generic minimizing behavior of simple algorithms for nonsmooth optimization and showcase the potential risk of their applications to our problem through several examples. Our techniques can potentially be applied to analyze the optimization landscapes of a variety of other more sophisticated nonsmooth learning problems, such as robust low-rank matrix recovery.

In this paper, we study equality-type Clarke subdifferential chain rules of matrix factorization and factorization machine. Specifically, we show for these problems that provided the latent dimension is larger than some multiple of the problem size (i.e., slightly overparameterized) and the loss function is locally Lipschitz, the subdifferential chain rules hold everywhere. In addition, we examine the tightness of the analysis through some interesting constructions and make some important observations from the perspective of optimization; e.g., we show that for all this type of problems, computing a stationary point is trivial. Some tensor generalizations and neural extensions are also discussed, albeit they remain mostly open.

This article is concerned with locally flatly immersed surfaces in simply-connected $4$-manifolds where the complement of the surface has fundamental group $\mathbb{Z}$. Once the genus and number of double points are fixed, we classify such immersed surfaces in terms of the equivariant intersection form of their exterior and a secondary invariant. Applications include criteria for deciding when an immersed $\mathbb{Z}$-surface in $S^4$ is isotopic to the standard immersed surface that is obtained from an unknotted surface by adding local double points. As another application, we enumerate $\mathbb{Z}$-disks in $D^4$ with a single double point and boundary a given knot; we prove that the number of such disks may be infinite. We also prove that a knot bounds a $\mathbb{Z}$-disk in $D^4$ with $c_+$ positive double points and $c_-$ negative double points if and only if it can be converted into an Alexander polynomial one knot via changing $c_+$ positive crossings and $c_-$ negative crossings. In $4$-manifolds other than $D^4$ and $S^4$, applications include measuring the extent to which immersed $\mathbb{Z}$-surfaces are determined by the equivariant intersection form of their exterior. Along the way, we prove that any two $\mathbb{Z}^2$-concordances between the Hopf link and an Alexander polynomial one link $L$ are homeomorphic rel. boundary., Comment: 76 pages, 15 figures

Death among subjects is common in observational studies evaluating the causal effects of interventions among geriatric or severely ill patients. High mortality rates complicate the comparison of the prevalence of adverse events (AEs) between interventions. This problem is often referred to as outcome "truncation" by death. A possible solution is to estimate the survivor average causal effect (SACE), an estimand that evaluates the effects of interventions among those who would have survived under both treatment assignments. However, because the SACE does not include subjects who would have died under one or both arms, it does not consider the relationship between AEs and death. We propose a Bayesian method which imputes the unobserved mortality and AE outcomes for each participant under the intervention they did not receive. Using the imputed outcomes we define a composite ordinal outcome for each patient, combining the occurrence of death and the AE in an increasing scale of severity. This allows for the comparison of the effects of the interventions on death and the AE simultaneously among the entire sample. We implement the procedure to analyze the incidence of heart failure among geriatric patients being treated for Type II diabetes with sulfonylureas or dipeptidyl peptidase-4 inhibitors., Comment: For supplemental materials, see https://github.com/AnthonySisti/Adverse-Effects-Estimation-in-Observational-Studies-with-Truncation-by-Death

Given a triangulation $G$ of a surface $\mathbb{D}$, a spanning disk is a disk $\mathbb{D} \subseteq \mathbb{S}$ containing all the vertices of $G$ such that the boundary of $\mathbb{D}$ is a cycle of $G$. In this paper, we consider the question of when a triangulation of a surface contains a spanning disk. We give a very short proof that every triangulation of the torus contains a spanning disk, which strengthens a theorem of Nevo and Tarabykin. For arbitrary surfaces, we prove that triangulations with sufficiently high facewidth always contain spanning disks. Finally, we exhibit triangulations which do not have spanning disks. This shows that a minimum facewidth condition is necessary. Our results are motivated by and have applications for rigidity questions in the plane., Comment: 7 pages, 2 figures

Competitive equilibrium (CE) for chores has recently attracted significant attention, with many algorithms proposed to approximately compute it. However, existing algorithms either lack iterate convergence guarantees to an exact CE or require solving high-dimensional linear or quadratic programming subproblems. This paper overcomes these issues by proposing a novel unconstrained difference-of-convex formulation, whose stationary points correspond precisely to the CE for chores. We show that the new formulation possesses the local error bound property and the Kurdyka-{\L}ojasiewicz property with an exponent of $1/2$. Consequently, we present the first algorithm whose iterates provably converge linearly to an exact CE for chores. Furthermore, by exploiting the max structure within our formulation and applying smoothing techniques, we develop a subproblem-free algorithm that finds an approximate CE in polynomial time. Numerical experiments demonstrate that the proposed algorithms outperform the state-of-the-art method., Comment: Accepted by WINE 2024

Large language models (LLMs) are stochastic, and not all models give deterministic answers, even when setting temperature to zero with a fixed random seed. However, few benchmark studies attempt to quantify uncertainty, partly due to the time and cost of repeated experiments. We use benchmarks designed for testing LLMs' capacity to reason about cardinal directions to explore the impact of experimental repeats on mean score and prediction interval. We suggest a simple method for cost-effectively quantifying the uncertainty of a benchmark score and make recommendations concerning reproducible LLM evaluation., Comment: 4 pages, 1 figure

Semantic text embedding is essential to many tasks in Natural Language Processing (NLP). While black-box models are capable of generating high-quality embeddings, their lack of interpretability limits their use in tasks that demand transparency. Recent approaches have improved interpretability by leveraging domain-expert-crafted or LLM-generated questions, but these methods rely heavily on expert input or well-prompt design, which restricts their generalizability and ability to generate discriminative questions across a wide range of tasks. To address these challenges, we introduce \algo{CQG-MBQA} (Contrastive Question Generation - Multi-task Binary Question Answering), a general framework for producing interpretable semantic text embeddings across diverse tasks. Our framework systematically generates highly discriminative, low cognitive load yes/no questions through the \algo{CQG} method and answers them efficiently with the \algo{MBQA} model, resulting in interpretable embeddings in a cost-effective manner. We validate the effectiveness and interpretability of \algo{CQG-MBQA} through extensive experiments and ablation studies, demonstrating that it delivers embedding quality comparable to many advanced black-box models while maintaining inherently interpretability. Additionally, \algo{CQG-MBQA} outperforms other interpretable text embedding methods across various downstream tasks., Comment: 19 pages, 5 figures, and 9 tables

In this paper we use level set-based topology optimisation to design three-dimensional periodic piezoelectric materials with enhanced properties. Our methodology is fully memory-distributed and written in Julia using the package GridapTopOpt. We compare and assess several existing iterative solvers with respect to their weak scalability and find that an approximate Schur complement preconditioned GMRES method demonstrates the best performance and scalability for solving the piezoelectric homogenisation equations. We use the developed techniques to computationally design high-resolution piezoelectric metamaterials with enhanced stiffness and piezoelectric properties that yield new insights into material design for sensor, hydrophone, and actuator applications. We suggest two robust structures with simple geometric features that exhibit enhanced piezoelectric properties several times larger than those of the base material. We find that level set-based topology optimisation is well suited to problems involving piezoelectricity and has the advantage of avoiding large regions of intermediate density material.

In this paper, we present a novel approach to coal mining question answering (QA) using large language models (LLMs) combined with tailored prompt engineering techniques. Coal mining is a complex, high-risk industry where accurate, context-aware information is critical for safe and efficient operations. Current QA systems struggle to handle the technical and dynamic nature of mining-related queries. To address these challenges, we propose a multi-turn prompt engineering framework designed to guide LLMs, such as GPT-4, in answering coal mining questions with higher precision and relevance. By breaking down complex queries into structured components, our approach allows LLMs to process nuanced technical information more effectively. We manually curated a dataset of 500 questions from real-world mining scenarios and evaluated the system's performance using both accuracy (ACC) and GPT-4-based scoring metrics. Experiments comparing ChatGPT, Claude2, and GPT-4 across baseline, chain-of-thought (CoT), and multi-turn prompting methods demonstrate that our method significantly improves both accuracy and contextual relevance, with an average accuracy improvement of 15-18\% and a notable increase in GPT-4 scores. The results show that our prompt-engineering approach provides a robust, adaptable solution for domain-specific question answering in high-stakes environments like coal mining.

Detectors at future high energy colliders will face enormous technical challenges. Disentangling the unprecedented numbers of particles expected in each event will require highly granular silicon pixel detectors with billions of readout channels. With event rates as high as 40 MHz, these detectors will generate petabytes of data per second. To enable discovery within strict bandwidth and latency constraints, future trackers must be capable of fast, power efficient, and radiation hard data-reduction at the source. We are developing a radiation hard readout integrated circuit (ROIC) in 28nm CMOS with on-chip machine learning (ML) for future intelligent pixel detectors. We will show track parameter predictions using a neural network within a single layer of silicon and hardware tests on the first tape-outs produced with TSMC. Preliminary results indicate that reading out featurized clusters from particles above a modest momentum threshold could enable using pixel information at 40 MHz., Comment: Contribution to the 42nd International Conference on High Energy Physics (ICHEP)

Optimal transport (OT) has an important role in transforming data distributions in a manner which engenders fairness. Typically, the OT operators are learnt from the unfair attribute-labelled data, and then used for their repair. Two significant limitations of this approach are as follows: (i) the OT operators for underrepresented subgroups are poorly learnt (i.e. they are susceptible to representation bias); and (ii) these OT repairs cannot be effected on identically distributed but out-of-sample (i.e.\ archival) data. In this paper, we address both of these problems by adopting a Bayesian nonparametric stopping rule for learning each attribute-labelled component of the data distribution. The induced OT-optimal quantization operators can then be used to repair the archival data. We formulate a novel definition of the fair distributional target, along with quantifiers that allow us to trade fairness against damage in the transformed data. These are used to reveal excellent performance of our representation-bias-tolerant scheme in simulated and benchmark data sets.

It is generally assumed that scalar field dark matter halos would contain solitonic cores -- spherically symmetric ground state configurations -- at their centers. This is especially interesting in the case of ultralight dark matter (ULDM), where the solitons sizes are on the order of galaxies. In this work, we show that the paradigm of a spherically symmetric soliton embedded in the center of each halo is not universally valid in a scenario with multiple interacting scalar fields. In particular, sufficiently strong repulsive interspecies interactions make the fields immiscible. In such models, the ground state configuration can fall into a number of different phases that depend on the fields' relative densities, masses, and interaction strengths. This raises the possibility that the inner regions of ULDM halos are more complex and diverse than previously assumed.

As Large Language Models (LLMs) become increasingly integrated into our daily lives, the potential harms from deceptive behavior underlie the need for faithfully interpreting their decision-making. While traditional probing methods have shown some effectiveness, they remain best for narrowly scoped tasks while more comprehensive explanations are still necessary. To this end, we investigate meta-models-an architecture using a "meta-model" that takes activations from an "input-model" and answers natural language questions about the input-model's behaviors. We evaluate the meta-model's ability to generalize by training them on selected task types and assessing their out-of-distribution performance in deceptive scenarios. Our findings show that meta-models generalize well to out-of-distribution tasks and point towards opportunities for future research in this area. Our code is available at https://github.com/acostarelli/meta-models-public ., Comment: 11 pages, 2 figures

Given $n$ buttons and $n$ bulbs so that the $i$th button toggles the $i$th bulb and perhaps some other bulbs, we compute the sharp lower bound on the number of bulbs that can be lit regardless of the action of the buttons. In the previous article we dealt with the case where each button affects at most 2 or 3 bulbs. In the present article we give sharp lower bounds for up to 4 or 5 wires per switch, and we show that the sharp asymptotic bound for an arbitrary number of wires is $\frac12$. (Even if you've found their buttons, you can please no more than half the people all the time!), Comment: 22 pages, 6 figures

We compute explicit reductions of crystalline representations of the absolute Galois group $\text{Gal}(\overline{\mathbb{Q}}_p/\mathbb{Q}_{p^f})$ with labeled Hodge-Tate weights in the range $p+2\le k_{0}\le 2p-4$ and $2\le k_i\le p-3$ for $1\le i\le f-1$.

The task of action spotting consists in both identifying actions and precisely localizing them in time with a single timestamp in long, untrimmed video streams. Automatically extracting those actions is crucial for many sports applications, including sports analytics to produce extended statistics on game actions, coaching to provide support to video analysts, or fan engagement to automatically overlay content in the broadcast when specific actions occur. However, before 2018, no large-scale datasets for action spotting in sports were publicly available, which impeded benchmarking action spotting methods. In response, our team built the largest dataset and the most comprehensive benchmarks for sports video understanding, under the umbrella of SoccerNet. Particularly, our dataset contains a subset specifically dedicated to action spotting, called SoccerNet Action Spotting, containing more than 550 complete broadcast games annotated with almost all types of actions that can occur in a football game. This dataset is tailored to develop methods for automatic spotting of actions of interest, including deep learning approaches, by providing a large amount of manually annotated actions. To engage with the scientific community, the SoccerNet initiative organizes yearly challenges, during which participants from all around the world compete to achieve state-of-the-art performances. Thanks to our dataset and challenges, more than 60 methods were developed or published over the past five years, improving on the first baselines and making action spotting a viable option for the sports industry. This paper traces the history of action spotting in sports, from the creation of the task back in 2018, to the role it plays today in research and the sports industry., Comment: 31 pages, 2 figures, 5 tables