Showing 1–50 of 260 results for author: Chandra, V

The Twentieth Data Release of the Sloan Digital Sky Survey: First All-Sky BOSS Spectra, eROSITA-SDSS-V Mapper Coordinated Observations, and a Preview of the Local Volume Mapper

Authors: SDSS Collaboration, Mojgan Aghakhanloo, David Aguilar, James Aird, Andrés Almeida, Bella Abigail Sanabria Alonso, Hillary Diane Andales, Scott F. Anderson, Stefan Arseneau, Consuelo González Ávila, Shir Aviram, Catarina Aydar, Carles Badenes, Carolina Andonie, Jorge K. Barrera-Ballesteros, Franz E. Bauer, Chad Bender, Michelle A. Berg, F. Besser, Binod Bhattarai, Christian Moni Bidin, Jonathan C. Bird, Dmitry Bizyaev, Guillermo A. Blanc, Alexandra Bonkoski , et al. (251 additional authors not shown)

Abstract: This paper presents the twentieth data release (DR20) from the Sloan Digital Sky Survey, the third data release of its fifth generation (SDSS-V). SDSS-V is a panoptic spectroscopy survey that is mapping the stars, gas, and galaxies through three scientific programs: the Milky Way Mapper (MWM), the Local Volume Mapper (LVM), and the Black Hole Mapper (BHM). DR20 presents the first optical (BOSS) SD… ▽ More This paper presents the twentieth data release (DR20) from the Sloan Digital Sky Survey, the third data release of its fifth generation (SDSS-V). SDSS-V is a panoptic spectroscopy survey that is mapping the stars, gas, and galaxies through three scientific programs: the Milky Way Mapper (MWM), the Local Volume Mapper (LVM), and the Black Hole Mapper (BHM). DR20 presents the first optical (BOSS) SDSS-V spectra from southern hemisphere for the MWM and BHM surveys; new optical MWM and BHM data from the northern hemisphere are also available, for a total over 3 million spectra of 1.5 million stars and half a million galaxies and quasars, with galactic and extragalactic x-ray targets coordinate with eROSITA DR2. DR20 includes integral field spectroscopy maps from LVM of six targets and 169 tiles, spanning Galactic HII regions, planetary nebulae, and nearby galaxies. Additionally, eighteen value added catalogs are also released with DR20, based on SDSS-V MWM and BHM data, and we present a new LVM visualization tool including an RGB HiPS map as a value added product. △ Less

Submitted 28 July, 2026; originally announced July 2026.

Comments: 81 pages, 13 figures, 8 tables

BOSS-CLAM: Utilizing a Constrained Linear Absorption Model to Infer Stellar Parameters from BOSS Spectra

Authors: Ilija Medan, Andrew R. Casey, Alexander P. Ji, Jonah M. Otto, Kayvon Sharifi, Zachary Way, Madeleine McKenzie, Natalie R. Myers, Keivan G. Stassun, Peter J. Smith, Andrew Tkachenko, Vedant Chandra, Michael R. Blanton, Peter M. Frinchaboy, Guy S. Stringfellow, Sean Morrison

Abstract: Large spectroscopic surveys require robust pipelines capable of inferring stellar parameters over a wide range of the Hertzsprung-Russell (HR) diagram from data of varying quality. SDSS-V is one such survey, where the data from the lower-resolution, optical BOSS spectrograph will provide a large dataset covering a wide range of Galactic stellar populations. To better analyze these data, we present… ▽ More Large spectroscopic surveys require robust pipelines capable of inferring stellar parameters over a wide range of the Hertzsprung-Russell (HR) diagram from data of varying quality. SDSS-V is one such survey, where the data from the lower-resolution, optical BOSS spectrograph will provide a large dataset covering a wide range of Galactic stellar populations. To better analyze these data, we present BOSS-CLAM, a generative, forward modeling pipeline for inferring effective temperature ($T_\mathrm{eff}$), surface gravity ($\log g$), metallicity ($[\mathrm{Fe/H}]$), and $α-$abundance ($[α/\mathrm{M}]$) from continuum-normalized BOSS spectra. BOSS-CLAM maps stellar labels to Non-negative Matrix Factorization (NMF) basis vector weights via a polynomial mapping jointly optimized with the spectral decomposition, which provides a more flexible framework for working with the lower-resolution BOSS data. Additionally, training labels are drawn from four complementary sources (ASPCAP, BOSS-MINESweeper, wide binaries, and a hot star validation sample), which enables coverage from cool M dwarfs through hot OB stars, and across a wide range of metallicity. We infer parameters for 1,708,214 BOSS spectra, with a recommended clean catalog of 915,514 sources. Validation against open and globular clusters demonstrates homogeneous, accurate abundances across a wide range of metallicity. Wide binary tests yield abundance uncertainties of $σ_{[\mathrm{Fe/H}]} \approx 0.15$ dex and $σ_{[α/\mathrm{M}]} \approx 0.06$ dex at SNR = 10. Finally, we demonstrate that the BOSS-CLAM catalog recovers known chemical structure of the Milky Way disk and is well-suited for Galactic archaeology, chemical tagging, and stellar population modeling. The pipeline, trained model, and catalog are publicly released as part of SDSS-V DR20. △ Less

Submitted 28 July, 2026; v1 submitted 24 July, 2026; originally announced July 2026.

Comments: 28 pages, 16 figures

Gravitational Effective Theories with Maximal Supersymmetry and a Peculiar Parity

Authors: Justin Berman, Simon Caron-Huot, Aditi V. Chandra, Henriette Elvang, Aidan Herderschee, Loki L. Lin, Roger Morales

Abstract: We study the space of four-dimensional ultraviolet completions for $\mathcal{N}=8$ supergravity that are described at low energies by weakly-coupled effective field theories (EFTs) with maximal supersymmetry and $\mathrm{SU}(4)\times\mathrm{SU}(4)$ R-symmetry. We show that tree-level factorization of the 4-, 5-, and 6-point EFT scattering amplitudes, together with a certain ``peculiar parity'' con… ▽ More We study the space of four-dimensional ultraviolet completions for $\mathcal{N}=8$ supergravity that are described at low energies by weakly-coupled effective field theories (EFTs) with maximal supersymmetry and $\mathrm{SU}(4)\times\mathrm{SU}(4)$ R-symmetry. We show that tree-level factorization of the 4-, 5-, and 6-point EFT scattering amplitudes, together with a certain ``peculiar parity'' condition, leads to nonlinear constraints on the 4-point Wilson coefficients. This peculiar parity is a property that can only be imposed on a subset of scalar amplitudes. Combining the nonlinear constraints with positivity, we find that the allowed region of 4-point Wilson coefficients is reduced to a non-convex domain with two sharp corners: one being the closed superstring Virasoro--Shapiro amplitude, the other an infinite spin tower amplitude exchanging states of every spin at the same mass. We show both numerically and analytically that requiring a finite number of states near the first mass level leaves only the Virasoro--Shapiro amplitude. △ Less

Submitted 15 July, 2026; originally announced July 2026.

Comments: 35+9 pages, 2 figures, 3 tables

Report number: LITP-26-13

Rotating magnetized pion gas of finite transverse size: condensation constraints and transport properties

Authors: Ankit Kumar, Diwakar Gaur, Vinod Chandra

Abstract: This work investigates the electric, thermal, and thermoelectric responses of a rotating pion gas of finite transverse radius in the presence of a background magnetic field, with the rotation axis aligned with the magnetic field. We explicitly calculate the parameter limits for $π^+$ condensation and restrict our working regime safely outside these boundaries, ensuring well-behaved transport coeff… ▽ More This work investigates the electric, thermal, and thermoelectric responses of a rotating pion gas of finite transverse radius in the presence of a background magnetic field, with the rotation axis aligned with the magnetic field. We explicitly calculate the parameter limits for $π^+$ condensation and restrict our working regime safely outside these boundaries, ensuring well-behaved transport coefficients. Notably, the system exhibits a condensation asymmetry, with $π^-$ remaining uncondensed at the parameters that induce $π^+$ condensation. Using the Boltzmann Transport Equation under the Relaxation Time Approximation, we calculate the longitudinal electrical conductivity, thermal conductivity, and the Seebeck coefficient. Our results reveal a competing interplay between the magnetic field and rotation, highlighting the substantial impact of rotation on the medium's transport properties: while the magnetic field suppresses the transport coefficients in a static medium, rotation, acting as an effective chemical potential, introduces an energy shift that favors their increase. Beyond an angular velocity, this rotational enhancement overpowers the magnetic suppression, leading to an increase in the transport coefficients with increasing magnetic field. Finally, we analyze the relative significance of charge and heat transport through the Lorenz number, providing further insight into the transport characteristics of the rotating magnetized pion medium. △ Less

Submitted 18 June, 2026; originally announced June 2026.

Comments: 13 pages, 7 figures, two-column

An Ultramassive White Dwarf with a Likely Oxygen-Neon Core

Authors: Stefan M. Arseneau, J. J. Hermes, Vedant Chandra, Roberto Raddi, Maria E. Camisassa, Alberto Rebassa-Mansergas, Santiago Torres

Abstract: The core composition of ultramassive white dwarfs remains an open question in stellar evolution. The carbon content of white dwarf cores is critical to their role as progenitors of Type Ia supernovae. However, because the stellar photosphere only extends to the outermost layer of the star, observational probes of core compositions are limited. Here we present gravitational redshift measurements of… ▽ More The core composition of ultramassive white dwarfs remains an open question in stellar evolution. The carbon content of white dwarf cores is critical to their role as progenitors of Type Ia supernovae. However, because the stellar photosphere only extends to the outermost layer of the star, observational probes of core compositions are limited. Here we present gravitational redshift measurements of an ultramassive white dwarf, SDSS J060851.44-005950.3, which indicate the likely presence of an oxygen-neon core. We measure the mass ($1.226_{-0.025}^{+0.024} M_\odot$) and radius ($0.491_{-0.009}^{+0.009}~R_\oplus$) of the white dwarf using gravitational redshifts from high-resolution UVES and MagE spectra paired with independent constraints from photometry. By comparing to state-of-the-art mass-radius relations for ultramassive white dwarfs, we find preference for a oxygen-neon core over a carbon-oxygen core, with a Bayes factor of $2.7$. This is a white dwarf which is likely structurally incapable of producing a Type Ia supernova, according to current understanding of supernova physics. This object provides evidence that white dwarfs which pass through the Q-branch without experiencing a delay in cooling compared to the normal white dwarf cooling sequence likely have oxygen-neon cores. △ Less

Submitted 17 June, 2026; originally announced June 2026.

Comments: 12 pages, 5 figures, 1 table, accepted for publication in ApJ

Tracing the very early disruption of the Sagittarius dwarf galaxy in the distant Milky Way halo

Authors: Manuel Bayer, Else Starkenburg, Akshara Viswanathan, Vedant Chandra, Alexander P. Ji, Guillaume F. Thomas

Abstract: Current models predict that at distances beyond 80 kpc in the Milky Way halo, we can find the earliest escaped stars from the merging Sagittarius dwarf galaxy. However, observational data on the Sagittarius stream at these distances is limited. This study examines an overdensity of red giant branch (RGB) stars potentially linked to Sagittarius merger debris. Using the Magellan Inamori Kyocera Eche… ▽ More Current models predict that at distances beyond 80 kpc in the Milky Way halo, we can find the earliest escaped stars from the merging Sagittarius dwarf galaxy. However, observational data on the Sagittarius stream at these distances is limited. This study examines an overdensity of red giant branch (RGB) stars potentially linked to Sagittarius merger debris. Using the Magellan Inamori Kyocera Echelle spectrograph of Las Campanas Observatory's Clay Telescope, we measured the radial velocities and metallicities of these stars. We compared their properties with model predictions of Sagittarius' disruption and other stellar tracers from the Dark Energy Spectroscopic Instrument Data Release 1 and RR Lyrae catalogs. Our spectral analysis confirms the significant tight clustering of four of these RGBs in full 6D phase space. This tight clump is embedded within a larger spur-like feature of the Sagittarius stream in the southern sky. A comparison with Sagittarius stream models further strengthens this hypothesis and shows that this far spur could be composed of stars originally in the halo of the Sagittarius dwarf galaxy, stripped in the earliest phases of the interaction. The metallicity dispersion of the four stars of $0.15 ^ {+0.17} _ {-0.08}$ around the average of [Fe/H] = $-1.46 ^ {+0.11} _ {-0.09}$ is very low. This study provides the first spectroscopic view of the distant southern spur of Sagittarius, composed of stars likely stripped from Sagittarius's halo. △ Less

Submitted 16 June, 2026; originally announced June 2026.

Comments: To be published in MNRAS, 18 pages, ten figures, five tables (with appendix of two pages, one table)

A systematic survey for hypervelocity runaways from thermonuclear supernovae

Authors: Kareem El-Badry, Klaus Werner, Ken J. Shen, Jay Strader, Antonio C. Rodriguez, Jiwon Jesse Han, Vedant Chandra, Laura Chomiuk, Zachary P. Vanderbosch, Lisa Blomberg, Natsuko Yamaguchi, Pranav Nagarajan, Ilaria Caiazzo, Jan van Roestel, Hila Glanz, Tin Long Sunny Wong, Aakash Bhat, Mark A. Hollands, Boris T. Gänsicke

Abstract: The explosion of a white dwarf (WD) in a close binary can launch a surviving runaway star at velocities of $\gtrsim 1000\, \rm km\,s^{-1}$. Such runaways provide a direct probe of thermonuclear supernovae (SNe) in double-degenerate binaries. Several candidate runaways are known, but their evolutionary states and the demographics of the broader population are uncertain. To enable robust population… ▽ More The explosion of a white dwarf (WD) in a close binary can launch a surviving runaway star at velocities of $\gtrsim 1000\, \rm km\,s^{-1}$. Such runaways provide a direct probe of thermonuclear supernovae (SNe) in double-degenerate binaries. Several candidate runaways are known, but their evolutionary states and the demographics of the broader population are uncertain. To enable robust population inference, we carry out a systematic survey for hypervelocity runaways with a simple selection function, selecting candidates based on large Gaia-inferred tangential velocities and blue colors. We classify 100% of the resulting 92 candidates using a combination of spectroscopic follow-up and archival data. The search yields ten suspected D$^6$ stars and three LP 40-365 stars. Three D$^6$ stars are new discoveries, including two hot ($T_{\rm eff} > 50,000$ K) objects and one cool ($T_{\rm eff}\approx 7,000$ K) object. We forward-model our survey under several proposed D$^6$ star evolutionary models, coupling each to a Galactic model and the survey selection function. No single model reproduces the observed diversity of D$^6$ stars, which likely reflects a range of remnant masses, ages, and heating mechanisms. Models in which runaway companions are heated by SN shocks alone are too faint and short-lived to explain most of the observed sample, while fully reheated models are too luminous and long-lived. Models with intermediate heating, as occurs in some simulations of violent mergers and partially disrupted remnants, best match the observed magnitude, distance, and kinematic-age distributions. The inferred D$^6$ star birth rate is model dependent, but the models that best match the observed population require rates of only a few percent of the Galactic SN Ia rate, perhaps implying that most SNe Ia result from WD binaries in which both components explode. △ Less

Submitted 26 June, 2026; v1 submitted 9 June, 2026; originally announced June 2026.

Comments: 22 pages, 11 figures, accepted to OJAp. Data at https://doi.org/10.22002/aff3x-28035

VLM3: Vision Language Models Are Native 3D Learners

Authors: Zhipeng Cai, Zhuang Liu, Yunyang Xiong, Zechun Liu, Vikas Chandra, Yangyang Shi

Abstract: Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) foc… ▽ More Vision Language Models (VLMs) enable a unified model to solve various vision tasks through prompting. They have shown promising performance in semantic understanding. However, 3D understanding still largely relies on expert vision models with complex task-specific designs. The key argument this work wants to make is that VLMs are native 3D learners. Our in-depth large scale study shows that 1) focal length unification, 2) text-based pixel reference and 3) data mixture and scaling, are all you need for effective 3D learning. Model architecture changes, large models, heavy data augmentations, and complex losses including the regression formulation, many of which form the foundation of expert vision models, are actually not necessary conditions. As a result, we propose VLM3, a scalable method with the simplest design that enables standard VLMs to master diverse 3D tasks. VLM3 not only advances the VLM depth estimation accuracy by a large margin (0.84 -> 0.9), but also enables diverse 3D tasks such as pixel correspondence, camera pose estimation and object-level 3D understanding, matching expert vision model accuracy while maintaining standard architectures and text-based training. We believe VLM3 opens up a new paradigm for simple and scalable 3D learning. △ Less

Submitted 28 May, 2026; originally announced May 2026.

MobileMoE: Scaling On-Device Mixture of Experts

Authors: Yanbei Chen, Hanxian Huang, Ernie Chang, Jacob Szwejbka, Digant Desai, Zechun Liu, Vikas Chandra, Raghuraman Krishnamoorthi

Abstract: Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on… ▽ More Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on-device LLMs. We first formulate an on-device MoE scaling law that jointly optimizes MoE architecture under mobile memory and compute constraints, identifying an on-device sweet spot - moderate sparsity with fine-grained and shared experts - that is simultaneously memory and compute-optimal. Building on the derived architectures, we train MobileMoE with a four-stage recipe covering pre-training, mid-training, instruction fine-tuning, and quantization-aware training, all on open-source datasets. Across 14 benchmarks, MobileMoE matches or exceeds leading on-device dense LLMs with 2-4$\times$ fewer inference FLOPs, and matches or surpasses the state-of-the-art MoE OLMoE-1B-7B with up to 60% fewer parameters. To bridge the last mile to mobile deployment, we provide the first efficient MoE inference on commodity smartphones with comprehensive on-device profiling. At comparable INT4 weight memory, MobileMoE-S delivers $1.8$-$3.8\times$ faster prefill and $2.2$-$3.4\times$ faster decode than the dense baseline MobileLLM-Pro. △ Less

Submitted 26 May, 2026; originally announced May 2026.

The DECam MAGIC Survey $-$ Mapping the Ancient Galaxy in CaHK: Overview and Summary of Early Science

Authors: A. Chiti, A. Drlica-Wagner, A. B. Pace, W. Cerny, K. R. Atzberger, F. O. Barbosa, J. A. Carballo-Bello, H. Q. Do, A. P. Ji, G. Limberg, A. M. Luna, C. E. Martínez-Vázquez, V. M. Placco, D. S. Prabhu, G. S. Stringfellow, A. K. Vivas, A. R. Walker, S. N. Campana, J. L. Carlin, V. Chandra, D. Crnojević, P. S. Ferguson, J. J. Hermes, N. Kallivayalil, G. E. Medina , et al. (20 additional authors not shown)

Abstract: We present the DECam Mapping the Ancient Galaxy in CaHK (MAGIC) survey, a 54-night NOIRLab Survey Program to image $\gtrsim$5,000$\,$deg$^2$ of the southern hemisphere using a metallicity-sensitive narrow-band filter covering the Ca$\,$ii$\,$H&K lines centered at 3955$\,$A. This filter is installed on the Dark Energy Camera (DECam), mounted on the 4-m NSF Víctor M. Blanco Telescope. The survey rea… ▽ More We present the DECam Mapping the Ancient Galaxy in CaHK (MAGIC) survey, a 54-night NOIRLab Survey Program to image $\gtrsim$5,000$\,$deg$^2$ of the southern hemisphere using a metallicity-sensitive narrow-band filter covering the Ca$\,$ii$\,$H&K lines centered at 3955$\,$A. This filter is installed on the Dark Energy Camera (DECam), mounted on the 4-m NSF Víctor M. Blanco Telescope. The survey reaches typical $10σ$ depths of $\text{mag}_{\text{CaHK}} \approx 22.5$, 3$-$4$\,$mag deeper than comparable surveys in the southern hemisphere. By combining photometry from this Ca$\,$ii$\,$H&K filter with existing DECam $g,r,i$ broadband photometry from the DECam Local Volume Exploration (DELVE) survey, MAGIC is deriving photometric metallicities for red giant branch stars down to the magnitude limit of usable proper motions from Gaia data release 3 (DR3). MAGIC has already imaged $\sim$3,000$\,$deg$^2$, supplemented by other affiliated observing programs that have used this filter to image star clusters, dwarf galaxies, and stellar streams. We overview MAGIC's survey strategy, describe data processing through the derivation of metallicities and photometric distances, and summarize early science results that have been published with this dataset. In addition, we present several new results, including the confirmation of a distant ($>5\,r_h$) member of the Reticulum II ultra-faint dwarf galaxy, on-sky density maps of low-metallicity stars into the distant Milky Way halo ($\sim150\,$kpc) recovering 13/14 ultra-faint dwarf galaxies in the current footprint, and a validation of our initial targeting of extremely metal-poor stars. Collectively, these results demonstrate that the MAGIC dataset enables cutting-edge studies of the faint, low-metallicity regime of the Milky Way and its substructures. △ Less

Submitted 26 May, 2026; originally announced May 2026.

Comments: 33 pages, 15 figures, 1 table, to be submitted to the Open Journal of Astrophysics

From protogalaxy through thick and thin: Why did the Milky Way evolve in three kinematic phases?

Authors: Olti Myrtaj, James S. Bullock, Michael Boylan-Kolchin, Vedant Chandra, Claude-André Faucher-Giguère, Robert Feldmann, Francisco J. Mercado, Jorge Moreno, Jonathan Stern, Andrew Wetzel, Pratik J. Gandhi

Abstract: APOGEE and Gaia data have revealed that the Milky Way's structure appears to have evolved through three distinct kinematic phases. First, at early cosmic times, the Milky Way was a disordered protogalaxy, which subsequently "spun up" to a second kinematic phase marked by star formation occurring in a rotating, thick stellar disk. The thick disk phase later transitioned to a third (and final) phase… ▽ More APOGEE and Gaia data have revealed that the Milky Way's structure appears to have evolved through three distinct kinematic phases. First, at early cosmic times, the Milky Way was a disordered protogalaxy, which subsequently "spun up" to a second kinematic phase marked by star formation occurring in a rotating, thick stellar disk. The thick disk phase later transitioned to a third (and final) phase with star formation occurring in a cold, thin stellar disk. In this paper, we use a suite of FIRE-2 simulations of Milky Way-mass galaxies to demonstrate that the same three phases arise in our cosmological zoom-in simulations, and study their physical origin. In all of our galaxies, the early disordered phase occurs when the rate of cool gas ($T \leq 10^4$ K) converting into stars is low, the star formation rate is bursty, and the baryonic mass "sloshes" within the host potential with respect to the center of mass motion. The gas in the galaxy begins to spin coherently after the sloshing phase ends, followed by the spin-up of young stars. The central potential of the galaxy is least concentrated just prior to gas spin-up. This second, thick disk phase coincides with a period when the rate of cool gas converting into stars is highest, even though the star formation rate remains bursty in this phase. The final transition to the thin disk phase occurs when the inner circumgalactic medium virializes. The thin disk phase is associated with a time of steady star formation and intermediate rates of cool gas converting into stars. Mergers do not appear to play a defining role in driving transitions between the three phases. The condition for the formation of a thick disk appears to be fairly minimal: a stable center of mass motion. The formation of a thin disk requires more: gas must accrete slowly enough for its angular momentum to mix and become coherent prior to joining the galaxy. △ Less

Submitted 21 May, 2026; originally announced May 2026.

Comments: 27 pages, 21 figures

Probing the IMF in the Early Universe -- Direct measurements in the Boötes I UFD with JWST/NIRCam

Authors: Keyi Ding, Mario Gennaro, Roberto J. Avila, Massimo Ricotti, Rachael L. Beaton, Martha L. Boyer, Thomas M. Brown, Annalisa Calamida, Santi Cassisi, Vedant Chandra, Roger E. Cohen, Matteo Correnti, Denija Crnojević, Kareem El-Badry, Marla Geha, Puragra Guhathakurta, Nitya Kallivayalil, Evan N. Kirby, Kristen. B. W. McQuinn, Alessandro Savino, Cheyanne Shariat, Joshua D. Simon, Daniel R. Weisz

Abstract: The dependence of the stellar initial mass function (IMF) on star-formation environment, particularly at low metallicities and high redshifts, remains poorly constrained. Ultra-faint dwarf galaxies (UFDs) are local fossils of high-redshift galaxies hosting old, metal-poor populations, and their resolved stellar populations provide unique pathways to constrain the sub-solar IMF. We investigate the… ▽ More The dependence of the stellar initial mass function (IMF) on star-formation environment, particularly at low metallicities and high redshifts, remains poorly constrained. Ultra-faint dwarf galaxies (UFDs) are local fossils of high-redshift galaxies hosting old, metal-poor populations, and their resolved stellar populations provide unique pathways to constrain the sub-solar IMF. We investigate the low-mass IMF in the Boötes I (Boo I) UFD with JWST/NIRCam, leveraging its capability to resolve over 10,000 stars reaching $\lesssim0.15 M_{\odot}$, obtaining one of the largest, deepest resolved stellar samples for UFDs. We explore three different functional forms of the IMF with machine learning and statistical techniques, combining forward modeling of synthetic color-magnitude diagrams with simulation-based inference. We find that a single power-law IMF fails to reproduce the observed luminosity function and also deviates from the canonical Salpeter IMF. Our best-fit broken power-law and lognormal IMF parameters are consistent with the Milky Way within 68% confidence level, providing evidence that star formation at metallicities as low as [Fe/H]$\approx-2.4$ follows a similar IMF as in the Milky Way. By treating Boo I as a local relic analogous to a high-redshift galaxy with a stellar mass of $\lesssim10^5 M_{\odot}$ at $z\gtrsim6$, our results provide evidence for the universality of the IMF across both local and high-redshift environments. △ Less

Submitted 14 May, 2026; originally announced May 2026.

Comments: 26 pages, 15 figures, submitted to ApJ

Exploring Audio Hallucination in Egocentric Video Understanding

Authors: Ashish Seth, Xinhao Mei, Changsheng Zhao, Varun Nagaraja, Ernie Chang, Gregory P. Meyer, Gael Le Lan, Yunyang Xiong, Vikas Chandra, Yangyang Shi, Dinesh Manocha, Zhipeng Cai

Abstract: Egocentric videos provide a distinctive setting in which sound serves as crucial cues to understand user activities and surroundings, particularly when visual information is unstable or occluded due to continuous camera movement. State-of-the-art large audio-visual language models (AV-LLMs) can generate multimodal descriptions. However, we show in this work that they are prone to audio hallucinati… ▽ More Egocentric videos provide a distinctive setting in which sound serves as crucial cues to understand user activities and surroundings, particularly when visual information is unstable or occluded due to continuous camera movement. State-of-the-art large audio-visual language models (AV-LLMs) can generate multimodal descriptions. However, we show in this work that they are prone to audio hallucinations, often inferring sounds from visual cues that are visible but not heard. We present a systematic and automatic evaluation framework for analyzing audio hallucinations in egocentric video through a targeted question-answering (Q/A) protocol. We curate a dataset of 300 egocentric videos and design 1,000 sound-focused questions to probe model outputs. To characterize hallucinations, we propose a grounded taxonomy that distinguishes between foreground action sounds from the user activities and background ambient sounds. Our evaluation shows that advanced AV-LLMs, such as Qwen2.5 Omni, exhibit high hallucination rates, achieving only 27.3% and 39.5% accuracy on Q/As related to foreground and background sounds, respectively. With this work, we highlight the need to measure the reliability of multimodal responses, emphasizing that robust evaluation of hallucinations is essential to develop reliable AV-LLMs. △ Less

Submitted 26 April, 2026; originally announced April 2026.

Comments: Accepted to ICASSP 2026

RPRA: Predicting an LLM-Judge for Efficient but Performant Inference

Authors: Dylan R. Ashley, Gaël Le Lan, Changsheng Zhao, Naina Dhingra, Zhipeng Cai, Ernie Chang, Mingchen Zhuge, Yangyang Shi, Vikas Chandra, Jürgen Schmidhuber

Abstract: Large language models (LLMs) face a fundamental trade-off between computational efficiency (e.g., number of parameters) and output quality, especially when deployed on computationally limited devices such as phones or laptops. One way to address this challenge is by following the example of humans and have models ask for help when they believe they are incapable of solving a problem on their own;… ▽ More Large language models (LLMs) face a fundamental trade-off between computational efficiency (e.g., number of parameters) and output quality, especially when deployed on computationally limited devices such as phones or laptops. One way to address this challenge is by following the example of humans and have models ask for help when they believe they are incapable of solving a problem on their own; we can overcome this trade-off by allowing smaller models to respond to queries when they believe they can provide good responses, and deferring to larger models when they do not believe they can. To this end, in this paper, we investigate the viability of Predict-Answer/Act (PA) and Reason-Predict-Reason-Answer/Act (RPRA) paradigms where models predict -- prior to responding -- how an LLM judge would score their output. We evaluate three approaches: zero-shot prediction, prediction using an in-context report card, and supervised fine-tuning. Our results show that larger models (particularly reasoning models) perform well when predicting generic LLM judges zero-shot, while smaller models can reliably predict such judges well after being fine-tuned or provided with an in-context report card. Altogether, both approaches can substantially improve the prediction accuracy of smaller models, with report cards and fine-tuning achieving mean improvements of up to 55% and 52% across datasets, respectively. These findings suggest that models can learn to predict their own performance limitations, paving the way for more efficient and self-aware AI systems. △ Less

Submitted 14 April, 2026; originally announced April 2026.

Comments: 10 pages in main text + 6 pages of references + 36 pages of appendices, 12 figures in main text + 37 figures in appendices, 2 tables in main text + 3 table in appendices, 13 prompts in appendices

MSC Class: 68T07 ACM Class: I.2.6; I.2.7; I.2.11

Small Vision-Language Models are Smart Compressors for Long Video Understanding

Authors: Junjie Fei, Jun Chen, Zechun Liu, Yunyang Xiong, Chong Zhou, Wei Wen, Junlin Han, Mingchen Zhuge, Saksham Suri, Qi Qian, Shuming Liu, Lemeng Wu, Raghuraman Krishnamoorthi, Vikas Chandra, Mohamed Elhoseiny, Chenchen Zhu

Abstract: Adapting Multimodal Large Language Models (MLLMs) for hour-long videos is bottlenecked by context limits. Dense visual streams saturate token budgets and exacerbate the lost-in-the-middle phenomenon. Existing heuristics, like sparse sampling or uniform pooling, blindly sacrifice fidelity by discarding decisive moments and wasting bandwidth on irrelevant backgrounds. We propose Tempo, an efficient… ▽ More Adapting Multimodal Large Language Models (MLLMs) for hour-long videos is bottlenecked by context limits. Dense visual streams saturate token budgets and exacerbate the lost-in-the-middle phenomenon. Existing heuristics, like sparse sampling or uniform pooling, blindly sacrifice fidelity by discarding decisive moments and wasting bandwidth on irrelevant backgrounds. We propose Tempo, an efficient query-aware framework compressing long videos for downstream understanding. Tempo leverages a Small Vision-Language Model (SVLM) as a local temporal compressor, casting token reduction as an early cross-modal distillation process to generate compact, intent-aligned representations in a single forward pass. To enforce strict budgets without breaking causality, we introduce Adaptive Token Allocation (ATA). Exploiting the SVLM's zero-shot relevance prior and semantic front-loading, ATA acts as a training-free $O(1)$ dynamic router. It allocates dense bandwidth to query-critical segments while compressing redundancies into minimal temporal anchors to maintain the global storyline. Extensive experiments show our 6B architecture achieves state-of-the-art performance with aggressive dynamic compression (0.5-16 tokens/frame). On the extreme-long LVBench (4101s), Tempo scores 52.3 under a strict 8K visual budget, outperforming GPT-4o and Gemini 1.5 Pro. Scaling to 2048 frames reaches 53.7. Crucially, Tempo compresses hour-long videos substantially below theoretical limits, proving true long-form video understanding relies on intent-driven efficiency rather than greedily padded context windows. △ Less

Submitted 9 April, 2026; originally announced April 2026.

Comments: Project page and demo are available at https://FeiElysia.github.io/tempo-page/

Neural Computers

Authors: Mingchen Zhuge, Changsheng Zhao, Haozhe Liu, Zijian Zhou, Shuming Liu, Wenyi Wang, Ernie Chang, Gael Le Lan, Junjie Fei, Wenxuan Zhang, Yasheng Sun, Zhipeng Cai, Zechun Liu, Yunyang Xiong, Yining Yang, Yuandong Tian, Yangyang Shi, Vikas Chandra, Jürgen Schmidhuber

Abstract: We propose a new frontier: Neural Computers (NCs) that unify computation, memory, and I/O of traditional computers in a learned runtime state. Our long-term goal is the Completely Neural Computer (CNC): the mature, general-purpose realization of this emerging machine form, with stable execution, explicit reprogramming, and durable capability reuse. As an initial step, we study whether elementary N… ▽ More We propose a new frontier: Neural Computers (NCs) that unify computation, memory, and I/O of traditional computers in a learned runtime state. Our long-term goal is the Completely Neural Computer (CNC): the mature, general-purpose realization of this emerging machine form, with stable execution, explicit reprogramming, and durable capability reuse. As an initial step, we study whether elementary NC primitives can be learned solely from collected I/O traces, without instrumented program state. Concretely, we instantiate NCs as video models that roll out screen frames from instructions, pixels, and user actions (when available) in CLI and GUI settings. We show that NCs can acquire elementary interface primitives, especially I/O alignment and short-horizon control, while routine reuse, controlled updates, and symbolic stability remain challenging. We outline a roadmap toward CNCs, to establish a new computing paradigm beyond today's agents and conventional computers. △ Less

Submitted 16 April, 2026; v1 submitted 7 April, 2026; originally announced April 2026.

Comments: Github (data pipeline): https://github.com/metauto-ai/NeuralComputer; Blogpost: https://metauto.ai/neuralcomputer/index_eng.html

Efficient Universal Perception Encoder

Authors: Chenchen Zhu, Saksham Suri, Cijo Jose, Maxime Oquab, Marc Szafraniec, Wei Wen, Yunyang Xiong, Patrick Labatut, Piotr Bojanowski, Raghuraman Krishnamoorthi, Vikas Chandra

Abstract: Running AI models on smart edge devices can unlock versatile user experiences, but presents challenges due to limited compute and the need to handle multiple tasks simultaneously. This requires a vision encoder with small size but powerful and versatile representations. We present our method, Efficient Universal Perception Encoder (EUPE), which offers both inference efficiency and universally good… ▽ More Running AI models on smart edge devices can unlock versatile user experiences, but presents challenges due to limited compute and the need to handle multiple tasks simultaneously. This requires a vision encoder with small size but powerful and versatile representations. We present our method, Efficient Universal Perception Encoder (EUPE), which offers both inference efficiency and universally good representations for diverse downstream tasks. We achieve this by distilling from multiple domain-expert foundation vision encoders. Unlike previous agglomerative methods that directly scale down from multiple teachers to an efficient encoder, we demonstrate the importance of first scaling up to a large proxy teacher and then scaling down from this single teacher. Experiments show that EUPE achieves on-par or better performance than individual domain experts of the same size on diverse task domains and also outperforms previous agglomerative encoders. We release the full family of EUPE models and the code to foster future research. △ Less

Submitted 31 March, 2026; v1 submitted 23 March, 2026; originally announced March 2026.

Comments: Code: https://github.com/facebookresearch/EUPE; Model: https://huggingface.co/collections/facebook/eupe

dTRPO: Trajectory Reduction in Policy Optimization of Diffusion Large Language Models

Authors: Wenxuan Zhang, Lemeng Wu, Changsheng Zhao, Ernie Chang, Mingchen Zhuge, Zechun Liu, Andy Su, Hanxian Huang, Jun Chen, Chong Zhou, Raghuraman Krishnamoorthi, Vikas Chandra, Mohamed Elhoseiny, Wei Wen

Abstract: Diffusion Large Language Models (dLLMs) introduce a new paradigm for language generation, which in turn presents new challenges for aligning them with human preferences. In this work, we aim to improve the policy optimization for dLLMs by reducing the cost of the trajectory probability calculation, thereby enabling scaled-up offline policy training. We prove that: (i) under reference policy regula… ▽ More Diffusion Large Language Models (dLLMs) introduce a new paradigm for language generation, which in turn presents new challenges for aligning them with human preferences. In this work, we aim to improve the policy optimization for dLLMs by reducing the cost of the trajectory probability calculation, thereby enabling scaled-up offline policy training. We prove that: (i) under reference policy regularization, the probability ratio of the newly unmasked tokens is an unbiased estimate of that of intermediate diffusion states, and (ii) the probability of the full trajectory can be effectively estimated with a single forward pass of a re-masked final state. By integrating these two trajectory reduction strategies into a policy optimization objective, we propose Trajectory Reduction Policy Optimization (dTRPO). We evaluate dTRPO on 7B dLLMs across instruction-following and reasoning benchmarks. Results show that it substantially improves the core performance of state-of-the-art dLLMs, achieving gains of up to 9.6% on STEM tasks, up to 4.3% on coding tasks, and up to 3.0% on instruction-following tasks. Moreover, dTRPO exhibits strong training efficiency due to its offline, single-forward nature, and achieves improved generation efficiency through high-quality outputs. △ Less

Submitted 13 April, 2026; v1 submitted 19 March, 2026; originally announced March 2026.

MobileLLM-Flash: Latency-Guided On-Device LLM Design for Industry Scale Deployment

Authors: Hanxian Huang, Igor Fedorov, Andrey Gromov, Bernard Beckerman, Naveen Suda, David Eriksson, Maximilian Balandat, Rylan Conway, Patrick Huber, Chinnadhurai Sankar, Ayushi Dalmia, Zechun Liu, Lemeng Wu, Tarek Elgamal, Adithya Sagar, Vikas Chandra, Raghuraman Krishnamoorthi

Abstract: Real-time AI experiences call for on-device large language models (OD-LLMs) optimized for efficient deployment on resource-constrained hardware. The most useful OD-LLMs produce near-real-time responses and exhibit broad hardware compatibility, maximizing user reach. We present a methodology for designing such models using hardware-in-the-loop architecture search under mobile latency constraints. T… ▽ More Real-time AI experiences call for on-device large language models (OD-LLMs) optimized for efficient deployment on resource-constrained hardware. The most useful OD-LLMs produce near-real-time responses and exhibit broad hardware compatibility, maximizing user reach. We present a methodology for designing such models using hardware-in-the-loop architecture search under mobile latency constraints. This system is amenable to industry-scale deployment: it generates models deployable without custom kernels and compatible with standard mobile runtimes like Executorch. Our methodology avoids specialized attention mechanisms and instead uses attention skipping for long-context acceleration. Our approach jointly optimizes model architecture (layers, dimensions) and attention pattern. To efficiently evaluate candidates, we treat each as a pruned version of a pretrained backbone with inherited weights, thereby achieving high accuracy with minimal continued pretraining. We leverage the low cost of latency evaluation in a staged process: learning an accurate latency model first, then searching for the Pareto-frontier across latency and quality. This yields MobileLLM-Flash, a family of foundation models (350M, 650M, 1.4B) for efficient on-device use with strong capabilities, supporting up to 8k context length. MobileLLM-Flash delivers up to 1.8x and 1.6x faster prefill and decode on mobile CPUs with comparable or superior quality. Our analysis of Pareto-frontier design choices offers actionable principles for OD-LLM design. △ Less

Submitted 27 April, 2026; v1 submitted 16 March, 2026; originally announced March 2026.

Comments: Accepted to ACL Industry Track 2026

The Stellar Initial Mass Function down to 0.16M$_{\odot}$ Towards the Small Magellanic Cloud

Authors: Roger E. Cohen, Mario Gennaro, Matteo Correnti, Kristen B. W. McQuinn, Vedant Chandra

Abstract: The presence (and nature) of variations in the stellar initial mass function (IMF) at substantially sub-solar masses and metallicities ($m$$<$0.5M$_{\odot}$, [M/H]$\lesssim$$-$1) remains poorly constrained. Predictions from simulations vary widely, while observationally, resolved star studies of ultra-faint dwarf galaxies (UFDs) suffer from small sample sizes and background galaxy contamination du… ▽ More The presence (and nature) of variations in the stellar initial mass function (IMF) at substantially sub-solar masses and metallicities ($m$$<$0.5M$_{\odot}$, [M/H]$\lesssim$$-$1) remains poorly constrained. Predictions from simulations vary widely, while observationally, resolved star studies of ultra-faint dwarf galaxies (UFDs) suffer from small sample sizes and background galaxy contamination due to low projected stellar densities. As an alternative metal-poor target, we measure the IMF from resolved stars towards a carefully selected field in the Small Magellanic Cloud (SMC), leveraging a plethora of independent constraints on the target field stellar population including distributions of distance, %extinction, age and metallicity. We resolve $>$15,000 stars down to 0.16M$_{\odot}$ within a single pointing of NIRCam onboard JWST, using an observing strategy that minimizes contamination from point-source-like background galaxies. We explore three different functional forms of the IMF, forward modeling observed color-magnitude diagrams (CMDs) and luminosity functions. We find a best-fit single power law IMF slope of $α$=$-$1.61$^{+0.03}_{-0.03}$, consistent with UFDs probed down to similar limiting masses. Fitting a broken power law IMF, we find low- and high-mass slopes of $α_{1}$=$-$1.44$^{+0.04}_{-0.04}$, $α_{2}$=$-$2.17$^{+0.11}_{-0.11}$ respectively, consistent with solar neighborhood values. Assuming a lognormal IMF, we find a characteristic mass and lognormal width of $m_{c}$=0.12$^{+0.03}_{-0.03}$M$_{\odot}$, $σ$=0.61$^{+0.07}_{-0.06}$M$_{\odot}$, allowing for characteristic masses lower than local values as seen in some simulations as well as low-metallicity Galactic clusters. Lastly, we quantify the impact of assumptions required in our analysis and discuss potential future improvements. △ Less

Submitted 16 March, 2026; originally announced March 2026.

Comments: 28 pages, 14 figures and 2 tables including appendices. ApJ in press

Quantifying the Milky Way, LMC and their interaction using all-sky kinematics of outer halo stars

Authors: Richard A. N. Brooks, Jason L. Sanders, Adam M. Dillamore, Nicolás Garavito-Camargo, Vedant Chandra, Adrian M. Price-Whelan, Phillip Cargile

Abstract: The recent pericentric passage of the Large Magellanic Cloud (LMC) has dislodged the Milky Way's (MW) centre of mass, inducing dynamical disequilibrium, the reflex motion, in the kinematics of outer stellar halo stars. Using data out to $160 \, \rm kpc$ from the combined H3+SEGUE+MagE outer halo survey, we constrain the mass of the MW and LMC, as well as the resulting reflex motion and the stellar… ▽ More The recent pericentric passage of the Large Magellanic Cloud (LMC) has dislodged the Milky Way's (MW) centre of mass, inducing dynamical disequilibrium, the reflex motion, in the kinematics of outer stellar halo stars. Using data out to $160 \, \rm kpc$ from the combined H3+SEGUE+MagE outer halo survey, we constrain the mass of the MW and LMC, as well as the resulting reflex motion and the stellar halo velocity anisotropy. Using a suite of 32,000 rigid MW--LMC simulations, each with a MW stellar halo evolved to the present day in the combined MW--LMC potential, we perform Simulation Based Inference by training a neural posterior estimator on the means and dispersions of the radial and tangential velocities of stars from the combined H3+SEGUE+MagE outer halo sample. Relative to halo stars at $100 \, \rm kpc$, we find the magnitude of the reflex velocity to be $v_{\rm travel} = 38.6^{+8.3}_{-7.8}\,\rm km \, s^{-1}$. Simultaneously, we determine the enclosed MW mass, $M_{\rm MW}(< 50 \, \rm kpc) = 3.36 \pm 0.15 \times 10^{11}\, \rm M_{\odot}$ and the enclosed LMC mass, $M_{\rm LMC}(< 50 \, \rm kpc) = 8.76^{+1.94}_{-1.77} \times 10^{10}\, \rm M_{\odot}$. Our results suggest that the total LMC mass must be at least $\sim20\%$ that of the MW. The velocity anisotropy prior to the LMC's infall is constrained to be $β_0 = 0.68 \pm 0.02$. Finally, we demonstrate that neglecting the LMC in models biases the estimated MW mass to prefer more massive values. △ Less

Submitted 13 May, 2026; v1 submitted 9 March, 2026; originally announced March 2026.

Comments: MNRAS accepted. 17 pages, 12 figures, 2 tables. arXiv admin note: text overlap with arXiv:2510.04735

Journal ref: Mon Not R Astron Soc (2026)

The Binary Populations of Stellar Streams are Set by Cluster Dynamics

Authors: Anya Phillips, Charlie Conroy, Jacob Nibauer, Long Wang, Vedant Chandra, Ana Bonaca, Jay Strader, Morgan MacLeod

Abstract: We present a suite of direct N-body simulations of low mass ($<10^4~M_{\odot}$) globular cluster streams initialized with observationally-motivated binary demographics in order to understand the effect of in-cluster dynamical processing on the stream binary population. The models are initialized with a range of stellar densities and cluster orbits, and Poisson variation in the number of massive an… ▽ More We present a suite of direct N-body simulations of low mass ($<10^4~M_{\odot}$) globular cluster streams initialized with observationally-motivated binary demographics in order to understand the effect of in-cluster dynamical processing on the stream binary population. The models are initialized with a range of stellar densities and cluster orbits, and Poisson variation in the number of massive and short-lived stars. Wide binaries are disrupted on short timescales by internal tides and on long timescales by two-body encounters. Tides are most important prior to impulsive mass loss-driven cluster expansion. Close binaries ($P_{\rm orb}<10^2~\rm yr$) are most abundant at the stream center due to cluster mass segregation. The wide binary fraction and the degree of binary segregation in the resulting stream are sensitive to the initial cluster density and massive star fraction. In mock radial velocity surveys of the simulated streams, undetectable binaries have velocity amplitudes of $\sim$$0.5$-$1~\rm km~s^{-1}$, adding $\sim0.1~\rm km\ s^{-1}$ of velocity dispersion to the streams, and are dynamically depleted by $\sim10$-$60\%$ compared to the initial binary population. Custom N-body models of Milky Way streams with binaries will allow a holistic understanding of their dynamical structures in advance of upcoming multi-epoch spectroscopic surveys. △ Less

Submitted 6 March, 2026; originally announced March 2026.

Comments: 27 pages, 16 Figures, 1 table, submitted to ApJ

EgoAVU: Egocentric Audio-Visual Understanding

Authors: Ashish Seth, Xinhao Mei, Changsheng Zhao, Varun Nagaraja, Ernie Chang, Gregory P. Meyer, Gael Le Lan, Yunyang Xiong, Vikas Chandra, Yangyang Shi, Dinesh Manocha, Zhipeng Cai

Abstract: Understanding egocentric videos plays a vital role for embodied intelligence. Recent multi-modal large language models (MLLMs) can accept both visual and audio inputs. However, due to the challenge of obtaining text labels with coherent joint-modality information, whether MLLMs can jointly understand both modalities in egocentric videos remains under-explored. To address this problem, we introduce… ▽ More Understanding egocentric videos plays a vital role for embodied intelligence. Recent multi-modal large language models (MLLMs) can accept both visual and audio inputs. However, due to the challenge of obtaining text labels with coherent joint-modality information, whether MLLMs can jointly understand both modalities in egocentric videos remains under-explored. To address this problem, we introduce EgoAVU, a scalable data engine to automatically generate egocentric audio-visual narrations, questions, and answers. EgoAVU enriches human narrations with multimodal context and generates audio-visual narrations through cross-modal correlation modeling. Token-based video filtering and modular, graph-based curation ensure both data diversity and quality. Leveraging EgoAVU, we construct EgoAVU-Instruct, a large-scale training dataset of 3M samples, and EgoAVU-Bench, a manually verified evaluation split covering diverse tasks. EgoAVU-Bench clearly reveals the limitations of existing MLLMs: they bias heavily toward visual signals, often neglecting audio cues or failing to correspond audio with the visual source. Finetuning MLLMs on EgoAVU-Instruct effectively addresses this issue, enabling up to 113% performance improvement on EgoAVU-Bench. Such benefits also transfer to other benchmarks such as EgoTempo and EgoIllusion, achieving up to 28% relative performance gain. Code will be released to the community. △ Less

Submitted 5 February, 2026; originally announced February 2026.

Evaluating Classifications of Extremely Metal-poor Candidates Selected from Gaia XP Spectra

Authors: Riley Thai, Andrew R. Casey, Alexander Ji, Vedant Chandra, Hans-Walter Rix

Abstract: Extremely metal-poor stars are intrinsically rare, but emerging methods exist to accurately classify them from all-sky Gaia XP low-resolution spectra. To assess their overall accuracy for targeting metal-poor stars, we present a high-resolution spectroscopic followup of 75 very metal-poor candidates selected from the catalog by R. Andrae, V. Chandra, and H. W. Rix. We discover 2 new extremely meta… ▽ More Extremely metal-poor stars are intrinsically rare, but emerging methods exist to accurately classify them from all-sky Gaia XP low-resolution spectra. To assess their overall accuracy for targeting metal-poor stars, we present a high-resolution spectroscopic followup of 75 very metal-poor candidates selected from the catalog by R. Andrae, V. Chandra, and H. W. Rix. We discover 2 new extremely metal-poor ($\rm{[Fe / H]}<-3$) stars and 20 new very metal-poor ($\rm{[Fe/H]} < -2$) stars. Abundances of up to 22 elements are derived from 1D local thermodynamic equilibrium analysis and kinematic parameters are derived using Gaia astrometry and spectroscopic radial velocities. The chemodynamical properties are mostly consistent with expectations for halo stars, but we discover an Mg-enhanced CEMP star ($\mathrm{[Mg/Fe]} = 0.89$) and an Mg-poor star from an accreted ultra-faint dwarf galaxy. The Gaia XP metallicity estimates are consistent with our $\rm{[Fe/H]}$ measurements down to $\rm{[Fe/H]}\sim -3.0$, but estimates worsen in highly extincted regions. We find that 4 other XP-based metallicity catalogs succeed in mitigating contaminants and can also classify metal-poor stars robustly to $\rm{[Fe/H]}\sim -3.0$. Our results demonstrate the utility of Gaia XP spectra for identifying the most metal-poor stars across the Galaxy. △ Less

Submitted 29 January, 2026; originally announced January 2026.

Comments: 26 pages, 12 figures. Submitted to AJ

SLAP: Scalable Language-Audio Pretraining with Variable-Duration Audio and Multi-Objective Training

Authors: Xinhao Mei, Gael Le Lan, Haohe Liu, Zhaoheng Ni, Varun Nagaraja, Yang Liu, Yangyang Shi, Vikas Chandra

Abstract: Contrastive language-audio pretraining (CLAP) has achieved notable success in learning semantically rich audio representations and is widely adopted for various audio-related tasks. However, current CLAP models face several key limitations. First, they are typically trained on relatively small datasets, often comprising a few million audio samples. Second, existing CLAP models are restricted to sh… ▽ More Contrastive language-audio pretraining (CLAP) has achieved notable success in learning semantically rich audio representations and is widely adopted for various audio-related tasks. However, current CLAP models face several key limitations. First, they are typically trained on relatively small datasets, often comprising a few million audio samples. Second, existing CLAP models are restricted to short and fixed duration, which constrains their usage in real-world scenarios with variable-duration audio. Third, the standard contrastive training objective operates on global representations, which may hinder the learning of dense, fine-grained audio features. To address these challenges, we introduce Scalable Language-Audio Pretraining (SLAP), which scales language-audio pretraining to 109 million audio-text pairs with variable audio durations and incorporates multiple training objectives. SLAP unifies contrastive loss with additional self-supervised and captioning losses in a single-stage training, facilitating the learning of richer dense audio representations. The proposed SLAP model achieves new state-of-the-art performance on audio-text retrieval and zero-shot audio classification tasks, demonstrating its effectiveness across diverse benchmarks. △ Less

Submitted 18 January, 2026; originally announced January 2026.

Comments: Accepted to ICASSP 2026

Discovery of the First Five Carbon-Enhanced Metal-Poor Stars in the LMC

Authors: Madeline Lucey, Vedant Chandra, Alexander Ji, Andrew Casey, David Nidever, Sean Morrison, Robyn Sanderson, Slater Oden, José Fernández-Trincado, Guilherme Limberg

Abstract: A substantial fraction of metal-poor stars in the local Milky Way halo exhibit large overabundances of carbon. These stars, dubbed Carbon-Enhanced Metal-Poor (CEMP) stars, provide crucial constraints on the nature of the early universe including the earliest nucleosynthetic events. Whether these stars exist at similar rates in nearby galaxies is a major open question with implications for the envi… ▽ More A substantial fraction of metal-poor stars in the local Milky Way halo exhibit large overabundances of carbon. These stars, dubbed Carbon-Enhanced Metal-Poor (CEMP) stars, provide crucial constraints on the nature of the early universe including the earliest nucleosynthetic events. Whether these stars exist at similar rates in nearby galaxies is a major open question with implications for the environmental dependence of early chemical evolution. Here, we present the discovery of the first five CEMP stars in the Milky Way's largest dwarf companion, the LMC, using SDSS-V spectra from the BOSS instrument. We measure metallicities ranging from [Fe/H] = -2.1 to -3.2 and evolutionary state corrected carbon enhancements of [C/Fe] = +1.2 to +2.4, placing these stars among the most metal-poor and carbon-rich ever identified in the LMC. Their absolute carbon abundances and metallicities classify them as Group I CEMP stars, suggesting binary mass-transfer origins, though neutron-capture abundance measurements are required to confirm whether this classification scheme applies beyond the Milky Way. Although these stars were selected as the most promising CEMP candidates from the SDSS-V sample, likely biasing this initial sample toward higher absolute carbon abundances, their discovery suggests that previous null detections of CEMP stars in the LMC were caused by metallicity-sensitive photometric targeting biases against high [C/H] stars. A forthcoming analysis of the full spectroscopic sample will push to lower carbon abundances, providing a more complete census and enabling critical tests of whether environmental differences shape the formation channels and frequencies of CEMP stars in this system. △ Less

Submitted 23 March, 2026; v1 submitted 15 January, 2026; originally announced January 2026.

Comments: 15 pages, 10 figures, 1 table, accepted to ApJL, comments welcome!

VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice

Authors: Shuming Liu, Mingchen Zhuge, Changsheng Zhao, Jun Chen, Lemeng Wu, Zechun Liu, Chenchen Zhu, Zhipeng Cai, Chong Zhou, Haozhe Liu, Ernie Chang, Saksham Suri, Hongyu Xu, Qi Qian, Wei Wen, Balakrishnan Varadarajan, Zhuang Liu, Hu Xu, Florian Bordes, Raghuraman Krishnamoorthi, Bernard Ghanem, Vikas Chandra, Yunyang Xiong

Abstract: Chain-of-thought (CoT) reasoning has emerged as a powerful tool for multimodal large language models on video understanding tasks. However, its necessity and advantages over direct answering remain underexplored. In this paper, we first demonstrate that for RL-trained video models, direct answering often matches or even surpasses CoT performance, despite CoT producing step-by-step analyses at a hi… ▽ More Chain-of-thought (CoT) reasoning has emerged as a powerful tool for multimodal large language models on video understanding tasks. However, its necessity and advantages over direct answering remain underexplored. In this paper, we first demonstrate that for RL-trained video models, direct answering often matches or even surpasses CoT performance, despite CoT producing step-by-step analyses at a higher computational cost. Motivated by this, we propose VideoAuto-R1, a video understanding framework that adopts a reason-when-necessary strategy. During training, our approach follows a Thinking Once, Answering Twice paradigm: the model first generates an initial answer, then performs reasoning, and finally outputs a reviewed answer. Both answers are supervised via verifiable rewards. During inference, the model uses the confidence score of the initial answer to determine whether to proceed with reasoning. Across video QA and grounding benchmarks, VideoAuto-R1 achieves state-of-the-art accuracy with significantly improved efficiency, reducing the average response length by ~3.3x, e.g., from 149 to just 44 tokens. Moreover, we observe a low rate of thinking-mode activation on perception-oriented tasks, but a higher rate on reasoning-intensive tasks. This suggests that explicit language-based reasoning is generally beneficial but not always necessary. △ Less

Submitted 21 March, 2026; v1 submitted 8 January, 2026; originally announced January 2026.

Comments: Accepted to CVPR 2026. Project page: https://ivul-kaust.github.io/projects/videoauto-r1/

An Ancient Brown Dwarf Transiting a Metal-Poor Thick Disk Star

Authors: Jéa Adams Redai, Vedant Chandra, Samuel W. Yee, Victoria DiTomasso, Sean Andrews, Karin Öberg, Rebecca Woody, David W. Latham, Allyson Bieryla, Samuel N. Quinn, David Charbonneau, Theron W. Carmichael, Chih-Chun Hsu, Noah Vowell, Jason J. Wang, Sebastian Zieba, Paul Benni, Karen A. Collins, David R. Ciardi, Julian van Eyken, William Fong, Michael B. Lund, Andrei M. Tatarnikov

Abstract: We report the discovery of TOI-7019b, the first transiting brown dwarf (BD) known to orbit a star that is part of the Milky Way's ancient thick disk, as defined chemically ([Fe/H] $= -0.79 \pm 0.05$ dex, [$α$/Fe] $= +0.26 \pm 0.05$ dex, [M/H] $= -0.59 \pm 0.06$ dex) and kinematically ($v_{\perp} \approx 150 \pm 1$ km s$^{-1}$). We estimate a system age $τ= 12 \pm 2$ Gyr by fitting the host star's… ▽ More We report the discovery of TOI-7019b, the first transiting brown dwarf (BD) known to orbit a star that is part of the Milky Way's ancient thick disk, as defined chemically ([Fe/H] $= -0.79 \pm 0.05$ dex, [$α$/Fe] $= +0.26 \pm 0.05$ dex, [M/H] $= -0.59 \pm 0.06$ dex) and kinematically ($v_{\perp} \approx 150 \pm 1$ km s$^{-1}$). We estimate a system age $τ= 12 \pm 2$ Gyr by fitting the host star's spectrum and spectral energy distribution to alpha-enhanced isochrones, and independently using the age-metallicity relation of the thick disk. This makes TOI-7019 by far the most metal-poor and ancient BD host known to date. We measure a BD mass of $61.3 \pm 2.1$ $M_{\rm J}$ and radius of $0.82 \pm 0.02$ $R_{\rm J}$ from a joint analysis of transit photometry and radial velocity measurements, along with an orbital period of $48.2592 \pm 0.0001$ days and an orbital eccentricity of $0.403 \pm 0.002$. The measured radius appears $12.3\% \pm 2.8\%$ larger than predicted relative to standard evolutionary models for old, metal-poor brown dwarfs, hinting at missing physics like the magnetic inhibition of convection. TOI-7019b lowers the probed metallicity regime for transiting BDs by over a factor of two, making it a benchmark system to test evolutionary models in the low-metallicity regime. Future measurements of TOI-7019b's atmosphere will test whether a brown dwarf's atmospheric composition tracks its host star's abundances, as expected for binary-like co-formation. △ Less

Submitted 5 December, 2025; originally announced December 2025.

Comments: Submitted to ApJL

MobileLLM-Pro Technical Report

Authors: Patrick Huber, Ernie Chang, Wei Wen, Igor Fedorov, Tarek Elgamal, Hanxian Huang, Naveen Suda, Chinnadhurai Sankar, Vish Vogeti, Yanghan Wang, Alex Gladkov, Kai Sheng Tai, Abdelrahman Elogeel, Tarek Hefny, Vikas Chandra, Ahmed Aly, Anuj Kumar, Raghuraman Krishnamoorthi, Adithya Sagar

Abstract: Efficient on-device language models around 1 billion parameters are essential for powering low-latency AI applications on mobile and wearable devices. However, achieving strong performance in this model class, while supporting long context windows and practical deployment remains a significant challenge. We introduce MobileLLM-Pro, a 1-billion-parameter language model optimized for on-device deplo… ▽ More Efficient on-device language models around 1 billion parameters are essential for powering low-latency AI applications on mobile and wearable devices. However, achieving strong performance in this model class, while supporting long context windows and practical deployment remains a significant challenge. We introduce MobileLLM-Pro, a 1-billion-parameter language model optimized for on-device deployment. MobileLLM-Pro achieves state-of-the-art results across 11 standard benchmarks, significantly outperforming both Gemma 3-1B and Llama 3.2-1B, while supporting context windows of up to 128,000 tokens and showing only minor performance regressions at 4-bit quantization. These improvements are enabled by four core innovations: (1) implicit positional distillation, a novel technique that effectively instills long-context capabilities through knowledge distillation; (2) a specialist model merging framework that fuses multiple domain experts into a compact model without parameter growth; (3) simulation-driven data mixing using utility estimation; and (4) 4-bit quantization-aware training with self-distillation. We release our model weights and code to support future research in efficient on-device language models. △ Less

Submitted 10 November, 2025; originally announced November 2025.

Comments: 17 pages

The Milky Way - Large Magellanic Cloud Interaction with Simulation Based Inference

Authors: Richard A. N. Brooks, Jason L. Sanders, Vedant Chandra, Nicolás Garavito-Camargo, Adam M. Dillamore, Adrian M. Price-Whelan, Yuan-Sen Ting

Abstract: The infall of the Large Magellanic Cloud (LMC) into the Milky Way (MW) has displaced the MW's centre of mass, manifesting as an observed reflex motion in the velocities of outer halo stars. We use a Simulation Based Inference framework to constrain properties of the MW, LMC and the induced reflex motion using the dynamics of outer MW halo stars. Specifically, we use the mean radial and tangential… ▽ More The infall of the Large Magellanic Cloud (LMC) into the Milky Way (MW) has displaced the MW's centre of mass, manifesting as an observed reflex motion in the velocities of outer halo stars. We use a Simulation Based Inference framework to constrain properties of the MW, LMC and the induced reflex motion using the dynamics of outer MW halo stars. Specifically, we use the mean radial and tangential velocities of outer halo stars calculated in a set of distance and on-sky bins. We train neural networks to estimate parameter posterior distributions using a set of $128,000$ rigid MW--LMC simulations conditioned upon velocity data from the Dark Energy Spectroscopic Instrument (DESI) and the combined H3+SEGUE+MagE outer halo surveys. We constrain the reflex motion velocity and the enclosed LMC mass within $50 \, \rm kpc$ using the DESI or H3+SEGUE+MagE dataset while varying the survey sky coverage and depth. Using the radial and tangential velocity data from the H3+SEGUE+MagE survey and on-sky quadrants, we report a distance-averaged reflex motion velocity for the outer halo samples, the speed at which the MW lurches towards the LMC, of $v_{\rm{travel}} = 26.4^{+5.5}_{-4.4} \, \rm km \, \rm s^{-1}$, while simultaneously finding an enclosed LMC mass of $M_{\rm LMC}(< 50 \, \rm kpc) = 9.2^{+1.9}_{-2.3} \times 10^{10}\, \rm M_{\odot}$. Quoted uncertainties are statistical. Our results suggest that the LMC's total mass is at least $\approx 10-15 \%$ of that of the MW. This inference framework is flexible such that it can provide rapid constraints when applied to any future survey measuring the velocities of outer halo stars. △ Less

Submitted 29 June, 2026; v1 submitted 6 October, 2025; originally announced October 2025.

Comments: MNRAS accepted. 20 pages, 11 figures, 2 tables

Journal ref: Mon Not R Astron Soc (2026)

DepthLM: Metric Depth From Vision Language Models

Authors: Zhipeng Cai, Ching-Feng Yeh, Hu Xu, Zhuang Liu, Gregory Meyer, Xinjie Lei, Changsheng Zhao, Shang-Wen Li, Vikas Chandra, Yangyang Shi

Abstract: Vision language models (VLMs) can flexibly address various vision tasks through text interactions. Although successful in semantic understanding, state-of-the-art VLMs including GPT-5 still struggle in understanding 3D from 2D inputs. On the other hand, expert pure vision models achieve super-human accuracy in metric depth estimation, a key 3D understanding task. However, they require task-specifi… ▽ More Vision language models (VLMs) can flexibly address various vision tasks through text interactions. Although successful in semantic understanding, state-of-the-art VLMs including GPT-5 still struggle in understanding 3D from 2D inputs. On the other hand, expert pure vision models achieve super-human accuracy in metric depth estimation, a key 3D understanding task. However, they require task-specific architectures and losses. Such difference motivates us to ask: Can VLMs reach expert-level accuracy without architecture or loss change? We take per-pixel metric depth estimation as the representative task and show that the answer is yes! Surprisingly, comprehensive analysis shows that text-based supervised-finetuning with sparse labels is sufficient for VLMs to unlock strong 3D understanding, no dense prediction head or complex regression/regularization loss is needed. The bottleneck for VLMs lies actually in pixel reference and cross-dataset camera ambiguity, which we address through visual prompting and intrinsic-conditioned augmentation. With much smaller models, our method DepthLM surpasses the accuracy of most advanced VLMs by over 2x, making VLMs for the first time comparable with pure vision models. Interestingly, without explicit enforcement during training, VLMs trained with DepthLM naturally avoids over-smoothing, having much fewer flying points at boundary regions than pure vision models. The simplicity of DepthLM also enables a single VLM to cover various 3D tasks beyond metric depth. Our code and model will be released at the link below. △ Less

Submitted 1 October, 2025; v1 submitted 29 September, 2025; originally announced September 2025.

MobileLLM-R1: Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes

Authors: Changsheng Zhao, Ernie Chang, Zechun Liu, Chia-Jung Chang, Wei Wen, Chen Lai, Sheng Cao, Yuandong Tian, Raghuraman Krishnamoorthi, Yangyang Shi, Vikas Chandra

Abstract: The paradigm shift in large language models (LLMs) from instinctive responses to chain-of-thought (CoT) reasoning has fueled two prevailing assumptions: (1) reasoning capabilities only emerge in sufficiently large models, and (2) such capabilities require training on massive datasets. While the first assumption has already been challenged by recent sub-billion-parameter reasoning models such as Qw… ▽ More The paradigm shift in large language models (LLMs) from instinctive responses to chain-of-thought (CoT) reasoning has fueled two prevailing assumptions: (1) reasoning capabilities only emerge in sufficiently large models, and (2) such capabilities require training on massive datasets. While the first assumption has already been challenged by recent sub-billion-parameter reasoning models such as Qwen3-0.6B and DeepSeek distilled variants, the second remains largely unquestioned. In this work, we revisit the necessity of scaling to extremely large corpora (>10T tokens) for reasoning emergence. By carefully curating and resampling open-source datasets that we identify as beneficial under our designed metrics, we demonstrate that strong reasoning abilities can emerge with far less data. Specifically, we show that only ~2T tokens of high-quality data are sufficient, and pre-training with 4.2T tokens on the dataset resampled from these ~2T tokens, followed by a established post-training procedure, enables the development of MobileLLM-R1, a series of sub-billion-parameter reasoning models that substantially outperform prior models trained on fully open-sourced data. For example, MobileLLM-R1-950M achieves an AIME score of 15.5, compared to just 0.6 for OLMo-2-1.48B and 0.3 for SmolLM-2-1.7B. Remarkably, despite being trained on only 11.7% of the tokens compared to Qwen3's proprietary 36T-token corpus for pretraining, MobileLLM-R1-950M matches or surpasses Qwen3-0.6B across multiple reasoning benchmarks. To facilitate further research in this direction, we have made the models (https://huggingface.co/collections/facebook/mobilellm-r1) and code (https://github.com/facebookresearch/MobileLLM-R1) publicly available, along with the complete training recipe, data sources, and data mixing ratios. △ Less

Submitted 27 February, 2026; v1 submitted 29 September, 2025; originally announced September 2025.

Comments: ICLR 2026

A nearly pristine star from the Large Magellanic Cloud

Authors: Alexander P. Ji, Vedant Chandra, Selenna Mejias-Torres, Zhongyuan Zhang, Philipp Eitner, Kevin C. Schlaufman, Hillary Diane Andales, Ha Do, Natalie M. Orrantia, Rithika Tudmilla, Pierre N. Thibodeaux, Keivan G. Stassun, Madeline Howell, Jamie Tayar, Maria Bergemann, Andrew R. Casey, Jennifer A. Johnson, Joleen K. Carlberg, William Cerny, Jose G. Fernandez-Trincado, Keith Hawkins, Juna A. Kollmeier, Chervin F. P. Laporte, Guilherme Limberg, Tadafumi Matsuno , et al. (6 additional authors not shown)

Abstract: The first stars formed out of pristine gas, causing them to be so massive that none are expected to have survived until today. If their direct descendants were sufficiently low-mass stars, such stars could exist today and would be recognizable by having the lowest metallicities (abundance of elements heavier than helium). We present the independent identification and detailed chemical analysis of… ▽ More The first stars formed out of pristine gas, causing them to be so massive that none are expected to have survived until today. If their direct descendants were sufficiently low-mass stars, such stars could exist today and would be recognizable by having the lowest metallicities (abundance of elements heavier than helium). We present the independent identification and detailed chemical analysis of the star SDSS J0715-7334, finding ultra-low elemental abundances of both iron and carbon ([Fe/H] = -4.3, [C/Fe] < -0.2) and total metallicity Z < 7.8 x 10^{-7} (log Z/Zsun < -4.3). The star's orbit indicates that it originates from the halo of the Large Magellanic Cloud. Its heavy element abundance pattern can be explained by a primordial supernova with an initial mass of 30 solar masses. This star is over ten times more chemically pristine than the most extreme high-redshift galaxies currently found by the James Webb Space Telescope. It is sufficiently metal-poor that current models of low-mass star formation require dust cooling to explain its existence. △ Less

Submitted 1 April, 2026; v1 submitted 25 September, 2025; originally announced September 2025.

Comments: Accepted to Nature Astronomy, author version

Wide binaries in an ultra-faint dwarf galaxy: discovery, population modeling, and a nail in the coffin of primordial black hole dark matter

Authors: Cheyanne Shariat, Kareem El-Badry, Mario Gennaro, Keyi Ding, Joshua D. Simon, Roberto J. Avila, Annalisa Calamida, Santi Cassisi, Matteo Correnti, Daniel R. Weisz, Marla Geha, Evan N. Kirby, Thomas M. Brown, Massimo Ricotti, Kristen B. W. McQuinn, Nitya Kallivayalil, Karoline Gilbert, Camilla Pacifici, Puragra Guhathakurta, Denija Crnojević, Martha L. Boyer, Rachael L. Beaton, Vedant Chandra, Roger E. Cohen, Alvio Renzini , et al. (2 additional authors not shown)

Abstract: We report the discovery and characterization of a wide binary population in the ultrafaint dwarf galaxy Boötes I using deep JWST/NIRCam imaging. Our sample consists of 52 candidate binaries with projected separations of 7,000 - 16,000 au and stellar masses from near the hydrogen-burning limit to the main-sequence turnoff ($\sim0.1$ - $0.8~{\rm M_\odot}$). By forward-modeling selection biases and c… ▽ More We report the discovery and characterization of a wide binary population in the ultrafaint dwarf galaxy Boötes I using deep JWST/NIRCam imaging. Our sample consists of 52 candidate binaries with projected separations of 7,000 - 16,000 au and stellar masses from near the hydrogen-burning limit to the main-sequence turnoff ($\sim0.1$ - $0.8~{\rm M_\odot}$). By forward-modeling selection biases and chance alignments, we find that $1.25\pm0.25\%$ of Boötes I stars are members of wide binaries with separations beyond 5,000 au. This fraction, along with the distributions of separations and mass ratios, matches that in the Solar neighborhood, suggesting that wide binary formation is largely insensitive to metallicity, even down to [Fe/H] $\approx -2.5$. The observed truncation in the separation distribution near 16,000 au is well explained by stellar flyby disruptions. We also discuss how the binaries can be used to constrain the galaxy's dark matter properties. We show that our detection places new limits on primordial black hole dark matter, finding that compact objects with $M \gtrsim 5~{\rm M_\odot}$ cannot constitute more than $\sim1\%$ of the dark matter content. In contrast to previous work, we find that wide binaries are unlikely to provide robust constraints on the dark matter profile of ultrafaint galaxies given the uncertainties in the initial binary population, flyby disruptions, and contamination from chance alignments. These findings represent the most robust detection of wide binaries in an external galaxy to date, opening a new avenue for studying binary star formation and survival in extreme environments. △ Less

Submitted 1 October, 2025; v1 submitted 4 September, 2025; originally announced September 2025.

Comments: re-submitted to PASP after minor changes

Double White Dwarf Binaries in SDSS-V DR19 : A catalog of DA white dwarf binaries and constraints on the binary population

Authors: Gautham Adamane Pallathadka, Vedant Chandra, Nadia L. Zakamska, Nicole R. Crumpler, Stefan M. Arseneau, Kareem El-Badry, Boris T. Gäensicke, Yossef Zenati, J. J. Hermes, Axel D. Schwope, Carles Badenes, Nicola Pietro Gentile Fusillo, Sean Morrison, Tim Cunningham, Priyanka Chakraborty, Gagik Tovmasian, Dmitry Bizyaev, Kaike Pan, Scott F. Anderson, Sebastian Demasi

Abstract: The fifth-generation Sloan Digital Sky Survey (SDSS-V) includes the first large-scale spectroscopic survey of white dwarfs (WDs) in the era of Gaia parallaxes. SDSS-V collects multiple exposures per target, making it ideal for binary detection. We present a search for hydrogen atmosphere (DA) double white dwarf (DWD) binaries in this rich dataset. We quantify radial velocity variations between sub… ▽ More The fifth-generation Sloan Digital Sky Survey (SDSS-V) includes the first large-scale spectroscopic survey of white dwarfs (WDs) in the era of Gaia parallaxes. SDSS-V collects multiple exposures per target, making it ideal for binary detection. We present a search for hydrogen atmosphere (DA) double white dwarf (DWD) binaries in this rich dataset. We quantify radial velocity variations between sub-exposures to identify binary candidates, and also measure the orbital period for a subset of DWD binary candidates. We find 63 DWD binary candidates, of which 43 are new discoveries, and we provide tentative periods for 10 binary systems. Using these measurements, we place constraints on the binary fraction of the Galactic WD population with $< 0.4$ AU separations $f_{\mathrm{bin,0.4}} = 9\%$, and the power-law index of the initial separation distribution $α= -0.62$. Using the simulated binary population, we estimate that $\leq 10$ super-Chandrasekhar binaries that merge within a Hubble time are expected in our sample. We predict that $\leq 5$ systems in our sample should be detectable via gravitational waves by LISA (Laser Interferometer Space Antenna), one of which has already been identified as a LISA verification source. We also estimate a total of about 10,000 - 20,000 LISA-detectable DWD binaries in the galaxy. Our catalog of WD+WD binary candidates in SDSS-V is now public, and promises to uncover a large number of exciting DWD systems. △ Less

Submitted 2 September, 2025; originally announced September 2025.

Comments: Submitted to AAS journals

Resolution-Corrected White Dwarf Gravitational Redshifts Validate SDSS-V Wavelength Calibration and Enable Accurate Mass-Radius Tests

Authors: Stefan M. Arseneau, J. J. Hermes, Nadia L. Zakamska, Kareem El-Badry, Nicole R. Crumpler, Vedant Chandra, Gautham Adamane Pallathadka, Carles Badenes, Boris T. Gaensicke, Nicola Gentile Fusillo

Abstract: Leveraging the large sample size of low-resolution spectroscopic surveys to constrain white dwarf stellar structure requires an accurate understanding of the shapes of hydrogen absorption lines, which are pressure broadened by the Stark effect. Using data from both the Sloan Digital Sky Survey and the Type Ia Supernova Progenitor Survey, we show that substantial biases (5-15 km/s) exist in radial… ▽ More Leveraging the large sample size of low-resolution spectroscopic surveys to constrain white dwarf stellar structure requires an accurate understanding of the shapes of hydrogen absorption lines, which are pressure broadened by the Stark effect. Using data from both the Sloan Digital Sky Survey and the Type Ia Supernova Progenitor Survey, we show that substantial biases (5-15 km/s) exist in radial velocity measurements made from observations at low spectral resolution relative to similar measurements from high-resolution spectra. Our results indicate that the physics of line formation in high-density plasmas, especially in the wings of the lines, are not fully accounted for in state-of-the-art white dwarf model atmospheres. We provide corrections to account for these resolution-induced redshifts in a way that is independent of an assumed mass-radius relation, and we demonstrate that statistical measurements of gravitational redshift with these corrections yield improved agreement with theoretical mass-radius relations. Our results provide a set of best practices for white dwarf radial velocity measurements from low-resolution spectroscopy, including those from the Sloan Digital Sky Survey, the Dark Energy Spectroscopic Instrument, the 4-meter Multi-Object Spectroscopic Telescope, and the Wide-Field Multiplexed Spectroscopic Facility. △ Less

Submitted 6 August, 2025; originally announced August 2025.

Comments: 17 pages, 11 figures, 2 tables, accepted for publication in ApJ

Mapping the Distant and Metal-Poor Milky Way with SDSS-V

Authors: Vedant Chandra, Phillip A. Cargile, Alexander P. Ji, Charlie Conroy, Hans-Walter Rix, Emily Cunningham, Bruno Dias, Chervin Laporte, William Cerny, Guilherme Limberg, Avrajit Bandyopadhyay, Ana Bonaca, Andrew R. Casey, John Donor, Jose G. Fernandez-Trincado, Peter M. Frinchaboy, Pramod Gupta, Keith Hawkins, Jennifer A. Johnson, Juna A. Kollmeier, Madeline Lucey, Ilija Medan, Szabolcs Meszaros, Sean Morrison, Jose Sanchez-Gallego , et al. (6 additional authors not shown)

Abstract: The fifth-generation Sloan Digital Sky Survey (SDSS-V) is conducting the first all-sky low-resolution spectroscopic survey of the Milky Way's stellar halo. We describe the stellar parameter pipeline for the SDSS-V halo survey, which simultaneously models spectra, broadband photometry, and parallaxes to derive stellar parameters, metallicities, alpha abundances, and distances. The resulting BOSS-MI… ▽ More The fifth-generation Sloan Digital Sky Survey (SDSS-V) is conducting the first all-sky low-resolution spectroscopic survey of the Milky Way's stellar halo. We describe the stellar parameter pipeline for the SDSS-V halo survey, which simultaneously models spectra, broadband photometry, and parallaxes to derive stellar parameters, metallicities, alpha abundances, and distances. The resulting BOSS-MINESweeper catalog is validated across a wide range of stellar parameters and metallicities using star clusters and a comparison to high-resolution spectroscopic surveys. We demonstrate several scientific capabilities of this dataset: identifying the most chemically peculiar stars in our Galaxy, discovering and mapping distant halo substructures, and measuring the all--sky dynamics of the Milky Way on the largest scales. The BOSS-MINESweeper catalog for SDSS DR19 is publicly available and will be updated for future data releases. △ Less

Submitted 1 August, 2025; originally announced August 2025.

Comments: 31 pages, 19 figures; Submitted to AAS Journals;

A Large Catalog of DA White Dwarf Characteristics Using SDSS and Gaia Observations

Authors: Nicole R. Crumpler, Vedant Chandra, Nadia L. Zakamska, Gautham Adamane Pallathadka, Stefan Arseneau, Nicola Gentile Fusillo, J. J. Hermes, Carles Badenes, Priyanka Chakraborty, Boris T. Gänsicke, Sean Morrison, Hans-Walter Rix, Stephen P. Schmidt, Axel Schwope, Keivan G. Stassun

Abstract: We present a catalog of 8545 and 19,257 unique DA white dwarfs observed in SDSS Data Release 19 and previous SDSS data releases, respectively. This is the largest catalog of both spectroscopic and photometric measurements of DA white dwarfs available to date, and we make this catalog and all code used to create it publicly available. We measure the apparent radial velocity, spectroscopic effective… ▽ More We present a catalog of 8545 and 19,257 unique DA white dwarfs observed in SDSS Data Release 19 and previous SDSS data releases, respectively. This is the largest catalog of both spectroscopic and photometric measurements of DA white dwarfs available to date, and we make this catalog and all code used to create it publicly available. We measure the apparent radial velocity, spectroscopic effective temperature and surface gravity, and photometric effective temperature and radius for all objects in our catalog. We validate our measurements against other published white dwarf catalogs. For apparent radial velocities, surface gravities, and effective temperatures measured from spectra with signal-to-noise ratios $>50$, our measurements agree with published SDSS white dwarf catalogs to within 7.5 km/s, 0.060 dex, and $2.4\%$, respectively. For radii and effective temperatures measured with Gaia photometry, our measurements agree with other published Gaia datasets to within $0.0005$ $R_\odot$ and $3\%$, respectively. We use this catalog to investigate systematic discrepancies between white dwarfs observed in SDSS-V and previous generations of SDSS. For objects observed in both SDSS-V and previous generations, we uncover systematic differences between measured spectroscopic parameters depending on which set of survey data is used. On average, the measured apparent radial velocity of a DA white dwarf is $11.5$ km/s larger and the surface gravity is $0.015$ dex smaller when a white dwarf's spectroscopic parameters are measured using SDSS-V data compared to using data from previous generations of SDSS. These differences may be due to changes in the wavelength solution across survey generations. △ Less

Submitted 1 August, 2025; originally announced August 2025.

Journal ref: ApJ 989 24 (2025)

Electric, thermal and thermoelectric response of a hot pion gas in a time dependent background magnetic field

Authors: Ankit Kumar, Gowthama K K, Vinod Chandra, Sadhana Dash

Abstract: The prime focus of the work is to determine the electric, thermal and thermoelectric transport coefficients of a hot pion gas in the presence of time-dependent background magnetic fields. The thermoelectric effect is analyzed by examining the magneto-Seebeck and Nernst coefficients in the hot pionic medium under such conditions. Furthermore, the phenomenologically relevant elliptic flow coefficien… ▽ More The prime focus of the work is to determine the electric, thermal and thermoelectric transport coefficients of a hot pion gas in the presence of time-dependent background magnetic fields. The thermoelectric effect is analyzed by examining the magneto-Seebeck and Nernst coefficients in the hot pionic medium under such conditions. Furthermore, the phenomenologically relevant elliptic flow coefficient, linked to the Knudsen number, is examined. The analysis reveals the significant impact of both the strength and time dependence of the magnetic field on the transport coefficients of the pionic medium. The results are analyzed in contrast to those obtained under a constant magnetic field. △ Less

Submitted 31 July, 2025; originally announced July 2025.

Comments: 11 pages, 7 figures, two-column

A Spectroscopic Search for Dormant Black Holes in Low-Metallicity Binaries

Authors: Pranav Nagarajan, Kareem El-Badry, Henrique Reggiani, Casey Y. Lam, Joshua D. Simon, Johanna Müller-Horn, Rhys Seeburger, Hans-Walter Rix, Howard Isaacson, Jessica Lu, Vedant Chandra, Rene Andrae

Abstract: The discovery of the massive black hole (BH) system Gaia BH3 in pre-release Gaia DR4 data suggests that wide BH binaries with luminous companions may be significantly overrepresented at low metallicities. Motivated by this finding, we have initiated a spectroscopic survey of low-metallicity stars exhibiting elevated RUWE values in Gaia DR3, using the FEROS and APF spectrographs. We identify promis… ▽ More The discovery of the massive black hole (BH) system Gaia BH3 in pre-release Gaia DR4 data suggests that wide BH binaries with luminous companions may be significantly overrepresented at low metallicities. Motivated by this finding, we have initiated a spectroscopic survey of low-metallicity stars exhibiting elevated RUWE values in Gaia DR3, using the FEROS and APF spectrographs. We identify promising BH binary candidates as objects with instantaneously measured radial velocities (RVs) that are very different from their mean RVs reported in Gaia DR3. Thus far, we have observed over 500 targets, including a nearly complete sample of stars with $\text{[Fe/H]} < -1.5$, RUWE $> 2$, and $G < 15$. Our search has yielded one promising target exhibiting slow acceleration and an RV more than 98 km s$^{-1}$ different from its DR3 mean RV, as well as dozens of other candidates with smaller RV discrepancies. We quantify the sensitivity of our search using simulations, demonstrating that it recovers at least half of the BH companions within our selection criteria. We make all the spectra and RVs from our survey publicly available and encourage further follow-up. △ Less

Submitted 27 August, 2025; v1 submitted 16 July, 2025; originally announced July 2025.

Comments: 40 pages, 12 figures, Accepted to PASP. GitHub repository at https://github.com/pranav-nagarajan/Low-Metallicity-Binaries . Spectra available at https://doi.org/10.5281/zenodo.16888415

Double White Dwarf Binaries in SDSS-V DR19 : The discovery of a rare DA+DQ white dwarf binary with 31 hour orbital period

Authors: Gautham Adamane Pallathadka, Vedant Chandra, Boris T. Gansicke, Nadia L. Zakamska, Detlev Koester, Yossef Zenati, Nicole R. Crumpler, Stefan M. Arseneau, J. J. Hermes, Matthias R. Schreiber, Keivan G. Stassun, Axel Schwope, Kareem El-Badry, Gagik Tovmassian, Tim Cunningham, Sean Morrison

Abstract: Binaries of two white dwarfs (WDs) are an important class of astrophysical objects that are theorized to lead to Type Ia supernovae and are also used to gain insight into complex processes involved in stellar binary evolution. We report the discovery of SDSS~J090618.44+022311.6, a rare post-common envelope binary of a hydrogen atmospheric DA WD and a DQ WD which shows carbon absorption features, a… ▽ More Binaries of two white dwarfs (WDs) are an important class of astrophysical objects that are theorized to lead to Type Ia supernovae and are also used to gain insight into complex processes involved in stellar binary evolution. We report the discovery of SDSS~J090618.44+022311.6, a rare post-common envelope binary of a hydrogen atmospheric DA WD and a DQ WD which shows carbon absorption features, and is only the fourth such binary known. We combine the available spectroscopic, photometric, and radial velocity data to provide a self-consistent model for the binary and discuss its history as a binary DA+DQ. The system has a period of 31.17 hours with masses of 0.42 M$_{\odot}$ for DA WD and 0.49 M$_{\odot}$ for DQ WD. The corresponding cooling ages point to an Algol type of evolution with the lower mass star evolving into a DA WD first and later the massive DQ WD is formed. The system has a merger timescale of 450 Gyrs and will lead to the formation of a massive WD. With this, the number of known DA+DQ WD binaries has increased to four, and we find that their stellar properties all lie in the same range. Detailed study of more such systems is vital to understand common processes involved in the formation of this rare class of binaries and give insights towards the broader picture of WD spectral evolution. △ Less

Submitted 15 July, 2025; originally announced July 2025.

Comments: Submitted to AAS journals

The Nineteenth Data Release of the Sloan Digital Sky Survey

Authors: SDSS Collaboration, Gautham Adamane Pallathadka, Mojgan Aghakhanloo, James Aird, Andrés Almeida, Singh Amrita, Friedrich Anders, Scott F. Anderson, Stefan Arseneau, Consuelo González Avila, Shir Aviram, Catarina Aydar, Carles Badenes, Jorge K. Barrera-Ballesteros, Franz E. Bauer, Aida Behmard, Michelle Berg, F. Besser, Christian Moni Bidin, Dmitry Bizyaev, Guillermo Blanc, Michael R. Blanton, Jo Bovy, William Nielsen Brandt, Joel R. Brownstein , et al. (187 additional authors not shown)

Abstract: Mapping the local and distant Universe is key to our understanding of it. For decades, the Sloan Digital Sky Survey (SDSS) has made a concerted effort to map millions of celestial objects to constrain the physical processes that govern our Universe. The most recent and fifth generation of SDSS (SDSS-V) is organized into three scientific ``mappers". Milky Way Mapper (MWM) that aims to chart the var… ▽ More Mapping the local and distant Universe is key to our understanding of it. For decades, the Sloan Digital Sky Survey (SDSS) has made a concerted effort to map millions of celestial objects to constrain the physical processes that govern our Universe. The most recent and fifth generation of SDSS (SDSS-V) is organized into three scientific ``mappers". Milky Way Mapper (MWM) that aims to chart the various components of the Milky Way and constrain its formation and assembly, Black Hole Mapper (BHM), which focuses on understanding supermassive black holes in distant galaxies across the Universe, and Local Volume Mapper (LVM), which uses integral field spectroscopy to map the ionized interstellar medium in the local group. This paper describes and outlines the scope and content for the nineteenth data release (DR19) of SDSS and the most substantial to date in SDSS-V. DR19 is the first to contain data from all three mappers. Additionally, we also describe nine value added catalogs (VACs) that enhance the science that can be conducted with the SDSS-V data. Finally, we discuss how to access SDSS DR19 and provide illustrative examples and tutorials. △ Less

Submitted 9 July, 2025; originally announced July 2025.

Comments: Submitted to AASJournals. 56 Pages, 9 Tables, 11 Figures

Sloan Digital Sky Survey-V: Pioneering Panoptic Spectroscopy

Authors: Juna A. Kollmeier, Hans-Walter Rix, Conny Aerts, James Aird, Pablo Vera Alfaro, Andrés Almeida, Scott F. Anderson, Óscar Jiménez Arranz, Stefan M. Arseneau, Roberto Assef, Shir Aviram, Catarina Aydar, Carles Badenes, Avrajit Bandyopadhyay, Kat Barger, Robert H. Barkhouser, Franz E. Bauer, Chad Bender, Felipe Besser, Binod Bhattarai, Pavaman Bilgi, Jonathan Bird, Dmitry Bizyaev, Guillermo A. Blanc, Michael R. Blanton , et al. (195 additional authors not shown)

Abstract: The Sloan Digital Sky Survey-V (SDSS-V) is pioneering panoptic spectroscopy: it is the first all-sky, multi-epoch, optical-to-infrared spectroscopic survey. SDSS-V is mapping the sky with multi-object spectroscopy (MOS) at telescopes in both hemispheres (the 2.5-m Sloan Foundation Telescope at Apache Point Observatory and the 100-inch du Pont Telescope at Las Campanas Observatory), where 500 zonal… ▽ More The Sloan Digital Sky Survey-V (SDSS-V) is pioneering panoptic spectroscopy: it is the first all-sky, multi-epoch, optical-to-infrared spectroscopic survey. SDSS-V is mapping the sky with multi-object spectroscopy (MOS) at telescopes in both hemispheres (the 2.5-m Sloan Foundation Telescope at Apache Point Observatory and the 100-inch du Pont Telescope at Las Campanas Observatory), where 500 zonal robotic fiber positioners feed light from a wide-field focal plane to an optical (R$\sim 2000$, 500 fibers) and a near-infrared (R$\sim 22,000$, 300 fibers) spectrograph. In addition to these MOS capabilities, the survey is pioneering ultra wide-field ($\sim$ 4000~deg$^2$) integral field spectroscopy enabled by a new dedicated facility (LVM-I) at Las Campanas Observatory, where an integral field spectrograph (IFS) with 1801 lenslet-coupled fibers arranged in a 0.5 degree diameter hexagon feeds multiple R$\sim$4000 optical spectrographs that cover 3600-9800 angstroms. SDSS-V's hardware and multi-year survey strategy are designed to decode the chemo-dynamical history of the Milky Way Galaxy and tackle fundamental open issues in stellar physics in its Milky Way Mapper program, trace the growth physics of supermassive black holes in its Black Hole Mapper program, and understand the self-regulation mechanisms and the chemical enrichment of galactic ecosystems at the energy-injection scale in its Local Volume Mapper program. The survey is well-timed to multiply the scientific output from major all-sky space missions. The SDSS-V MOS programs began robotic operations in 2021; IFS observations began in 2023 with the completion of the LVM-I facility. SDSS-V builds upon decades of heritage of SDSS's pioneering advances in data analysis, collaboration spirit, infrastructure, and product deliverables in astronomy. △ Less

Submitted 9 July, 2025; originally announced July 2025.

Comments: 76 pages, submitted to the Astronomical Journal

AutoMixer: Checkpoint Artifacts as Automatic Data Mixers

Authors: Ernie Chang, Yang Li, Patrick Huber, Vish Vogeti, David Kant, Yangyang Shi, Vikas Chandra

Abstract: In language model training, it is desirable to equip models with capabilities from various tasks. However, it is not clear how to directly obtain the right data mixtures for these capabilities as the relationship between data and tasks is difficult to be modeled. In this work, we observe that checkpoint models exhibit emerging capabilities at different points in the training trajectory. Often, the… ▽ More In language model training, it is desirable to equip models with capabilities from various tasks. However, it is not clear how to directly obtain the right data mixtures for these capabilities as the relationship between data and tasks is difficult to be modeled. In this work, we observe that checkpoint models exhibit emerging capabilities at different points in the training trajectory. Often, the training process saves checkpoints as artifacts that are under-utilized as a source of in-training data signals. We identify these artifact models based on their respective capabilities on the benchmarks and leverage them as data mixers by using their aggregated first-order influence approximation over source data. We demonstrated on eight reasoning benchmarks that the proposed framework shows significant improvements in the pretraining setting, with performance improvements of up to 1.93%. Overall, this shows the potential of checkpoint models to enhance data quality and optimize data mixtures. △ Less

Submitted 8 February, 2026; v1 submitted 27 June, 2025; originally announced June 2025.

Comments: Accepted at ACL 2025

SPLATART: Articulated Gaussian Splatting with Estimated Object Structure

Authors: Stanley Lewis, Vishal Chandra, Tom Gao, Odest Chadwicke Jenkins

Abstract: Representing articulated objects remains a difficult problem within the field of robotics. Objects such as pliers, clamps, or cabinets require representations that capture not only geometry and color information, but also part seperation, connectivity, and joint parametrization. Furthermore, learning these representations becomes even more difficult with each additional degree of freedom. Complex… ▽ More Representing articulated objects remains a difficult problem within the field of robotics. Objects such as pliers, clamps, or cabinets require representations that capture not only geometry and color information, but also part seperation, connectivity, and joint parametrization. Furthermore, learning these representations becomes even more difficult with each additional degree of freedom. Complex articulated objects such as robot arms may have seven or more degrees of freedom, and the depth of their kinematic tree may be notably greater than the tools, drawers, and cabinets that are the typical subjects of articulated object research. To address these concerns, we introduce SPLATART - a pipeline for learning Gaussian splat representations of articulated objects from posed images, of which a subset contains image space part segmentations. SPLATART disentangles the part separation task from the articulation estimation task, allowing for post-facto determination of joint estimation and representation of articulated objects with deeper kinematic trees than previously exhibited. In this work, we present data on the SPLATART pipeline as applied to the syntheic Paris dataset objects, and qualitative results on a real-world object under spare segmentation supervision. We additionally present on articulated serial chain manipulators to demonstrate usage on deeper kinematic tree structures. △ Less

Submitted 13 June, 2025; originally announced June 2025.

Comments: 7 pages, Accepted to the 2025 RSS Workshop on Gaussian Representations for Robot Autonomy. Contact: Stanley Lewis, stanlew@umich.edu

On the flow topology of swirl jets upon impingement

Authors: Premchand V. Chandra, Pradip Dutta

Abstract: Jet impingement enhances heat transfer and is characterised by the complex flow patterns formed when a jet impacts a plate aligned normal to it. While traditional round jet impingement has been extensively studied to understand flow and associated heat transfer, there is still room for research in investigating flow structures in swirl jet impingement. This paper focuses on the flow topology of sw… ▽ More Jet impingement enhances heat transfer and is characterised by the complex flow patterns formed when a jet impacts a plate aligned normal to it. While traditional round jet impingement has been extensively studied to understand flow and associated heat transfer, there is still room for research in investigating flow structures in swirl jet impingement. This paper focuses on the flow topology of swirl jets generated by a 45-degree vane swirler, impinging on a flat plate studied at dimensionless jet-plate distances (H/D=1-4) and Reynolds numbers (Re = 16600 and 23000). The flow structures, mean velocity components, and turbulence characteristics are presented using a 2D Particle Image Velocimetry (PIV) experiment at the front and top planes. Furthermore, results from the 3D numerical simulations are presented to support the results where the PIV study had experimental limitations. The effect of impingement distance or jet-plate distance on the mean flow properties and turbulence parameters is discussed. A Proper Orthogonal Decomposition (POD) analysis has been performed to understand the dominant coherent structures in different cases of impingement distance. We show that the turbulence parameters are more pronounced at smaller jet-plate distances $(H/D \leq 2)$, which could explain the enhanced heat transfer for these jets. △ Less

Submitted 29 May, 2025; originally announced May 2025.

Comments: 27 pages, 30 figures, The manuscript is yet to be submitted to a peer review journal for publication

The Cocytos Stream: A Disrupted Globular Cluster from our Last Major Merger?

Authors: Christian Aganze, Vedant Chandra, Risa H. Wechsler, Ting S. Li, Sergey E. Koposov, Leandro Beraldo Silva, Andreia Carrillo, Alexander H. Riley, Monica Valluri, Oleg Y. Gnedin, Mairead Heiger, Constance Rockosi, Raymond Carlberg, Amanda Byström, Namitha Kizhuprakkat, Mika Lambert, Bokyoung Kim, Gustavo Medina Toledo, Carlos Allende Prieto, Jessica Nicole Aguilar, Steven Ahlen, Davide Bianchi, David Brooks, Todd T. Claybaugh, Andrew P. Cooper , et al. (25 additional authors not shown)

Abstract: The census of stellar streams and dwarf galaxies in the Milky Way provides direct constraints on galaxy formation models and the nature of dark matter. The DESI Milky Way survey (with a footprint of 14,000$~deg{^2}$ and a depth of $r<19$ mag) delivers the largest sample of distant metal-poor stars compared to previous optical fiber-fed spectroscopic surveys. This makes DESI an ideal survey to sear… ▽ More The census of stellar streams and dwarf galaxies in the Milky Way provides direct constraints on galaxy formation models and the nature of dark matter. The DESI Milky Way survey (with a footprint of 14,000$~deg{^2}$ and a depth of $r<19$ mag) delivers the largest sample of distant metal-poor stars compared to previous optical fiber-fed spectroscopic surveys. This makes DESI an ideal survey to search for previously undetected streams and dwarf galaxies. We present a detailed characterization of the Cocytos stream, which was re-discovered using a clustering analysis with a catalog of giants in the DESI year 3 data, supplemented with Magellan/MagE spectroscopy. Our analysis reveals a relatively metal-rich ([Fe/H]$=-1.3$) and thick stream (width$=1.5^\circ$) at a heliocentric distance of $\approx 25$ kpc, with an internal velocity dispersion of 6.5-9 km s$^{-1}$. The stream's metallicity, radial orbit, and proximity to the Virgo stellar overdensities suggest that it is most likely a disrupted globular cluster that came in with the Gaia-Enceladus merger. We also confirm its association with the Pyxis globular cluster. Our result showcases the ability of wide-field spectroscopic surveys to kinematically discover faint disrupted dwarfs and clusters, enabling constraints on the dark matter distribution in the Milky Way. △ Less

Submitted 15 April, 2025; originally announced April 2025.

Comments: To be submitted to ApJ, comments welcome

Nonperturbative heavy quark diffusion coefficients in a weakly magnetized thermal QCD medium

Authors: Debarshi Dey, Aritra Bandyopadhyay, Santosh K. Das, Sadhana Dash, Vinod Chandra, Basanta K. Nandi

Abstract: In this work, the perturbative and non-perturbative contributions to the heavy quark (HQ) momentum ($κ$) as well as spatial ($D_s$) diffusion coefficients are computed in a weak background magnetic field. The formalism adopted here involves calculation of the in-medium potential of the HQ in a weak magnetic field, which then serves as a proxy for the resummed gluon propagator in the calculation of… ▽ More In this work, the perturbative and non-perturbative contributions to the heavy quark (HQ) momentum ($κ$) as well as spatial ($D_s$) diffusion coefficients are computed in a weak background magnetic field. The formalism adopted here involves calculation of the in-medium potential of the HQ in a weak magnetic field, which then serves as a proxy for the resummed gluon propagator in the calculation of HQ self-energy ($Σ$). The self-energy determines the scattering rate of HQs with light thermal partons, which is subsequently used to evaluate $κ$ and $D_s$. It is observed that non-perturbative effects play a dominant role at low temperature. The spatial diffusion coefficient $2πT D_s$, exhibits good agreement with recent LQCD results. These findings can be applied to calculate the heavy quark directed flow at RHIC and LHC energies. An extension of this formalism to the case of finite HQ momentum has also been attempted. △ Less

Submitted 9 July, 2025; v1 submitted 3 April, 2025; originally announced April 2025.

Comments: 17 pages, 7 figures, version 2

Journal ref: Physical Review D 112, 016011 (2025)

Self-Vocabularizing Training for Neural Machine Translation

Authors: Pin-Jie Lin, Ernie Chang, Yangyang Shi, Vikas Chandra

Abstract: Past vocabulary learning techniques identify relevant vocabulary before training, relying on statistical and entropy-based assumptions that largely neglect the role of model training. Empirically, we observe that trained translation models are induced to use a byte-pair encoding (BPE) vocabulary subset distinct from the original BPE vocabulary, leading to performance improvements when retrained wi… ▽ More Past vocabulary learning techniques identify relevant vocabulary before training, relying on statistical and entropy-based assumptions that largely neglect the role of model training. Empirically, we observe that trained translation models are induced to use a byte-pair encoding (BPE) vocabulary subset distinct from the original BPE vocabulary, leading to performance improvements when retrained with the induced vocabulary. In this paper, we analyze this discrepancy in neural machine translation by examining vocabulary and entropy shifts during self-training--where each iteration generates a labeled dataset by pairing source sentences with the model's predictions to define a new vocabulary. Building on these insights, we propose self-vocabularizing training, an iterative method that self-selects a smaller, more optimal vocabulary, yielding up to a 1.49 BLEU improvement. Moreover, we find that deeper model architectures lead to both an increase in unique token usage and a 6-8% reduction in vocabulary size. △ Less

Submitted 31 March, 2025; v1 submitted 17 March, 2025; originally announced March 2025.

Comments: Accepted to NAACL SRW 2025

Spin wave interactions in the pyrochlore Heisenberg antiferromagnet with Dzyaloshinskii-Moriya interactions

Authors: V. V. Jyothis, Kallol Mondal, Himanshu Mavani, V. Ravi Chandra

Abstract: We study the effect of magnon interactions on the spin wave spectra of the all-in-all-out phase of the pyrochlore nearest neighbour antiferromagnet with a Dzyaloshinskii-Moriya interaction ($D$). The leading order corrections to spin wave energies indicate a significant renormalisation for commonly encountered strengths of the Dzyaloshinskii-Moriya term. For low values of $D$ we find a potential i… ▽ More We study the effect of magnon interactions on the spin wave spectra of the all-in-all-out phase of the pyrochlore nearest neighbour antiferromagnet with a Dzyaloshinskii-Moriya interaction ($D$). The leading order corrections to spin wave energies indicate a significant renormalisation for commonly encountered strengths of the Dzyaloshinskii-Moriya term. For low values of $D$ we find a potential instability of the phase itself, indicated by the renormalisation of magnon frequencies to negative values. We have also studied the renormalized spectra in the presence of magnetic fields along three high symmetry directions of the lattice, namely the $[111]$, $[100]$ and $[110]$ directions. Generically, we find that for a fixed value of the Dzyaloshinskii-Moriya interaction renormalized spectra for the lowest band decrease with an increasing strength of the field. We have also analyzed the limits of the two magnon continuum and probed the possibility of magnon decay. For a range of $D$ and the field strength we identify possible parameter regimes where the decay of the higher bands of the system are kinematically allowed. △ Less

Submitted 23 July, 2026; v1 submitted 13 February, 2025; originally announced February 2025.

Comments: This version is close to the published version in Phys. Rev. B. The text has been expanded at several places and more steps/expressions have been provided explicitly for clarity. An addition to the analysis classical ground states pertaining to the field dependence of canting angles is presented in the Appendix

Journal ref: Phys. Rev. B 114, 024418 (2026)