arXiv subjects

Jason Wang

Publications and source records attributed to Jason Wang.

At least 19 recordsLinked to original sources

arXiv2026

The Roman Coronagraph Community Participation Program: corgisim - a simulation suite for the Nancy Grace Roman Space Telescope Coronagraph Instrument

NASA's Roman Space Telescope will feature a pathfinder Coronagraph Instrument to demonstrate advanced high-contrast imaging from space, paving the way for future missions like the Habitable Worlds Observatory. The Coronagraph Instrument could obtain imaging, polarimetry and spectroscopy of Jupiter analogs in reflected visible light for the first time. We present the development of an open-source simulation package ``corgisim'' as part of the Roman Coronagraph Community Participate Program. Built on established optical propagation libraries including PROPER and CGISim, corgisim provides a user-friendly, publicly available Python framework for end-to-end simulations of the Coronagraph Instrument observations. The package produces high-fidelity, format-compliant data for pre-launch calibration, pipeline testing, and community applications such as target selection and observation planning. We will give an overview of corgisim's infrastructure, functionalities, and current implementation across planned imaging, polarimetry, and spectroscopy modes, including the ability to simulate host stars, injected companions, and extended disks. We will also highlight suitable applications of corgisim and provide guidance on how users can access and employ the software.

astro-ph.IM↗

arXiv2026

The Roman Coronagraph Community Participation Program: data reduction pipeline astrometric calibration

The Nancy Grace Roman Space Telescope will be equipped with a Technology Demonstration Coronagraph Instrument that will push the current limits of high contrast imaging for exoplanets ($10^{-9}$ contrast). The Roman Coronagraph Community Participation Program has developed corgidrp, a python-based data reduction pipeline for the Roman Coronagraph Instrument that will perform essential data processing and calibration steps for coronagraphic observations. The astrometric calibration function within corgidrp allows us to understand the on-sky angular size and distance scale of science observations by characterizing essential detector parameters: boresight, plate scale, north angle, and optical distortion. Measuring these astrometric calibration products not only helps us understand the science output of our data, but it is what allows us to point the Roman coronagraph accurately at our science target in the first place. Here, we describe the techniques used within the astrometric calibration and demonstrate that our algorithm meets Technology Demonstration Threshold Requirements: (1) compute the on-sky location of the center of CGI EXCAM detector to better than 30 [mas] and (2) compute the on-sky position angles of the camera axes to within 0.3 [deg].

astro-ph.IM↗

arXiv2026

MobileMem: Learning from a Year of Mobile Experiences

The next generation of AI agents is increasingly moving beyond systems that answer isolated questions toward persistent personal assistants that can understand, remember, and continuously learn from users' experiences. Such assistants require long-term memory to accumulate and leverage user-specific experiences over time, yet existing benchmarks remain inadequate for realistic mobile settings, where experiences are heterogeneous, multimodal, evolving, and deeply personal. We introduce MobileMem, a benchmark and framework for studying on-device long-term memory, grounded in a year-scale collection of mobile experiences. MobileMem employs a knowledge-grounded synthesis pipeline to construct coherent and temporally consistent long-horizon trajectories from user-app sessions. It provides complementary text and multimodal settings covering multi-hop and temporal reasoning, knowledge updating, and implicit preference inference. Specifically, MobileMem enables agents to remember the past, understand the present, and adapt to the future. By modeling experiences rather than isolated facts, MobileMem moves memory beyond information retrieval toward experiential intelligence for continuous personal learning.

cs.AI↗

arXiv2026

Jointly Modeling Roman Coronagraph Astrometry and Photometry Improves Orbital Parameter Estimates

Launching in 2027, the Nancy Roman Grace Space Telescope (Roman) has the potential to directly image exoplanets in reflected light for the first time. Roman imaging will introduce new constraints on exoplanet orbital parameters, since reflected-light intensity depends on orbital phase. In this Note, we discuss an addition to the open-source Python package orbitize!, which allows users to model exoplanet orbits using joint constraints from astrometry and photometric variations due to orbital phase. To investigate the impact of adding photometric data into the orbital model, we simulated realistic measurements of partial orbits, both including and excluding photometry in our model, and computed orbital posteriors. We found that fitting both astrometry and photometry improves posterior precision relative to fitting astrometry alone. This effect was more pronounced for higher-SNR images; for example, photometric data with SNR=10 yielded 33% improvement in inclination precision when including photometry, while SNR=3 data yielded only 12% improvement.

astro-ph.IM↗

arXiv2026

Direct Imaging Discovery of Giant Exoplanet $\beta$ Pictoris d: A Decade-Long Game of Hide-and-Seek

We report the direct imaging discovery of a third exoplanet in the $\beta$ Pictoris system. We detect $\beta$ Pictoris d ($\beta$ Pic d) in non-coronagraphic observations obtained with VLT/ERIS as well as multi-epoch archival datasets from JWST/NIRCam and VLT/SPHERE. Astrometric measurements over an 11-year baseline demonstrate that it is consistent with a gravitationally-bound source with orbital motion. Joint multi-planet orbit fits of all three planets in the system yield a semi-major axis of $26.0^{+2.2}_{-6.1}$ au and inclination $89.0^{+0.7}_{-0.6}$ deg for planet d. $\beta$ Pic d has a larger orbital semi-major axis than the other known planets in the system, but is coplanar with the inner two planets, and its orbit is consistent with sculpting the inner edge of the debris disk. $\beta$ Pic d has a contrast of $\Delta L^{\prime}=12.11\pm0.15$ mag, with colors and luminosity that closely match those of 51 Eri b, another exoplanet in the $\beta$ Pictoris moving group. Its VLT/ERIS and JWST/NIRCam colors are distinct from those of free-floating planetary-mass objects of a similar age and temperature. Its red $F410M-F444W$ color indicates strong CO$_2$ absorption in its atmosphere and suggests significant enhancement in metals compared to free-floating objects. From the ATMO hot-start evolutionary models, we estimate an effective temperature of $600^{+45}_{-60}$ K and mass of $2.4\pm0.6$ $M_{\rm Jup}$, which also closely matches similar estimates for 51 Eri b. $\beta$ Pic d is among the lowest-mass exoplanets imaged from the ground. This discovery highlights the deep sensitivity achievable with ground-based imaging in the mid-infrared and the discovery potential of future high-contrast observations with the Extremely Large Telescope.

astro-ph.EP↗

arXiv2026

CheckMIABench: Firm Foundations For Membership Inference Attacks on Language Models

Membership inference attacks (MIAs) are a canonical way to assess a machine learning model's privacy properties. Although several attempts have been made to evaluate MIAs on language models, the extant literature has suffered numerous difficulties in constructing clean evaluations to test new techniques. In particular, subtle distribution shifts between member and non-member sets can undermine the statistical validity of MIAs; recent work has underscored this by showing that "blind" methods with no access to the underlying model can perform far better than published methods on the same benchmarks. This paper constructs a benchmark for principled evaluation of MIAs against LLMs, by leveraging the insight that training data before and after a fixed point during training are drawn from the same distribution. Therefore, all open-source models with intermediate checkpoints and public training data can be converted into MIA testbeds. We apply our framework to a half-dozen published attacks on the Pythia and OLMo family of models, from 70M to 7B parameters. To facilitate further privacy research, we open-source a modular library for designing and implementing attacks in this setting: https://github.com/safr-ai-lab/pandora_llm.

cs.LG↗

arXiv2026

$^{13}$CO and potential variability in $\beta$ Pictoris b with GRAVITY+

The $^{12}$CO/$^{13}$CO ratio was introduced as an indicator for where in the disk a planet has formed. Previously a lower value compared to the host star's was suggested to show that a planet accreted CO ice beyond the disk's CO ice line. In this letter we aim to determine the $^{12}$CO/$^{13}$CO value of the directly imaged planet $\beta$ Pictoris b, and whether we can link it to its formation. Its apparent brightness results in an exceptional S/N of up to ~60 per wavelength point. We present the first science observations with the upgraded GRAVITY+ instrument at a spectral resolution of R ~ 4000, which we analyse with petitRADTRANS. Our retrievals robustly indicate the presence of $^{13}$CO with a $^{12}$CO/$^{13}$CO ratio of 91$^{+24}_{-17}$, consistent with both a solar to ISM-like value. Our $^{12}$CO/$^{13}$CO value corroborates recent interpretations that $^{13}$CO may be a less useful tracer of formation location in the disk than previously thought; nonetheless, we discuss theories with which this value is consistent. As our observations span ~7 hours, this enabled us to search for atmospheric variability in $\beta$ Pictoris b; we report a tentative constraint on the variability amplitude of about 1.4$^{+0.6}_{-0.7}$%.

astro-ph.EP↗

arXiv2026

Non-Learning Low-Light Stereo Vision

We present a non-learning stereo framework for disparity estimation from severely noisy images. Using the Field of Junctions (FoJ), it retains coarse visual features stable under severe noise for cost volume construction while discarding fine textures inseparable from photon noise. The resulting structural information guides boundary-aware Semi-Global Matching (SGM) that dynamically adapts smoothness penalties to preserve true disparity discontinuities. The output is a sparse disparity map more accurate than those of recent stereo algorithms over unmasked pixels on widely-used benchmark datasets.

cs.CV↗

arXiv2026

${\pi}_{0.7}$: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities

We present a new robotic foundation model, called ${\pi}_{0.7}$, that can enable strong out-of-the-box performance in a wide range of scenarios. ${\pi}_{0.7}$ can follow diverse language instructions in unseen environments, including multi-stage tasks with various kitchen appliances, provide zero-shot cross-embodiment generalization, for example enabling a robot to fold laundry without seeing the task before, and perform challenging tasks such as operating an espresso machine out of the box at a level of performance that matches much more specialized RL-finetuned models. The main idea behind ${\pi}_{0.7}$ is to use diverse context conditioning during training. This conditioning information, contained in the prompt, makes it possible to steer the model precisely to perform many tasks with different strategies. It is conditioned not just on a language command that describes what it should do, but on additional multimodal information that also describes the manner or strategy in which it should do it, including metadata about task performance and subgoal images. This enables ${\pi}_{0.7}$ to use very diverse data, including demonstrations, potentially suboptimal (autonomous) data including failures, and data from non-robot sources. Our experiments evaluate ${\pi}_{0.7}$ across numerous tasks with multiple robot platforms, on tasks that require speed and dexterity, language following, and compositional task generalization.

cs.LG↗

arXiv2026

Discovery of a Low-Mass Companion to the Accelerating Star HIP 53005 with Strongly Conflicting Mass Estimates

We present the discovery of a low-mass companion located at $\rho$ $\sim$ 0\farcs{}85 ($r_{\rm proj} \approx 62~au$) from the early-type 1.2 Gyr-old star HIP 53005 using direct imaging data from the Subaru and Keck Telescopes and astrometry from the Hipparcos-Gaia Catalog of Accelerations. The companion, HIP 53005 C, is a component of a multiple system also including a $\approx$ 12\farcs{}4-separation M dwarf companion inducing a negligible proper motion acceleration. HIP~53005 C's position on color-magnitude diagrams, the fit of its spectral energy distribution to atmosphere models, and its location on an empirical mass-magnitude diagram all suggest that it lies at the M/L transition and near the hydrogen-burning limit ($\sim80~M_{\rm Jup}$). However, our orbital fitting combining direct-imaging relative astrometry with proper motion acceleration favors a much higher dynamical mass of $\sim185\ M_{\rm Jup}$. An additional unseen, more closely-orbiting companion below the detection limit (at $\rho\lesssim0\farcs2$)) may explain this discrepancy. Alternatively, HIP~53005C could be a low-mass binary like Gliese~229Bab, making this system an intriguing laboratory for studying multiple star formation.

astro-ph.SR↗

arXiv2026

Worlds Next Door. IV. Mapping the Late Stages of Giant Planet Evolution with a Precise Dynamical Mass and Luminosity for $\epsilon$ Ind Ab

We present new JWST/NIRCam 4-5 $\mu$m (F410M, F430M) and JWST/MIRI 18-25 $\mu$m (F1800W, F2100W, F2550W) imaging detections of the nearby (3.6 pc) cold (275 K) gas giant exoplanet $\epsilon$ Ind Ab. The F2550W detection of $\epsilon$ Ind Ab constitutes the longest wavelength image of an exoplanet acquired to date. Combining three decades of radial velocity monitoring, Gaia-Hipparcos absolute astrometry, and relative astrometry from direct imaging (including the new NIRCam astrometry), we conduct a comprehensive re-analysis of $\epsilon$ Ind Ab's orbit and obtain a dynamical mass $M_{\rm Ab} = 6.5^{+0.7}_{-0.6}\;M_{\rm Jup}$. Using $\epsilon$ Ind Ab's NIRCam and MIRI photometry, we assemble the first 4-25 $\mu$m spectral energy distribution (SED) of a cold gas giant outside the Solar System. The NIRCam photometry supports a metal-enriched atmosphere for $\epsilon$ Ind Ab based on analysis with atmospheric model grids, consistent with predictions from the giant planet mass-metallicity relation. While the current data do not provide definitive evidence for or against the presence of water ice clouds, we tentatively find that the H$_2$O vapor absorption-dominated F2550W photometry is systematically brighter ($>1\sigma$, but $<2\sigma$) than predictions from cloud-free/rainout chemistry models and better explained by a cloudy model. We calculate a bolometric luminosity of $\log L_{\rm bol}/L_\odot = -7.23 \pm 0.03$ dex by directly integrating $\epsilon$ Ind Ab's SED. Combining this with the planet's dynamical mass and age ($3.5 \pm 1.0$ Gyr), we demonstrate excellent agreement with evolutionary model predictions in a new regime of low luminosities, low masses, and old ages. Our results establish $\epsilon$ Ind Ab as a benchmark system for planetary evolution studies and set the stage for the detailed atmospheric characterization of this temperate extrasolar world.

astro-ph.EP↗

arXiv2026

Evidence for a Peak at $\sim$0.3 in the Eccentricity Distribution of Typical Super-Jovian Exoplanets

In this study, we compute completeness-corrected occurrence rates of giant exoplanets as a function of mass, semimajor axis, and eccentricity, using the approximately uniform California Legacy Survey sample of RV-discovered planets published in Rosenthal et al. 2021. We recover the previously-detected rise in occurrence with semimajor axis for both lower- and higher-mass subsets of the population out to $\sim$5 au. When restricting to planets with semimajor axes between 0.1 and 4.5 au (roughly speaking, the "peak" of giant planet occurrence), we find evidence for distinct eccentricity distributions for each of two mass sub-populations. Most strikingly, we observe a peak in the eccentricity distribution of super-Jovian planets (3-20~M$_{\rm J}$) at 0.3, which is apparent using two different parameterizations of the eccentricity distribution model. A hierarchical histogram model reveals that $\sim$92% of posterior samples indicate an elevated occurrence rate of super-Jupiters with modest eccentricities (0.2-0.4) compared to lower or higher eccentricities (i.e. evidence for a moderate eccentricity "peak"), and 99% of samples indicate super-Jupiters with modest eccentricities are more common than those with lower eccentricities (i.e. evidence that moderate eccentricities are more common than low eccentricities). We use a truncated Gaussian model fit to pinpoint the location of the super-Jupiter eccentricity peak with more precision, finding a maximum a posterior (MAP) peak location of $e=0.3$. This low but elevated characteristic eccentricity could be the result of dynamically hot histories, perhaps involving a giant impacts phase. All analysis code for this project is publicly available on Zenodo (https://zenodo.org/records/18089157) and GitHub (github.com/sblunt/eccentricities).

astro-ph.EP↗

arXiv2026

Yuan3.0 Ultra: A Trillion-Parameter Enterprise-Oriented MoE LLM

We introduce Yuan3.0 Ultra, an open-source Mixture-of-Experts (MoE) large language model featuring 68.8B activated parameters and 1010B total parameters, specially designed to enhance performance on enterprise scenarios tasks while maintaining competitive capabilities on general purpose tasks. We propose Layer-Adaptive Expert Pruning (LAEP) algorithm designed for the pre-training stage of MoE LLMs. In contrast to previous expert pruning approaches that operate primarily in the post-training phase, the proposed algorithm enhances training efficiency by selectively pruning underutilized experts and reorganizing experts across computing devices according to token distribution statistics. Comprehensive experiments demonstrate that LAEP effectively reduces model size and substantially improves pre-training efficiency. When pre-training Yuan3.0 Ultra from scratch original with 1515B parameters, this algorithm delivers a 49\% boost in pre-training efficiency and a 33.3\% reduction in total parameters, while preserving the model's outstanding multi-domain performance. On enterprise scenario benchmarks including Docmatix, ChatRAG, SummEval and MMTab, Yuan3.0 Ultra achieves leading accuracy. The model and codes are publicly available at https://github.com/Yuan-lab-LLM/Yuan3.0-Ultra.

cs.LG↗

arXiv2026

Yuan3.0 Flash: An Open Multimodal Large Language Model for Enterprise Applications

We introduce Yuan3.0 Flash, an open-source Mixture-of-Experts (MoE) MultiModal Large Language Model featuring 3.7B activated parameters and 40B total parameters, specifically designed to enhance performance on enterprise-oriented tasks while maintaining competitive capabilities on general-purpose tasks. To address the overthinking phenomenon commonly observed in Large Reasoning Models (LRMs), we propose Reflection-aware Adaptive Policy Optimization (RAPO), a novel RL training algorithm that effectively regulates overthinking behaviors. In enterprise-oriented tasks such as retrieval-augmented generation (RAG), complex table understanding, and summarization, Yuan3.0 Flash consistently achieves superior performance. Moreover, it also demonstrates strong reasoning capabilities in domains such as mathematics, science, etc., attaining accuracy comparable to frontier model while requiring only approximately 1/4 to 1/2 of the average tokens. Yuan3.0 Flash has been fully open-sourced to facilitate further research and real-world deployment: https://github.com/Yuan-lab-LLM/Yuan3.0.

cs.AI↗

arXiv2025

Direct imaging characterization of cool gaseous planets

Cool gas giant exoplanets, particularly those with properties similar to those of Jupiter and Saturn, remain poorly characterized due to current observational limitations. This white paper outlines the transformative science case for the Habitable Worlds Observatory (HWO) to directly image and spectroscopically characterize a broad range of gaseous exoplanets with effective temperatures below 400 K. The study focuses on determining key atmospheric properties, including molecular composition, cloud and haze characteristics, and temperature structure, across planets of varying sizes and orbital separations. Leveraging reflected light spectroscopy and polarimetry, HWO will enable comparative planetology of cool gas giants orbiting both solar-type and M-dwarf stars, bridging the observational gap between hot exoplanets and Solar System giants. We present observational requirements and survey strategies necessary to uncover correlations between atmospheric properties and planetary or stellar parameters. This effort will establish critical constraints on planetary formation, cloud microphysics, and the role of photochemistry under diverse irradiation conditions. The unique capabilities of HWO will make it the first facility capable of characterizing true exo-Jupiters in reflected light, thus offering an unprecedented opportunity to place the Solar System in a broader galactic context.

astro-ph.IM↗

arXiv2025

Chemical and Isotopic Homogeneity Between the L Dwarf CD-35 2722 B and its Early M Host Star

CD-35 2722 B is an L dwarf companion to the nearby, $\sim 50-200$ Myr old M1 dwarf CD-35 2722 A. We present a detailed analysis of both objects using high-resolution ($R \sim 35,000$) $K$ band spectroscopy from the Keck Planet Imager and Characterizer (KPIC) combined with archival photometry. With a mass of $30^{+5}_{-4} M_{\mathrm{Jup}}$ (planet-to-host mass ratio 0.05) and projected separation of $67\pm4$ AU from its host, CD-35 2722 B likely formed via gravitational instability. We explore whether the chemical composition of the system tells a similar story. Accounting for systematic uncertainties, we find $\mathrm{[M/H]}=-0.16^{+0.03}_{-0.02} \mathrm{(stat)} \pm 0.25 \mathrm{(sys)}$ dex and $^{12}\mathrm{C}/^{13}\mathrm{C}=132^{+20}_{-14}$ for the host, and $\mathrm{[M/H]}=0.27^{+0.07}_{-0.06} (\mathrm{stat}) \pm 0.12 (\mathrm{sys})$ dex, $^{12}\mathrm{CO}/^{13}\mathrm{CO}=159^{+33}_{-24} \mathrm{(stat)}^{+40}_{-33} \mathrm{(sys)}$, and $\mathrm{C/O} = 0.55 \pm 0.01 (\mathrm{stat}) \pm 0.04 (\mathrm{sys})$ for the companion. The chemical compositions for the brown dwarf and host star agree within the $1.5\sigma$ level, supporting a scenario where CD-35 2722 B formed via gravitational instability. We do not find evidence for clouds on CD-35 2722 B despite it being a photometrically red mid-L dwarf and thus expected to be quite cloudy. We retrieve a temperature structure which is more isothermal than models and investigate its impact on our measurements, finding that constraining the temperature structure to self-consistent models does not significantly impact our retrieved chemical properties. Our observations highlight the need for data from complementary wavelength ranges to verify the presence of aerosols in likely cloudy L dwarfs.

astro-ph.EP↗

arXiv2025

Characterizing the Time Variability of 2M1207 A+b with JWST NIRSpec/PRISM

We present JWST NIRSpec/PRISM IFU time-resolved observations of 2M1207 A and b (TWA 27), a $\sim 10$ Myr binary system consisting of a $\sim 2500$ K sub-stellar primary hosting a $\sim 1300$ K companion. Our data provide 20 time-resolved spectra over an observation spanning 12.56 hours. We provide an empirical characterization for the spectra of both objects across time. For 2M1207 A, non-linear trend models are statistically favored within the ranges 0.6-2.3 $\mu$m and 3.8-5.3 $\mu$m. However, most of the periods constrained from sinusoidal models exceed the observing window, setting a lower limit of 12.56 hours. We find the data at H$\alpha$ and beyond 4.35 $\mu$m show a moderate time correlation, as well as a pair of light curves at 0.73-0.80 $\mu$m and 3.36-3.38 $\mu$m. For 2M1207 b, light curves integrated across 0.86-1.77 $\mu$m and 3.29-4.34 $\mu$m support linear trend models. Following the interpretation of Zhang et. al. (2025), we model the 2M1207 b data with two 1D atmospheric components, both with silicate and iron condensates. The model of time variability as changes to the cloud filling factor shows broad consistency with the variability amplitudes derived from our data. Our amplitudes, however, disagree with the models at $\approx$0.86-1 $\mu$m. While an additional model component such as rainout chemistry may be considered here, our analysis is limited by a low signal-to-noise ratio. Our results demonstrate the capability of JWST to simultaneously monitor the spectral variability of a planetary-mass companion and host at low contrast.

astro-ph.EP↗

arXiv2025

YSES 2b is a background star: Differential astrometric M-dwarf measurements in time

We wish to confirm the nature of YSES 2b, a purportedly faint companion of the young star YSES 2. We used on-sky observations from SPHERE and GRAVITY to measure the astrometric position of 2b with respect to the star YSES 2, and examined the competing hypotheses of (i) a bound substellar companion versus (ii) a distant unrelated background source with a non-zero proper motion. YSES 2b appears to be a late-type M-dwarf star over 2 kiloparsecs behind the star YSES 2. It has a transverse velocity of about 300 km/s and is located within one of the spiral arms of the Galaxy. The main discriminant was multiple epochs of GRAVITY astrometry that identified the sub-milliarcsecond parallactic motion of the star.

astro-ph.EP↗