arXiv subjects

Jason Wang

Publications and source records attributed to Jason Wang.

At least 37 records · Page 2Linked to original sources

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 $μ$m and 3.8-5.3 $μ$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$α$ and beyond 4.35 $μ$m show a moderate time correlation, as well as a pair of light curves at 0.73-0.80 $μ$m and 3.36-3.38 $μ$m. For 2M1207 b, light curves integrated across 0.86-1.77 $μ$m and 3.29-4.34 $μ$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 $μ$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↗

arXiv2025

Follow-Up Exploration of the TWA 7 Planet-Disk System with JWST NIRCam

The young M-star TWA 7 hosts a bright and near face-on debris disk, which has been imaged from the optical to the submillimeter. The disk displays multiple complex substructures such as three disk components, a large dust clump, and spiral arms, suggesting the presence of planets to actively sculpt these features. The evidence for planets in this disk was further strengthened with the recent detection of a point-source compatible with a Saturn-mass planet companion using JWST/MIRI at 11 $μ$m, at the location a planet was predicted to reside based on the disk morphology. In this paper, we present new observations of the TWA 7 system with JWST/NIRCam in the F200W and F444W filters. The disk is detected at both wavelengths and presents many of the same substructures as previously imaged, although we do not robustly detect the southern spiral arm. Furthermore, we detect two faint potential companions in the F444W filter at the 2-3$σ$ level. While one of these companions needs further followup to determine its nature, the other one coincides with the location of the planet candidate imaged with MIRI, providing further evidence that this source is a sub-Jupiter mass planet companion rather than a background galaxy. Such discoveries make TWA 7 only the second system, after $β$ Pictoris, in which a planet predicted by the debris disk morphology has been detected.

astro-ph.EP↗

arXiv2025

Differentiating Formation Models with New Dynamical Masses for the PDS 70 Protoplanets

Hot- and cold-start planet formation models predict differing luminosities for the young, bright planets that direct imaging surveys are most sensitive to. However, precise mass estimates are required to distinguish between these models observationally. The presence of two directly imaged planets, PDS 70 b and c, in the PDS 70 protoplanetary disk provides us a unique opportunity for dynamical mass measurements, since the masses for these planets are currently poorly constrained. Fitting orbital parameters to new astrometry of these planets, taken with VLTI/GRAVITY in the $K$~band, we find $2σ$ dynamical upper mass limits of 4.9 $M_{\rm Jup}$ for b and 13.6 $M_{\rm Jup}$ for c. Adding astrometry from the newly proposed planet candidate PDS 70 d into our model, we determine $2σ$ dynamical upper mass limits of 5.3 $M_{\rm Jup}$, 7.5 $M_{\rm Jup}$ and 2.2 $M_{\rm Jup}$ for b, c, and the candidate d respectively. However, $N$-body analysis of the orbits fit in this case suggest that the inclusion of $d$ makes the system unstable. Using the upper mass limits for b and c we rule out the coldest-start formation models for both planets, calculating minimum post-formation entropies ($S_i$) of 9.5 $k_{\rm B}/{\rm baryon}$ and 8.4 $k_{\rm B}/{\rm baryon}$ respectively. This places PDS 70 b and c on the growing list of directly-imaged planets inconsistent with cold-start formation.

astro-ph.EP↗

arXiv2025

Automated Unit Test Case Generation: A Systematic Literature Review

Software is omnipresent within all factors of society. It is thus important to ensure that software are well tested to mitigate bad user experiences as well as the potential for severe financial and human losses. Software testing is however expensive and absorbs valuable time and resources. As a result, the field of automated software testing has grown of interest to researchers in past decades. In our review of present and past research papers, we have identified an information gap in the areas of improvement for the Genetic Algorithm and Particle Swarm Optimisation. A gap in knowledge in the current challenges that face automated testing has also been identified. We therefore present this systematic literature review in an effort to consolidate existing knowledge in regards to the evolutionary approaches as well as their improvements and resulting limitations. These improvements include hybrid algorithm combinations as well as interoperability with mutation testing and neural networks. We will also explore the main test criterion that are used in these algorithms alongside the challenges currently faced in the field related to readability, mocking and more.

cs.SE↗

arXiv2025

Towards Interpretable Soft Prompts

Soft prompts have been popularized as a cheap and easy way to improve task-specific LLM performance beyond few-shot prompts. Despite their origin as an automated prompting method, however, soft prompts and other trainable prompts remain a black-box method with no immediately interpretable connections to prompting. We create a novel theoretical framework for evaluating the interpretability of trainable prompts based on two desiderata: faithfulness and scrutability. We find that existing methods do not naturally satisfy our proposed interpretability criterion. Instead, our framework inspires a new direction of trainable prompting methods that explicitly optimizes for interpretability. To this end, we formulate and test new interpretability-oriented objective functions for two state-of-the-art prompt tuners: Hard Prompts Made Easy (PEZ) and RLPrompt. Our experiments with GPT-2 demonstrate a fundamental trade-off between interpretability and the task-performance of the trainable prompt, explicating the hardness of the soft prompt interpretability problem and revealing odd behavior that arises when one optimizes for an interpretability proxy.

cs.LG↗

arXiv2025

HD 206893 B at High Spectral Resolution with the Keck Planet Imager and Characterizer (KPIC)

We present an atmospheric characterization and orbital analysis of HD 206893 B, an exceptionally red, L/T-transition substellar companion in a multiplanetary system, via Keck Planet Imager and Characterizer (KPIC) high-resolution (R $\sim$ 35,000) K-band spectroscopy. Using PHOENIX atmospheric models in a forward-model framework that fits the spectrum of the companion and diffracted starlight simultaneously, we detect HD 206893 B at $>8σ$ significance via cross-correlation in two epochs. We find an effective temperature for the companion of $1634^{+72}_{-38}$ K and a log(g) of $4.55^{+0.17}_{-0.22}$. Only accounting for statistical uncertainties, we measure the carbon-oxygen ratio (C/O) of this companion to be $0.57 \pm 0.02$, or near-solar while assuming solar metallicity. The C/O ratio we measure fits the tentative trend of $>4 M_{Jup}$ companions having near-solar C/O ratios while less massive companions have greater-than-solar C/O ratios. Using substellar evolution models, we find an age of $112^{+36}_{-22}$ Myr, a mass of $22.7^{+2.5}_{-1.7} M_{Jup}$, and a radius of $1.11 \pm 0.03 R_{Jup}$ for this companion. We also use KPIC radial velocity data to fit the orbit of HD 206893 B and analyze the orbital stability of this system. We find that the orbital stability is relatively independent of the mass of HD 206893 B, and favors an orbital configuration where B and its interior planetary companion, HD 206893 c, are co-planar. The measured C/O ratio coupled with the current architecture of the system cannot rule out a core accretion scenario, nor a disk fragmentation scenario regarding the formation pathway of HD 206893 B.

astro-ph.EP↗

arXiv2024

Understanding Warmup-Stable-Decay Learning Rates: A River Valley Loss Landscape Perspective

Training language models currently requires pre-determining a fixed compute budget because the typical cosine learning rate schedule depends on the total number of steps. In contrast, the Warmup-Stable-Decay (WSD) schedule uses a constant learning rate to produce a main branch of iterates that can in principle continue indefinitely without a pre-specified compute budget. Then, given any compute budget, one can branch out from the main branch at a proper time with a rapidly decaying learning rate to produce a strong model. Empirically, WSD generates a non-traditional loss curve: the loss remains elevated during the stable phase but sharply declines during the decay phase. Towards explaining this phenomenon, we conjecture that pretraining loss exhibits a river valley landscape, which resembles a deep valley with a river at its bottom. Under this assumption, we show that during the stable phase, the iterate undergoes large oscillations due to the high learning rate, yet it progresses swiftly along the river. During the decay phase, the rapidly dropping learning rate minimizes the iterate's oscillations, moving it closer to the river and revealing true optimization progress. Therefore, the sustained high learning rate phase and fast decaying phase are responsible for progress in the river and the mountain directions respectively, and are both critical. Our analysis predicts phenomenons consistent with empirical observations and shows that this landscape can emerge from pretraining on a simple bi-gram dataset. Inspired by the theory, we introduce WSD-S, a variant of WSD that reuses previous checkpoints' decay phases and keeps only one main branch, where we resume from a decayed checkpoint. WSD-S empirically outperforms WSD and Cyclic-Cosine in obtaining multiple language model checkpoints across various compute budgets in a single run for parameters scaling from 0.1B to 1.2B.

cs.LG↗

arXiv2024

The Roman coronagraph community participation program: observation planning

The Coronagraphic Instrument onboard the Nancy Grace Roman Space Telescope is an important stepping stone towards the characterization of habitable, rocky exoplanets. In a technology demonstration phase conducted during the first 18 months of the mission (expected to launch in late 2026), novel starlight suppression technology may enable direct imaging of a Jupiter analog in reflected light. Here we summarize the current activities of the Observation Planning working group formed as part of the Community Participation Program. This working group is responsible for target selection and observation planning of both science and calibration targets in the technology demonstration phase of the Roman Coronagraph. We will discuss the ongoing efforts to expand target and reference catalogs, and to model astrophysical targets (exoplanets and circumstellar disks) within the Coronagraph's expected sensitivity. We will also present preparatory observations of high priority targets.

astro-ph.IM↗

arXiv2024

Scalable Indoor Novel-View Synthesis using Drone-Captured 360 Imagery with 3D Gaussian Splatting

Scene reconstruction and novel-view synthesis for large, complex, multi-story, indoor scenes is a challenging and time-consuming task. Prior methods have utilized drones for data capture and radiance fields for scene reconstruction, both of which present certain challenges. First, in order to capture diverse viewpoints with the drone's front-facing camera, some approaches fly the drone in an unstable zig-zag fashion, which hinders drone-piloting and generates motion blur in the captured data. Secondly, most radiance field methods do not easily scale to arbitrarily large number of images. This paper proposes an efficient and scalable pipeline for indoor novel-view synthesis from drone-captured 360 videos using 3D Gaussian Splatting. 360 cameras capture a wide set of viewpoints, allowing for comprehensive scene capture under a simple straightforward drone trajectory. To scale our method to large scenes, we devise a divide-and-conquer strategy to automatically split the scene into smaller blocks that can be reconstructed individually and in parallel. We also propose a coarse-to-fine alignment strategy to seamlessly match these blocks together to compose the entire scene. Our experiments demonstrate marked improvement in both reconstruction quality, i.e. PSNR and SSIM, and computation time compared to prior approaches.

cs.CV↗

arXiv2024

Efficiently Searching for Close-in Companions around Young M Dwarfs using a Multi-year PSF Library

We present Super-RDI, a unique framework for the application of reference star differential imaging (RDI) to Keck/NIRC2 high-contrast imaging observations with the vortex coronagraph. Super-RDI combines frame selection and signal-to-noise ratio (S/N) optimization techniques with a large multi-year reference point spread function (PSF) library to achieve optimal PSF subtraction at small angular separations. We compile a $\sim$7000 frame reference PSF library based on a set of 288 new Keck/NIRC2 $L'$ sequences of 237 unique targets acquired between 2015 and 2019 as part of two planet-search programs, one focusing on nearby young M dwarfs and the other targeting members of the Taurus star-forming region. For our dataset, synthetic companion injection-recovery tests reveal that frame selection with the mean-squared error (MSE) metric combined with KLIP-based PSF subtraction using 1000-3000 frames and $<$500 principal components yields the highest average S/N for injected synthetic companions. We uniformly reduce targets in the young M-star survey with both Super-RDI and angular differential imaging (ADI). For the typical parallactic angle rotation of our dataset ($\sim$10$^\circ$), Super-RDI performs better than a widely used implementation of ADI at separations $\lesssim$0.4" ($\approx$5 $λ$/$D$) gaining an average of 0.25 mag in contrast at 0.25" and 0.4 mag in contrast at 0.15". This represents a performance improvement in separation space over RDI with single-night reference star observations ($\sim$100 frame PSF libraries) applied to a similar Keck/NIRC2 dataset in previous work. We recover two known brown dwarf companions and provide detection limits for 155 targets in the young M-star survey. Our results demonstrate that increasing the PSF library size with careful selection of reference frames can improve the performance of RDI with the Keck/NIRC2 vortex coronagraph in $L'$.

astro-ph.EP↗

arXiv2024

Pandora's White-Box: Precise Training Data Detection and Extraction in Large Language Models

In this paper we develop state-of-the-art privacy attacks against Large Language Models (LLMs), where an adversary with some access to the model tries to learn something about the underlying training data. Our headline results are new membership inference attacks (MIAs) against pretrained LLMs that perform hundreds of times better than baseline attacks, and a pipeline showing that over 50% (!) of the fine-tuning dataset can be extracted from a fine-tuned LLM in natural settings. We consider varying degrees of access to the underlying model, pretraining and fine-tuning data, and both MIAs and training data extraction. For pretraining data, we propose two new MIAs: a supervised neural network classifier that predicts training data membership on the basis of (dimensionality-reduced) model gradients, as well as a variant of this attack that only requires logit access to the model by leveraging recent model-stealing work on LLMs. To our knowledge this is the first MIA that explicitly incorporates model-stealing information. Both attacks outperform existing black-box baselines, and our supervised attack closes the gap between MIA attack success against LLMs and the strongest known attacks for other machine learning models. In fine-tuning, we find that a simple attack based on the ratio of the loss between the base and fine-tuned models is able to achieve near-perfect MIA performance; we then leverage our MIA to extract a large fraction of the fine-tuning dataset from fine-tuned Pythia and Llama models. Our code is available at github.com/safr-ai-lab/pandora-llm.

cs.CR↗

arXiv2024

Hardness of Learning Neural Networks under the Manifold Hypothesis

The manifold hypothesis presumes that high-dimensional data lies on or near a low-dimensional manifold. While the utility of encoding geometric structure has been demonstrated empirically, rigorous analysis of its impact on the learnability of neural networks is largely missing. Several recent results have established hardness results for learning feedforward and equivariant neural networks under i.i.d. Gaussian or uniform Boolean data distributions. In this paper, we investigate the hardness of learning under the manifold hypothesis. We ask which minimal assumptions on the curvature and regularity of the manifold, if any, render the learning problem efficiently learnable. We prove that learning is hard under input manifolds of bounded curvature by extending proofs of hardness in the SQ and cryptographic settings for Boolean data inputs to the geometric setting. On the other hand, we show that additional assumptions on the volume of the data manifold alleviate these fundamental limitations and guarantee learnability via a simple interpolation argument. Notable instances of this regime are manifolds which can be reliably reconstructed via manifold learning. Looking forward, we comment on and empirically explore intermediate regimes of manifolds, which have heterogeneous features commonly found in real world data.

cs.LG↗

arXiv2024

Octofitter: Fast, Flexible, and Accurate Orbit Modelling to Detect Exoplanets

As next-generation imaging instruments and interferometers search for planets closer to their stars, they must contend with increasing orbital motion and longer integration times. These compounding effects make it difficult to detect faint planets but also present an opportunity. Increased orbital motion makes it possible to move the search for planets into the orbital domain, where direct images can be freely combined with the radial velocity and proper motion anomaly, even without a confirmed detection in any single epoch. In this paper, we present a fast and differentiable multimethod orbit-modeling and planet detection code called Octofitter. This code is designed to be highly modular and allows users to easily adjust priors, change parameterizations, and specify arbitrary function relations between the parameters of one or more planets. Octofitter further supplies tools for examining model outputs including prior and posterior predictive checks and simulation-based calibration. We demonstrate the capabilities of Octofitter on real and simulated data from different instruments and methods, including HD 91312, simulated JWST/NIRISS aperture masking interferometry observations, radial velocity curves, and grids of images from the Gemini Planet Imager. We show that Octofitter can reliably recover faint planets in long sequences of images with arbitrary orbital motion. This publicly available tool will enable the broad application of multiepoch and multimethod exoplanet detection, which could improve how future targeted ground- and space-based surveys are performed. Finally, its rapid convergence makes it a useful addition to the existing ecosystem of tools for modeling the orbits of directly imaged planets.

astro-ph.EP↗

arXiv2023

Closed Drafting as a Case Study for First-Principle Interpretability, Memory, and Generalizability in Deep Reinforcement Learning

Closed drafting or "pick and pass" is a popular game mechanic where each round players select a card or other playable element from their hand and pass the rest to the next player. In this paper, we establish first-principle methods for studying the interpretability, generalizability, and memory of Deep Q-Network (DQN) models playing closed drafting games. In particular, we use a popular family of closed drafting games called "Sushi Go Party", in which we achieve state-of-the-art performance. We fit decision rules to interpret the decision-making strategy of trained DRL agents by comparing them to the ranking preferences of different types of human players. As Sushi Go Party can be expressed as a set of closely-related games based on the set of cards in play, we quantify the generalizability of DRL models trained on various sets of cards, establishing a method to benchmark agent performance as a function of environment unfamiliarity. Using the explicitly calculable memory of other player's hands in closed drafting games, we create measures of the ability of DRL models to learn memory.

cs.LG↗

arXiv2023

MoPe: Model Perturbation-based Privacy Attacks on Language Models

Recent work has shown that Large Language Models (LLMs) can unintentionally leak sensitive information present in their training data. In this paper, we present Model Perturbations (MoPe), a new method to identify with high confidence if a given text is in the training data of a pre-trained language model, given white-box access to the models parameters. MoPe adds noise to the model in parameter space and measures the drop in log-likelihood at a given point $x$, a statistic we show approximates the trace of the Hessian matrix with respect to model parameters. Across language models ranging from $70$M to $12$B parameters, we show that MoPe is more effective than existing loss-based attacks and recently proposed perturbation-based methods. We also examine the role of training point order and model size in attack success, and empirically demonstrate that MoPe accurately approximate the trace of the Hessian in practice. Our results show that the loss of a point alone is insufficient to determine extractability -- there are training points we can recover using our method that have average loss. This casts some doubt on prior works that use the loss of a point as evidence of memorization or unlearning.

cs.LG↗

arXiv2023

The development of HISPEC for Keck and MODHIS for TMT: science cases and predicted sensitivities

HISPEC is a new, high-resolution near-infrared spectrograph being designed for the W.M. Keck II telescope. By offering single-shot, R=100,000 between 0.98 - 2.5 um, HISPEC will enable spectroscopy of transiting and non-transiting exoplanets in close orbits, direct high-contrast detection and spectroscopy of spatially separated substellar companions, and exoplanet dynamical mass and orbit measurements using precision radial velocity monitoring calibrated with a suite of state-of-the-art absolute and relative wavelength references. MODHIS is the counterpart to HISPEC for the Thirty Meter Telescope and is being developed in parallel with similar scientific goals. In this proceeding, we provide a brief overview of the current design of both instruments, and the requirements for the two spectrographs as guided by the scientific goals for each. We then outline the current science case for HISPEC and MODHIS, with focuses on the science enabled for exoplanet discovery and characterization. We also provide updated sensitivity curves for both instruments, in terms of both signal-to-noise ratio and predicted radial velocity precision.

astro-ph.IM↗

arXiv2023

Vortex Fiber Nulling for Exoplanet Observations: Implementation and First Light

Vortex fiber nulling (VFN) is a single-aperture interferometric technique for detecting and characterizing exoplanets separated from their host star by less than a diffracted beam width. VFN uses a vortex mask and single mode fiber to selectively reject starlight while coupling off-axis planet light with a simple optical design that can be readily implemented on existing direct imaging instruments that can feed light to an optical fiber. With its axially symmetric coupling region peaking within the inner working angle of conventional coronagraphs, VFN is more efficient at detecting new companions at small separations than conventional direct imaging, thereby increasing the yield of on-going exoplanet search campaigns. We deployed a VFN mode operating in K band ($2.0{-}2.5~μ$m) on the Keck Planet Imager and Characterizer (KPIC) instrument at the Keck II Telescope. In this paper we present the instrument design of this first on-sky demonstration of VFN and the results from on-sky commissioning, including planet and star throughput measurements and predicted flux-ratio detection limits for close-in companions. The instrument performance is shown to be sufficient for detecting a companion $10^3$ times fainter than a $5^{\mathrm{th}}$ magnitude host star in 1 hour at a separation of 50 mas (1.1$λ/D$). This makes the instrument capable of efficiently detecting substellar companions around young stars. We also discuss several routes for improvement that will reduce the required integration time for a detection by a factor ${>}$3.

astro-ph.IM↗