arXiv subjects

Peter Chang

Publications and source records attributed to Peter Chang.

17 recordsLinked to original sources

arXiv2026

Integrated ytterbium gain for visible-near-infrared photonics

Rare-earth gain media form the foundation of modern optical communications, emerging quantum hardware, and ultrafast optics. While chip-scale integration can enable fiber-like, and potentially beyond-fiber, functionality with unprecedented scalability, development in the visible and near-infrared remains in its early stages. Here, we demonstrate ytterbium-based optical gain integrated into an aluminum oxide photonic platform, achieving both single-mode lasing and optical amplification in the near-infrared regime. This platform delivers optical amplification with output powers exceeding 0.5 W, an optical-to-optical conversion efficiency above 70%, and a noise figure of 3.3 dB, approaching the quantum limit for phase-insensitive amplification. Furthermore, we achieve femtosecond pulse amplification to a record peak power of 14 kW, enabling supercontinuum generation with visible dispersive waves extending from 780 to 476 nm in conjunction with nonlinear photonic devices. This platform is compatible with heterogeneous integration into standard photonic circuits, laying the foundation for scalable visible-near-infrared photonic systems, including coherent laser arrays, mode-locked lasers, optical clocks, and microwave oscillators.

physics.optics↗

arXiv2025

Nonparametric Bayesian Multi-Treatment Mixture Cure Survival Model with Application in Pediatric Oncology

Heterogeneous treatment effect estimation is critical in oncology, particularly in multi-arm trials with overlapping therapeutic components and long-term survivors. These shared mechanisms pose a central challenge to identifying causal effects in precision medicine. We propose a novel covariate-dependent nonparametric Bayesian multi-treatment cure survival model that jointly accounts for common structures among treatments and cure fractions. Through latent link functions, our model leverages sharing among treatments through a flexible modeling approach, enabling individualized survival inference. We adopt a Bayesian route for inference and implement an efficient MCMC algorithm for approximating the posterior. Simulation studies demonstrate the method's robustness and superiority in various specification scenarios. Finally, application to the AALL0434 trial reveals clinically meaningful differences in survival across methotrexate-based regimens and their associations with different covariates, underscoring its practical utility for learning treatment effects in real-world pediatric oncology data.

stat.ME↗

arXiv2025

Magentic-UI: Towards Human-in-the-loop Agentic Systems

AI agents powered by large language models are increasingly capable of autonomously completing complex, multi-step tasks using external tools. Yet, they still fall short of human-level performance in most domains including computer use, software development, and research. Their growing autonomy and ability to interact with the outside world, also introduces safety and security risks including potentially misaligned actions and adversarial manipulation. We argue that human-in-the-loop agentic systems offer a promising path forward, combining human oversight and control with AI efficiency to unlock productivity from imperfect systems. We introduce Magentic-UI, an open-source web interface for developing and studying human-agent interaction. Built on a flexible multi-agent architecture, Magentic-UI supports web browsing, code execution, and file manipulation, and can be extended with diverse tools via Model Context Protocol (MCP). Moreover, Magentic-UI presents six interaction mechanisms for enabling effective, low-cost human involvement: co-planning, co-tasking, multi-tasking, action guards, and long-term memory. We evaluate Magentic-UI across four dimensions: autonomous task completion on agentic benchmarks, simulated user testing of its interaction capabilities, qualitative studies with real users, and targeted safety assessments. Our findings highlight Magentic-UI's potential to advance safe and efficient human-agent collaboration.

cs.AI↗

arXiv2025

Robust Perspective Correction for Real-World Crack Evolution Tracking in Image-Based Structural Health Monitoring

Accurate image alignment is essential for monitoring crack evolution in structural health monitoring (SHM), particularly under real-world conditions involving perspective distortion, occlusion, and low contrast. However, traditional feature detectors such as SIFT and SURF, which rely on Gaussian-based scale spaces, tend to suppress high-frequency edges, making them unsuitable for thin crack localization. Lightweight binary alternatives like ORB and BRISK, while computationally efficient, often suffer from poor keypoint repeatability on textured or shadowed surfaces. This study presents a physics-informed alignment framework that adapts the open KAZE architecture to SHM-specific challenges. By utilizing nonlinear anisotropic diffusion to construct a crack-preserving scale space, and integrating RANSAC-based homography estimation, the framework enables accurate geometric correction without the need for training, parameter tuning, or prior calibration. The method is validated on time-lapse images of masonry and concrete acquired via handheld smartphone under varied field conditions, including shadow interference, cropping, oblique viewing angles, and surface clutter. Compared to classical detectors, the proposed framework reduces crack area and spine length errors by up to 70 percent and 90 percent, respectively, while maintaining sub-5 percent alignment error in key metrics. Unsupervised, interpretable, and computationally lightweight, this approach supports scalable deployment via UAVs and mobile platforms. By tailoring nonlinear scale-space modeling to SHM image alignment, this work offers a robust and physically grounded alternative to conventional techniques for tracking real-world crack evolution.

cs.CV↗

arXiv2024

Magentic-One: A Generalist Multi-Agent System for Solving Complex Tasks

Modern AI agents, driven by advances in large foundation models, promise to enhance our productivity and transform our lives by augmenting our knowledge and capabilities. To achieve this vision, AI agents must effectively plan, perform multi-step reasoning and actions, respond to novel observations, and recover from errors, to successfully complete complex tasks across a wide range of scenarios. In this work, we introduce Magentic-One, a high-performing open-source agentic system for solving such tasks. Magentic-One uses a multi-agent architecture where a lead agent, the Orchestrator, plans, tracks progress, and re-plans to recover from errors. Throughout task execution, the Orchestrator directs other specialized agents to perform tasks as needed, such as operating a web browser, navigating local files, or writing and executing Python code. We show that Magentic-One achieves statistically competitive performance to the state-of-the-art on three diverse and challenging agentic benchmarks: GAIA, AssistantBench, and WebArena. Magentic-One achieves these results without modification to core agent capabilities or to how they collaborate, demonstrating progress towards generalist agentic systems. Moreover, Magentic-One's modular design allows agents to be added or removed from the team without additional prompt tuning or training, easing development and making it extensible to future scenarios. We provide an open-source implementation of Magentic-One, and we include AutoGenBench, a standalone tool for agentic evaluation. AutoGenBench provides built-in controls for repetition and isolation to run agentic benchmarks in a rigorous and contained manner -- which is important when agents' actions have side-effects. Magentic-One, AutoGenBench and detailed empirical performance evaluations of Magentic-One, including ablations and error analysis are available at https://aka.ms/magentic-one

cs.AI↗

arXiv2024

Dual-Comb Photothermal Microscopy

We introduce a new parallelized approach to photothermal microscopy that utilizes mid-infrared dual-comb lasers, enabling simultaneous measurements at hundreds of wavelengths. This technology allows for high-speed, label-free chemical identification with super-resolution infrared imaging, overcoming the limitations of slow, single-wavelength-laser methods.

physics.optics↗

arXiv2024

Mid-Infrared Hyperspectral Microscopy with Broadband 1-GHz Dual Frequency Combs

Mid-infrared microscopy is an important tool for biological analyses, allowing a direct probe of molecular bonds in their low energy landscape. In addition to the label-free extraction of spectroscopic information, the application of broadband sources can provide a third dimension of chemical specificity. However, to enable widespread deployment, mid-infrared microscopy platforms need to be compact and robust while offering high speed, broad bandwidth and high signal-to-noise ratio (SNR). In this study, we experimentally showcase the integration of a broadband, high-repetition-rate dual-comb spectrometer (DCS) in the mid-infrared range with a scanning microscope. We employ a set of 1-GHz mid-infrared frequency combs, demonstrating their capability for high-speed and broadband hyperspectral imaging of polymers and ovarian tissue. The system covers 1000 $\mathrm{cm^{-1}}$ at $\mathrm{\nu_c=2941 \; cm^{-1}}$ with 12.86 kHz spectra acquisition rate and 5 $\mathrm{\mu m}$ spatial resolution. Taken together, our experiments and analysis elucidate the trade-off between bandwidth and speed in DCS as it relates to microscopy. This provides a roadmap for the future advancement and application of high-repetition-rate DCS hyperspectral imaging.

physics.optics↗

arXiv2024

Bayesian Online Natural Gradient (BONG)

We propose a novel approach to sequential Bayesian inference based on variational Bayes (VB). The key insight is that, in the online setting, we do not need to add the KL term to regularize to the prior (which comes from the posterior at the previous timestep); instead we can optimize just the expected log-likelihood, performing a single step of natural gradient descent starting at the prior predictive. We prove this method recovers exact Bayesian inference if the model is conjugate. We also show how to compute an efficient deterministic approximation to the VB objective, as well as our simplified objective, when the variational distribution is Gaussian or a sub-family, including the case of a diagonal plus low-rank precision matrix. We show empirically that our method outperforms other online VB methods in the non-conjugate setting, such as online learning for neural networks, especially when controlling for computational costs.

stat.ML↗

arXiv2024

Dual-comb correlation spectroscopy of thermal light

The detection of light of thermal origin is the principal means by which humanity has learned about our world and the cosmos. In optical astronomy, in particular, direct detection of thermal photons and the resolution of their spectra have enabled discoveries of the broadest scope and impact. Such measurements, however, do not capture the phase of the thermal fields--a parameter that has proven crucial to transformative techniques in radio astronomy such as synthetic aperture imaging. Over the last 25 years, tremendous progress has occurred in laser science, notably in the phase-sensitive, broad bandwidth, high resolution, and traceable spectroscopy enabled by the optical frequency comb. In this work, we directly connect the fields of frequency comb laser spectroscopy and passive optical sensing as applied to astronomy, remote sensing, and atmospheric science. We provide fundamental sensitivity analysis of dual-comb correlation spectroscopy (DCCS), whereby broadband thermal light is measured via interferometry with two optical frequency combs. We define and experimentally verify the sensitivity scaling of DCCS at black body temperatures relevant for astrophysical observations. Moreover, we provide comparison with direct detection techniques and more conventional laser heterodyne radiometry. Our work provides the foundation for future exploration of comb-based broadband synthetic aperture hyperspectral imaging across the infrared and optical spectrum.

physics.optics↗

arXiv2024

Individualized Multi-Treatment Response Curves Estimation using RBF-net with Shared Neurons

Heterogeneous treatment effect estimation is an important problem in precision medicine. Specific interests lie in identifying the differential effect of different treatments based on some external covariates. We propose a novel non-parametric treatment effect estimation method in a multi-treatment setting. Our non-parametric modeling of the response curves relies on radial basis function (RBF)-nets with shared hidden neurons. Our model thus facilitates modeling commonality among the treatment outcomes. The estimation and inference schemes are developed under a Bayesian framework using thresholded best linear projections and implemented via an efficient Markov chain Monte Carlo algorithm, appropriately accommodating uncertainty in all aspects of the analysis. The numerical performance of the method is demonstrated through simulation experiments. Applying our proposed method to MIMIC data, we obtain several interesting findings related to the impact of different treatment strategies on the length of ICU stay and 12-hour SOFA score for sepsis patients who are home-discharged.

stat.ME↗

arXiv2023

A Multi-Harmonic NIR-UV Dual-Comb Spectrometer

Dual-comb spectroscopy in the ultraviolet (UV) and visible would enable broad bandwidth electronic spectroscopy with unprecedented frequency resolution. However, there are significant challenges in generation, detection and processing of dual-comb data that have restricted its progress in this spectral region. In this work, we leverage robust 1550 nm few-cycle pulses to generate frequency combs in the UV-visible. We couple this source to a wavelength multiplexed dual-comb spectrometer and simultaneously retrieve 100 MHz comb-mode-resolved spectra over three distinct harmonics spanning 380-800 nm. The experiments highlight the path to continuous dual-comb coverage spanning 200-750 nm, offering extensive access to electronic transitions in atoms, molecules, and solids.

physics.optics↗

arXiv2023

Privacy Issues in Large Language Models: A Survey

This is the first survey of the active area of AI research that focuses on privacy issues in Large Language Models (LLMs). Specifically, we focus on work that red-teams models to highlight privacy risks, attempts to build privacy into the training or inference process, enables efficient data deletion from trained models to comply with existing privacy regulations, and tries to mitigate copyright issues. Our focus is on summarizing technical research that develops algorithms, proves theorems, and runs empirical evaluations. While there is an extensive body of legal and policy work addressing these challenges from a different angle, that is not the focus of our survey. Nevertheless, these works, along with recent legal developments do inform how these technical problems are formalized, and so we discuss them briefly in Section 1. While we have made our best effort to include all the relevant work, due to the fast moving nature of this research we may have missed some recent work. If we have missed some of your work please contact us, as we will attempt to keep this survey relatively up to date. We are maintaining a repository with the list of papers covered in this survey and any relevant code that was publicly available at https://github.com/safr-ml-lab/survey-llm.

cs.AI↗

arXiv2023

Complete reactants-to-products observation of a gas-phase chemical reaction with broad, fast mid-infrared frequency combs

Molecular diagnostics are a primary tool of modern chemistry, enabling researchers to map chemical reaction pathways and rates to better design and control chemical systems. Many chemical reactions are complex and fast, and existing diagnostic approaches provide incomplete information. For example, mass spectrometry is optimized to gather snapshots of the presence of many chemical species, while conventional laser spectroscopy can quantify a single chemical species through time. Here we optimize for multiple objectives by introducing a high-speed and broadband, mid-infrared dual frequency comb absorption spectrometer. The optical bandwidth of >1000 cm-1 covers absorption fingerprints of many species with spectral resolution <0.03 cm-1 to accurately discern their absolute quantities. Key to this advance are 1 GHz pulse repetition rate frequency combs covering the 3-5 um region that enable microsecond tracking of fast chemical process dynamics. We demonstrate this system to quantify the abundances and temperatures of each species in the complete reactants-to-products breakdown of 1,3,5-trioxane, which exhibits a formaldehyde decomposition pathway that is critical to modern low temperature combustion systems. By maximizing the number of observed species and improving the accuracy of temperature and concentration measurements, this spectrometer advances understanding of chemical reaction pathways and rates and opens the door for novel developments such as combining high-speed chemistry with machine learning.

physics.chem-ph↗

arXiv2022

1-GHz mid-infrared frequency comb spanning 3 to 13 {\mu}m

Mid-infrared (MIR) spectrometers are invaluable tools for molecular fingerprinting and hyper-spectral imaging. Among the available spectroscopic approaches, GHz MIR dual-comb absorption spectrometers have the potential to simultaneously combine the high-speed, high spectral resolution, and broad optical bandwidth needed to accurately study complex, transient events in chemistry, combustion, and microscopy. However, such a spectrometer has not yet been demonstrated due to the lack of GHz MIR frequency combs with broad and full spectral coverage. Here, we introduce the first broadband MIR frequency comb laser platform at 1 GHz repetition rate that achieves spectral coverage from 3 to 13 {\mu}m. This frequency comb is based on a commercially available 1.56 {\mu}m mode-locked laser, robust all-fiber Er amplifiers and intra-pulse difference frequency generation (IP-DFG) of few-cycle pulses in \c{hi}(2) nonlinear crystals. When used in a dual comb spectroscopy (DCS) configuration, this source will simultaneously enable measurements with {\mu}s time resolution, 1 GHz (0.03 cm-1) spectral point spacing and a full bandwidth of >5 THz (>166 cm-1) anywhere within the MIR atmospheric windows. This represents a unique spectroscopic resource for characterizing fast and non-repetitive events that are currently inaccessible with other sources.

physics.optics↗

arXiv2009

Analytic torsion on spherical factors and tessellations

The analytic torsion is computed on fixed-point free and non fixed-point free factors (tessellations) of the three--sphere. We repeat the standard computation on spherical space forms (Clifford-Klein spaces) by an improved technique. The transformation to a simpler form of the spectral expression of the torsion on spherical factors effected by Ray is shown to be more general than his derivation implies. It effectively allows the eigenvalues to be considered as squares of integers, and applies also to trivial twistings. The analytic torsions compute to algebraic numbers, as expected. In the case of icosahedral space, the quaternion twisting gives a torsion proportional to the fundamental unit of Q(5^(1/2)). As well as a direct calculation, the torsions are obtained from the lens space values by a character inducing procedure.On tessellations, terms occur due to edge conical singularities.

math.DG↗

arXiv1992

Vacuum energy on orbifold factors of spheres

The vacuum energy is calculated for a free, conformally-coupled scalar field on the orbifold space-time \R$\times §^2/Γ$ where $Γ$ is a finite subgroup of O(3) acting with fixed points. The energy vanishes when $Γ$ is composed of pure rotations but not otherwise. It is shown on general grounds that the same conclusion holds for all even-dimensional factored spheres and the vacuum energies are given as generalised Bernoulli functions (i.e. Todd polynomials). The relevant $ζ$- functions are analysed in some detail and several identities are incidentally derived. The general discussion is presented in terms of finite reflection groups.

hep-th↗

arXiv1992

Polyhedral Cosmic Strings

Quantum field theory is discussed in Möbius corner kaleidoscopes using the method of images. The vacuum average of the stress-energy tensor of a free field is derived and is shown to be a simple sum of straight cosmic string expressions, the strings running along the edges of the corners. It does not seem possible to set up a spin-half theory easily.

hep-th↗