skip to main content

Attention:

The NSF Public Access Repository (NSF-PAR) system and access will be unavailable from 11:00 PM ET on Thursday, May 23 until 2:00 AM ET on Friday, May 24 due to maintenance. We apologize for the inconvenience.


Search for: All records

Creators/Authors contains: "Zhang, X"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Large Language Models (LLMs) have achieved remarkable success across a wide array of tasks. Due to the impressive planning and reasoning abilities of LLMs, they have been used as autonomous agents to do many tasks automatically. Recently, based on the development of using one LLM as a single planning or decision-making agent, LLM-based multi-agent systems have achieved considerable progress in complex problem-solving and world simulation. To provide the community with an overview of this dynamic field, we present this survey to offer an in-depth discussion on the essential aspects of multi-agent systems based on LLMs, as well as the challenges. Our goal is for readers to gain substantial insights on the following questions: What domains and environments do LLM-based multi-agents simulate? How are these agents profiled and how do they communicate? What mechanisms contribute to the growth of agents' capacities? For those interested in delving into this field of study, we also summarize the commonly used datasets or benchmarks for them to have convenient access. To keep researchers updated on the latest studies, we maintain an open-source GitHub repository, dedicated to outlining the research on LLM-based multi-agent systems. 
    more » « less
    Free, publicly-accessible full text available August 3, 2025
  2. Predicting chemical reaction yields is pivotal for efficient chemical synthesis, an area that focuses on the creation of novel compounds for diverse uses. Yield prediction demands accurate representations of reactions for forecasting practical transformation rates. Yet, the uncertainty issues broadcasting in real-world situations prohibit current models to excel in this task owing to the high sensitivity of yield activities and the uncertainty in yield measurements. Existing models often utilize single-modal feature representations, such as molecular fingerprints, SMILES sequences, or molecular graphs, which is not sufficient to capture the complex interactions and dynamic behavior of molecules in reactions. In this paper, we present an advanced Uncertainty-Aware Multimodal model (UAM) to tackle these challenges. Our approach seamlessly integrates data sources from multiple modalities by encompassing sequence representations, molecular graphs, and expert-defined chemical reaction features for a comprehensive representation of reactions. Additionally, we address both the model and data-based uncertainty, refining the model’s predictive capability. Extensive experiments on three datasets, including two high throughput experiment (HTE) datasets and one chemist-constructed Amide coupling reaction dataset, demonstrate that UAM outperforms the stateof-the-art methods. The code and used datasets are available at https://github.com/jychen229/Multimodal-reaction-yieldprediction. 
    more » « less
    Free, publicly-accessible full text available February 27, 2025
  3. Abstract

    In planetary radiation belts, the Kennel‐Petschek flux limit is expected to set an upper limit on trapped electron fluxes at 80–600 keV in the presence of efficient electron loss through pitch‐angle diffusion by whistler‐mode chorus waves generated around the magnetic equator by the same 80–600 keV electron population. Comparisons with maximum measured fluxes have been relatively successful, but several key assumptions of the Kennel‐Petschek model have not been experimentally tested. The Kennel‐Petschek model notably assumes an exponential growth of chorus waves as the trapped electron flux increases, and a fixed maximum wave power gain of about 3. Here, we describe a method for inferring the near‐equatorial wave power gain using only measurements of trapped, precipitating, and backscattered electron fluxes at low altitude. Next, we make use of Electron Losses and Fields Investigation (ELFIN) CubeSats measurements of such electron fluxes during two moderate geomagnetic storms with sustained electron injections to infer the corresponding chorus wave power gains as a function of time, energy, and equatorial trapped electron flux. We show that wave power increases exponentially with trapped flux, with a wave power gain roughly proportional to the theoretical linear convective gain, and that the maximum inferred gain near the upper flux limit is roughly 10, with a factor of 2 uncertainty. Therefore, two key theoretical underpinnings of the Kennel‐Petschek model are borne out by the present results, although the strong inferred gains should correspond to higher flux limits than in traditional estimates.

     
    more » « less
    Free, publicly-accessible full text available February 1, 2025
  4. Free, publicly-accessible full text available March 1, 2025
  5. Abstract

    Precipitation of relativistic electrons into the Earth's atmosphere regulates the outer radiation belt fluxes and contributes to magnetosphere‐atmosphere coupling. One of the main drivers of such precipitation is electron scattering by whistler‐mode waves. Such waves typically originate at the equator, where they can resonate with and scatter sub‐relativistic (tens to a few hundred keV) electrons. However, they can occasionally propagate far away from the equator along field lines, reaching middle latitudes, where they can resonate with and scatter relativistic (>500 keV) electrons. Such a propagation is typical for the dayside, but statistically has not been found on the nightside where the waves are quickly damped along their propagation due to Landau damping. Here we explore two events of relativistic electron precipitation from low‐altitude observations on the nightside. Combining measurements of whistler‐mode waves from ground observatories, relativistic electron precipitation from low‐altitude satellites, total electron content maps from GPS receivers, and magnetic field and electron flux from equatorial satellites, we show wave ducting by plasma density gradients is the possible channel that allows the waves to reach middle latitudes and scatter relativistic electrons. We suggest that both whistler‐mode wave generation and ducting can be driven by equatorial mesoscale (with spatial scales of about one Earth radius) transient structures during nightside injections. We also compare these nightside events with observations of ducted waves and relativistic electron precipitation at the dayside, where wave generation and ducting are driven by ultra‐low‐frequency waves. This study demonstrates the potential importance of mesoscale transients in relativistic electron precipitation, but does not however unequivocally establish that ducted whistler‐mode waves are the primary cause of the observed electron precipitation.

     
    more » « less
    Free, publicly-accessible full text available February 1, 2025
  6. Free, publicly-accessible full text available January 1, 2025
  7. Free, publicly-accessible full text available January 1, 2025
  8. Large Language Models (LLMs) with strong abilities in natural language processing tasks have emerged and have been applied in various kinds of areas such as science, finance and software engineering. However, the capability of LLMs to advance the field of chemistry remains unclear. In this paper, rather than pursuing state-of-the-art performance, we aim to evaluate capabilities of LLMs in a wide range of tasks across the chemistry domain. We identify three key chemistryrelated capabilities including understanding, reasoning and explaining to explore in LLMs and establish a benchmark containing eight chemistry tasks. Our analysis draws on widely recognized datasets facilitating a broad exploration of the capacities of LLMs within the context of practical chemistry. Five LLMs (GPT-4, GPT-3.5, Davinci-003, Llama and Galactica) are evaluated for each chemistry task in zero-shot and few-shot in-context learning settings with carefully selected demonstration examples and specially crafted prompts. Our investigation found that GPT-4 outperformed other models and LLMs exhibit different competitive levels in eight chemistry tasks. In addition to the key findings from the comprehensive benchmark analysis, our work provides insights into the limitation of current LLMs and the impact of in-context learning settings on LLMs’ performance across various chemistry tasks. The code and datasets used in this study are available at https://github.com/ChemFoundationModels/ChemLLMBench. 
    more » « less
    Free, publicly-accessible full text available December 16, 2024
  9. Free, publicly-accessible full text available November 1, 2024
  10. Inference of joule-class THz radiation sources from microchannel targets driven with hundreds of joule, picosecond lasers is reported. THz sources of this magnitude are useful for nonlinear pumping of matter and for charged-particle acceleration and manipulation. Microchannel targets demonstrate increased laser–THz conversion efficiency compared to planar foil targets, with laser energy to THz energy conversion up to ∼0.9% in the best cases.

     
    more » « less