arXiv subjects

Yuqi Cao

Publications and source records attributed to Yuqi Cao.

5 recordsLinked to original sources

arXiv2026

ToSCA: Leveraging Hierarchical Reinforcement Learning on Temporal and Strategic Abstractions of Conversational Agents

Humans naturally exhibit multiple forms of abstraction in reasoning and interaction, including temporal abstraction across decision timescales and strategic abstraction over communicative intents. Inspired by these complementary abstractions, we propose a two-level hierarchical reinforcement learning (HRL) framework for conversational agents that bridges the gap between existing token-level and utterance-level RL methods. Built upon a two-level Markov decision process (MDP), our framework conditions token-level response generation on utterance-level actions represented by explicit textual strategies. Based on theoretical analysis and efficiency considerations, we employ DQN to optimize the high-level Q-network and PPO to train the low-level actor-critic. To further alleviate reward sparsity and facilitate convergence, we introduce a dual-granularity reward mechanism that combines the utterance-level satisfaction score with token-level intrinsic self-consistency and a KL-divergence penalty. Experiments on both daily-life and emotional support conversations demonstrate that our method consistently outperforms a wide range of baselines in both strategy determination and response quality. Our implementation is available at https://github.com/AaronJi/ToSCA.

cs.CL↗

arXiv2026

SimART: A Unified and Open Real-world Multimodal Simulation Platform for 6G Integrated Sensing and Communication

Research on sixth-generation (6G) integrated sensing and communication (ISAC) increasingly depends on multimodal datasets. These datasets need to jointly characterize wireless propagation, onboard sensing, and platform mobility. Existing tools cover only part of these aspects. Robotics simulators model physics and perception but not site-specific channels, while ray tracing and link level tools lack vehicle dynamics and onboard sensors. Combining them manually leads to workflows that are fragile and hard to reproduce. Rather than introducing another standalone simulator, this article presents SimART. It integrates mature robotics, ray tracing, and wireless evaluation engines into a single reproducible pipeline. The key idea is a robot operating system (ROS) backbone that both synchronizes and organizes all multimodal streams. A shared clock, a common coordinate frame, and timestamped messages keep the streams aligned in time and space, and a single rosbag recording captures the full session into one reproducible file. This design decouples the sensing front end from the wireless back end, so that any ROS-compatible simulator can be plugged in while reusing the same back end across aerial, ground, indoor, and maritime ISAC settings. On top of this backbone, SimART contributes a scene construction pipeline that converts both OpenStreetMap extracts and user-defined layouts into spatially aligned visual and electromagnetic assets, and a channel knowledge map (CKM) generator that aggregates ray tracing and system level outputs into spatial priors for ISAC algorithms. A case study on vision and position aided beam prediction demonstrates the utility of the platform. The code is publicly available at https://github.com/guchuanv-alt/SimART.

eess.SP↗

arXiv2025

Supercurrent Growth in Nonequilibrium Superconductors

In ultrafast experiments on superconductors, a pump laser pulse often heats up the electronic system and suppresses the density of superfluid electrons. Subsequently, the electrons undergo a cooling process because of electron-phonon thermalization so that the superfluid density recovers in time. We study the nonequilibrium electromagnetic response of the system in this cooling process. We show that if a supercurrent is initiated by a probe electric field pulse, an intriguing phenomenon of `supercurrent growth' occurs, meaning that the net current grows in time with the increasing superfluid density. Using the Boltzmann kinetic equation, we uncover its microscopic origin as the momentum-relaxing scattering of Bogoliubov quasiparticles by impurities and phonons, in stark contrast to the widely accepted intuition that impurities always attenuate currents. We further show that supercurrent growth has important experimental manifestations, including the ultrafast Meissner effect and an optical reflectivity exceeding unity.

cond-mat.supr-con↗

arXiv2020

Flowing cryogenic liquid target for terahertz wave generation

Terahertz wave emission from condensed matter excited by intense laser pulses not only reflects the details in laser-matter interaction but also offers bright terahertz wave sources. Flowing liquid targets possess the advantage of providing a fresh area for each laser pulse. To demonstrate a debris-free target under laser excitation, we investigate the use of liquid nitrogen as a target. By creating a flowing liquid nitrogen line in the ambient environment, we successfully observe broadband terahertz wave emission under short pulse excitation. Our cryogenic line is able to sustain the excitation of a high-repetition-rate (1 kHz) laser. The terahertz peak field emitted from liquid nitrogen is comparable to that from liquid water, yet a broader bandwidth is observed. This demonstration prompts new opportunities in choosing potential materials for studying terahertz wave generation process and in understanding laser-induced ionization in different liquids.

physics.optics↗

arXiv2020

Broadband Terahertz Wave Emission from Liquid Metal

Metals have been studied as terahertz sources for decades. Recent research has shown the potential of metals in generating extremely high THz pulse energy excited by intense laser pulses. To avoid the metal surface debris caused by laser pulses, here, we report the results of the broadband terahertz wave emission from a flowing liquid metal line excited by sub-picosecond laser pulses. The THz signal emitted from the liquid gallium line shows stronger field with broader bandwidth comparing with the signal from water under the identical optical excitation conditions. Our preliminary study suggests that the liquid metals have the potential to serve as efficient and powerful THz sources for the intense lasers with a high repetition rate.

physics.optics↗