Multi-agent reinforcement learning(MARL) has long been a significant research topic in both machine learning and control systems. Recent development of(single-agent) deep reinforcement learning has created a resurgenc...Multi-agent reinforcement learning(MARL) has long been a significant research topic in both machine learning and control systems. Recent development of(single-agent) deep reinforcement learning has created a resurgence of interest in developing new MARL algorithms, especially those founded on theoretical analysis. In this paper, we review recent advances on a sub-area of this topic: decentralized MARL with networked agents.In this scenario, multiple agents perform sequential decision-making in a common environment, and without the coordination of any central controller, while being allowed to exchange information with their neighbors over a communication network. Such a setting finds broad applications in the control and operation of robots, unmanned vehicles, mobile sensor networks, and the smart grid. This review covers several of our research endeavors in this direction, as well as progress made by other researchers along the line. We hope that this review promotes additional research efforts in this exciting yet challenging area.展开更多
Laser-plasma accelerators(LPAs)have great potential to realize a compact X-ray free-electron laser(FEL),which is limited by the beam properties currently.Two-color high-intensity X-ray FEL provides a powerful tool for...Laser-plasma accelerators(LPAs)have great potential to realize a compact X-ray free-electron laser(FEL),which is limited by the beam properties currently.Two-color high-intensity X-ray FEL provides a powerful tool for probing ultrafast dynamic systems.In this paper,we present a simple and feasible method to generate a two-color X-ray FEL pulse based on an LPA beam.In this scheme,time-dependent mismatch along the bunch is generated and manipulated by the designed lattice system,enabling FEL lasing at different wavelength within two undulator sections.The time separation between the two pulses can be precisely adjusted by varying the time-delay chicane.Numerical simulations show that two-color soft X-ray FELs with gigawatt-level peak power and femtosecond duration can be generated,which confirm the validity and feasibility of the scheme.展开更多
X-ray free-electron lasers(FELs)provide cutting-edge tools for fundamental researches to study nature down to the atomic level at a time-scale that fits this resolution.A precise knowledge of temporal information of F...X-ray free-electron lasers(FELs)provide cutting-edge tools for fundamental researches to study nature down to the atomic level at a time-scale that fits this resolution.A precise knowledge of temporal information of FEL pulses is the central issue for its applications.Here we proposed and demonstrated a novel method to determine the FEL temporal profiles online.This robust method,designed for ultrafast FELs,allows researchers to acquire real-time longitudinal profiles of FEL pulses as well as their arrive times with respect to the external optical laser with a resolution better than 6 fs.Based on this method,we can also directly measure various properties of FEL pulses and correlations between them online.This helps us to further understand the FEL lasing processes and realize the generation of stable,nearly fully coherent soft X-ray laser pulses at the Shanghai Soft X-ray FEL facility.This method will enhance the experimental opportunities for ultrafast science in various areas.展开更多
基金supported by the National Key Research and Development Plan(Grant No.2020YFB1709401)the National Natural Science Foundation of China(Grant Nos.12202092,12032008,and 11821202)the China Postdoctoral Science Foundation(Grant No.ZX20220734).
基金Project supported in part by the US Army Research Laboratory(ARL) Cooperative Agreement (No. W911NF-17-2-0196)the Air Force Office of Scientific Research (AFOSR)Grant (No. FA9550-19-1-0353)。
文摘Multi-agent reinforcement learning(MARL) has long been a significant research topic in both machine learning and control systems. Recent development of(single-agent) deep reinforcement learning has created a resurgence of interest in developing new MARL algorithms, especially those founded on theoretical analysis. In this paper, we review recent advances on a sub-area of this topic: decentralized MARL with networked agents.In this scenario, multiple agents perform sequential decision-making in a common environment, and without the coordination of any central controller, while being allowed to exchange information with their neighbors over a communication network. Such a setting finds broad applications in the control and operation of robots, unmanned vehicles, mobile sensor networks, and the smart grid. This review covers several of our research endeavors in this direction, as well as progress made by other researchers along the line. We hope that this review promotes additional research efforts in this exciting yet challenging area.
基金supported by the National Key Research and Development Program of China(2018YFE0103100)the National Natural Science Foundation of China(11905277,12125508,11935020,and 11975300)+2 种基金Shanghai Science and Technology Committee Rising-Star Program(20QA1410100)Program of Shanghai Aca-demic/Technology Research Leader(21XD1404100)Shanghai Pilot Program for Basic Research-Chinese Academy of Sciences,Shanghai Branch(JCYJ-SHFY-2021-010).
文摘Laser-plasma accelerators(LPAs)have great potential to realize a compact X-ray free-electron laser(FEL),which is limited by the beam properties currently.Two-color high-intensity X-ray FEL provides a powerful tool for probing ultrafast dynamic systems.In this paper,we present a simple and feasible method to generate a two-color X-ray FEL pulse based on an LPA beam.In this scheme,time-dependent mismatch along the bunch is generated and manipulated by the designed lattice system,enabling FEL lasing at different wavelength within two undulator sections.The time separation between the two pulses can be precisely adjusted by varying the time-delay chicane.Numerical simulations show that two-color soft X-ray FELs with gigawatt-level peak power and femtosecond duration can be generated,which confirm the validity and feasibility of the scheme.
文摘X-ray free-electron lasers(FELs)provide cutting-edge tools for fundamental researches to study nature down to the atomic level at a time-scale that fits this resolution.A precise knowledge of temporal information of FEL pulses is the central issue for its applications.Here we proposed and demonstrated a novel method to determine the FEL temporal profiles online.This robust method,designed for ultrafast FELs,allows researchers to acquire real-time longitudinal profiles of FEL pulses as well as their arrive times with respect to the external optical laser with a resolution better than 6 fs.Based on this method,we can also directly measure various properties of FEL pulses and correlations between them online.This helps us to further understand the FEL lasing processes and realize the generation of stable,nearly fully coherent soft X-ray laser pulses at the Shanghai Soft X-ray FEL facility.This method will enhance the experimental opportunities for ultrafast science in various areas.