This paper carries out a game-theoretic analysis of a single-server queueing system with setup times under N-policy by considering both the partially observable and the partially unobservable information scenarios. Th...This paper carries out a game-theoretic analysis of a single-server queueing system with setup times under N-policy by considering both the partially observable and the partially unobservable information scenarios. The server switches off whenever the system becomes empty, and is resumed when the number of customers reaches a certain threshold value. Customers decide whether to join or to balk the system upon arrival based on their available information. The equilibrium joining strategy of customers as well as the systemzs performance measures are derived under different information levels. We find that both Follow-the-Crowd (FTC) and Avoid-the-Crowd (ATC) behaviors exist in our system. Numerical results show that the social welfare is unimodal in the threshold, and is decreasing in the waiting cost.展开更多
Multi-agent reinforcement learning is difficult to apply in practice,partially because of the gap between simulated and real-world scenarios.One reason for the gap is that simulated systems always assume that agents c...Multi-agent reinforcement learning is difficult to apply in practice,partially because of the gap between simulated and real-world scenarios.One reason for the gap is that simulated systems always assume that agents can work normally all the time,while in practice,one or more agents may unexpectedly“crash”during the coordination process due to inevitable hardware or software failures.Such crashes destroy the cooperation among agents and lead to performance degradation.In this work,we present a formal conceptualization of a cooperative multi-agent reinforcement learning system with unexpected crashes.To enhance the robustness of the system to crashes,we propose a coach-assisted multi-agent reinforcement learning framework that introduces a virtual coach agent to adjust the crash rate during training.We have designed three coaching strategies(fixed crash rate,curriculum learning,and adaptive crash rate)and a re-sampling strategy for our coach agent.To our knowledge,this work is the first to study unexpected crashes in a multi-agent system.Extensive experiments on grid-world and StarCraft II micromanagement tasks demonstrate the efficacy of the adaptive strategy compared with the fixed crash rate strategy and curriculum learning strategy.The ablation study further illustrates the effectiveness of our re-sampling strategy.展开更多
The article“Equilibrium Joining Strategies in the M/M/1 Queues with Setup Times under N-Policy”unfortunately contained a mistake about the first author’s affiliation.In the original publication of the paper,this af...The article“Equilibrium Joining Strategies in the M/M/1 Queues with Setup Times under N-Policy”unfortunately contained a mistake about the first author’s affiliation.In the original publication of the paper,this affiliation was“Department of Mathematics,Beijing Jiaotong University,Beijing 100044,China”.It should be“School of Economics and Management,University of Chinese Academy of Sciences,Beijing,100190,China”.展开更多
Photocrosslinkable polymers have been exploited to attain impressive advantages in printing freestanding,micrometer-scale,mechanically compliant features.However,a more integrated understanding of both the polymer pho...Photocrosslinkable polymers have been exploited to attain impressive advantages in printing freestanding,micrometer-scale,mechanically compliant features.However,a more integrated understanding of both the polymer photochemistry and the microfabrication processes could enable new strategic design avenues,unlocking far-reaching applications of the light-based modality of additive manufacturing.One promising approach for achieving high-aspect-ratio structures is to leverage the phenomenon of light self-trapping during the photopolymerization process.In this review,we discuss the design of materials that facilitate this optical behavior,the computational modeling and practical processing considerations to achieve high aspect-ratio structures,and the range of applications that can benefit from architectures fabricated using light self-trapping-especially those demanding free-standing structures and materials of stiffnesses relevant in biological applications.Coupled interactions exist among material attributes,including polymer composition,and processing parameters such as light intensity.We identify strong opportunities for predictive design of both the material and the process.Overall,this perspective describes the wide range of existing polymers and additive manufacturing approaches,and highlights various future directions to enable constructs with new complexities and functionalities through the development of next-generation photocrosslinkable materials and micromanufacturing methods.展开更多
基金the National Natural Science Foundation of China under Grant Nos.71871008 and 71571014.
文摘This paper carries out a game-theoretic analysis of a single-server queueing system with setup times under N-policy by considering both the partially observable and the partially unobservable information scenarios. The server switches off whenever the system becomes empty, and is resumed when the number of customers reaches a certain threshold value. Customers decide whether to join or to balk the system upon arrival based on their available information. The equilibrium joining strategy of customers as well as the systemzs performance measures are derived under different information levels. We find that both Follow-the-Crowd (FTC) and Avoid-the-Crowd (ATC) behaviors exist in our system. Numerical results show that the social welfare is unimodal in the threshold, and is decreasing in the waiting cost.
基金Project supported by the National Natural Science Foundation of China(No.61836011)the Youth Innovation Promotion Association of the Chinese Academy of Sciences(No.2018497)the GPU cluster built by the MCC Lab of Information Science and Technology Institution,USTC,China。
文摘Multi-agent reinforcement learning is difficult to apply in practice,partially because of the gap between simulated and real-world scenarios.One reason for the gap is that simulated systems always assume that agents can work normally all the time,while in practice,one or more agents may unexpectedly“crash”during the coordination process due to inevitable hardware or software failures.Such crashes destroy the cooperation among agents and lead to performance degradation.In this work,we present a formal conceptualization of a cooperative multi-agent reinforcement learning system with unexpected crashes.To enhance the robustness of the system to crashes,we propose a coach-assisted multi-agent reinforcement learning framework that introduces a virtual coach agent to adjust the crash rate during training.We have designed three coaching strategies(fixed crash rate,curriculum learning,and adaptive crash rate)and a re-sampling strategy for our coach agent.To our knowledge,this work is the first to study unexpected crashes in a multi-agent system.Extensive experiments on grid-world and StarCraft II micromanagement tasks demonstrate the efficacy of the adaptive strategy compared with the fixed crash rate strategy and curriculum learning strategy.The ablation study further illustrates the effectiveness of our re-sampling strategy.
文摘The article“Equilibrium Joining Strategies in the M/M/1 Queues with Setup Times under N-Policy”unfortunately contained a mistake about the first author’s affiliation.In the original publication of the paper,this affiliation was“Department of Mathematics,Beijing Jiaotong University,Beijing 100044,China”.It should be“School of Economics and Management,University of Chinese Academy of Sciences,Beijing,100190,China”.
基金M.Y.acknowledges the Angela Leong Fellowship Fund 2021-2022 from the Massachusetts Institute of Technology.K.K.acknowledges the financial support of the Natural Sciences and Engineering Research Council of Canada(award no.PDF-529703-2019)S.K.acknowledges the National Research Foundation of Korea(NRF)grant funded by the Korean government(MSIT)(No.2019R1A5A8083201 and 2022R1C1C1003966)。
文摘Photocrosslinkable polymers have been exploited to attain impressive advantages in printing freestanding,micrometer-scale,mechanically compliant features.However,a more integrated understanding of both the polymer photochemistry and the microfabrication processes could enable new strategic design avenues,unlocking far-reaching applications of the light-based modality of additive manufacturing.One promising approach for achieving high-aspect-ratio structures is to leverage the phenomenon of light self-trapping during the photopolymerization process.In this review,we discuss the design of materials that facilitate this optical behavior,the computational modeling and practical processing considerations to achieve high aspect-ratio structures,and the range of applications that can benefit from architectures fabricated using light self-trapping-especially those demanding free-standing structures and materials of stiffnesses relevant in biological applications.Coupled interactions exist among material attributes,including polymer composition,and processing parameters such as light intensity.We identify strong opportunities for predictive design of both the material and the process.Overall,this perspective describes the wide range of existing polymers and additive manufacturing approaches,and highlights various future directions to enable constructs with new complexities and functionalities through the development of next-generation photocrosslinkable materials and micromanufacturing methods.
