CAREER: Next Generation Online Resource Allocation

这项学院早期职业发展计划(Career)补助金将通过支持研究动态资源分配的新模型和算法，为促进国家繁荣和经济福利做出贡献。这项工作将开发新颖、实用的算法解决方案，以适应共享移动性、电动汽车电池更换和在线广告等应用中出现的复杂功能、目标和限制。这项研究将(I)系统的算法洞察导致实际实现和更健壮的在线平台；(Ii)识别和开发在线资源分配问题与其他领域(如排队论)之间的联系，以及(Iii)能够发现分析在线算法的新方法。随附的教育计划旨在扩大STEM的兴趣，并通过培训社区大学教师与在线资源分配和运营工程相关的主题，并合作开发社区大学课程的互动学习模块，为代表不足的社区提供机会。由该奖项支持的研究将为现代在线资源分配环境制定下一代模型，并为其解决方案设计直观的算法，强调可扩展到大型问题实例和尽可能最佳的理论性能保证。这将通过确定一般结构属性来实现，这些属性导致算法具有广泛的适用性和对不断变化的环境的健壮性。一个主要的技术重点将是开发分析自适应在线算法的方法，这些算法可以对问题实例中随机不确定性的实现做出反应。自适应算法通常具有最好的实际性能，但其理论分析具有挑战性，除非在某些特定环境中，否则难以理解。结果将包括最坏情况下的性能界限和数值实验，将建议的算法与最先进的方法进行比较。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This Faculty Early Career Development Program (CAREER) grant will contribute to the advancement of national prosperity and economic welfare by supporting research to study new models and algorithms for dynamic allocation of resources. This work will develop novel, practical algorithmic solutions that can accommodate the complex features, objectives, and constraints arising in applications such as shared mobility, battery swapping for electric vehicles and online advertising. The research will (i) systematic algorithmic insights leading to practical implementations and more robust online platforms; (ii) identify and exploit connections between online resource allocation problems and other areas such as queueing theory, and (iii) enable discovery of new methods for analyzing online algorithms. The accompanying educational plan aims to broaden STEM interest and to provide opportunities for underrepresented communities by training community college instructors on topics related to online resource allocation and operations engineering and collaboratively developing interactive learning modules for community college courses.The research supported by this award will formulate the next generation of models for modern online resource allocation environments and design intuitive algorithms for their solution, emphasizing scalability to large problem instances and the best possible theoretical performance guarantees. This will be accomplished by identifying general structural properties that lead to algorithms that are broadly applicable and robust to changing environments. A major technical emphasis will be on developing methods to analyze adaptive online algorithms that can react to realization of stochastic uncertainties in the problem instance. Adaptive algorithms typically have the best practical performance, but their theoretical analysis is challenging and poorly understood except in some specific settings. Results will include worst case performance bounds and numerical experiments comparing the proposed algorithms with state-of-the-art approaches.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.