Reflections on Dabeen Lee's talk at #MIP2023 on "Non-smooth and robust submodular maximization." Submodularity - the principle of diminishing returns - is an essential concept in optimization, surfacing across various fields, including machine learning, network design, and sensor placement. Lee's presentation tackled the problem of maximizing continuous DR-submodular functions, which may not always be smooth. Drawing potential parallels to energy systems planning, the submodular nature of benefits can be observed in the deployment of Distributed Energy Resources (DERs) or Electric Vehicle (EV) charging stations. As additional units are installed, the marginal benefit in certain locations may decrease. The goal, when positioning DERs, might be to augment the total energy supply to the grid or curtail grid losses without having to curtain wind or shed demand. Similarly, the planning of EV charging stations could aim to maximize coverage or minimize driver inconvenience. Robust and distributionally robust optimization methods could aid decision-making, ensuring performance remains reliable under varying scenarios, thereby leading to more resilient energy and transportation systems. While I speculate on these potential applications, the talk opens up an interesting avenue for thought with the broader impact and adaptability of optimization methodologies in our changing world. #SubmodularOptimization #RobustOptimization #EnergySystems #MIP2023

Using #apricot for #submodularoptimization be like.. @lkpino @wfondrie

Randomized distributed algorithms for submodular optimization problem #discreteoptimization #optimization #submodularoptimization #combinatorialoptimization #submodularity Great talk by Professor Alina Ene, junior @BU_Computing faculty fellow. @BUCompSci