Registered Data

[CT166]

[02443] Enhanced charge-based algorithm and its application in reliability-redundancy allocation problems

  • Session Date & Time : 3C (Aug.23, 13:20-15:00)
  • Type : Contributed Talk
  • Abstract : Reliability-redundancy allocation problems (RRAP) require selecting components with diverse choices and redundancy levels that maximize profit within constraints. The nonlinearity and non-smoothness in RRAP have defeated many traditional mathematical approaches. Therefore, in this paper, a new version of the charge-based artificial electric field algorithm (AEFA) is proposed, incorporating a novel Coulomb's constant and bounds. This restructuring improves the adaptability of AEFA on RRAPs. The suggested algorithm outperforms other existing algorithms on seven RRAPs.
  • Classification : 90C15, 90B25, 90C29, 90C06, 90C26
  • Author(s) :
    • Dikshit Chauhan (Dr B R Ambedkar National Institute of Technology Jalandhar)
    • Anupam Yadav (Dr B R Ambedkar National Institute of Technology Jalandhar)

[01034] Machine Learning for Two-stage Robust Optimization

  • Session Date & Time : 3C (Aug.23, 13:20-15:00)
  • Type : Contributed Talk
  • Abstract : When dealing with problems under uncertainty, two-stage robust optimization is one of the key approaches: you obtain an optimal robust solution, and adapt to the real scenario. However, these problems are one of the hardest optimization problem classes. To accelerate finding high-quality solutions, we propose a machine learning-based strategy. We experimentally show that with using our strategy you can train based on small problems, and apply them to bigger problems, while still getting good results.
  • Classification : 90C17, 90C90
  • Author(s) :
    • Esther Julien (Delft University of Technology)
    • Krzysztof Postek (Delft University of Technology)
    • Ilker Birbil (University of Amsterdam)

[01586] A robust optimization approach to a repair shop network planning

  • Session Date & Time : 3C (Aug.23, 13:20-15:00)
  • Type : Contributed Talk
  • Abstract : A robust optimization model is a paradigm for decision-making under uncertainty, where parameters are given in the form of uncertainty sets. First, we develop a robust optimization model for a repair shop network planning problem that involves the product of uncertain parameters in the constraints. We formulate its robust counterpart with the help of max-min regret and Lagrangian dual approach, considering the partial information of uncertain parameters is given in the different types of uncertainty sets. We apply the robust optimization model to a bi-objective multi-plant repair shop network planning problem where multiple pieces of equipment are repaired and overhauled using several resources over a multi-period planning horizon.
  • Classification : 90C17, 90C90
  • Author(s) :
    • Shubham Singh (IIT kanpur)

[02298] New semidefinite relaxations for a class of complex quadratic programming problems

  • Session Date & Time : 3C (Aug.23, 13:20-15:00)
  • Type : Contributed Talk
  • Abstract : In this talk, we propose some new semide finite relaxations for a class of nonconvex complex quadratic programming problems, widely appear in signal processing and power system. By deriving new valid constraints to the matrix variables in the lifted space, we derive some enhanced semide finite relaxations of complex quadratic programming problems. Then, we compare the proposed semide finite relaxations with existing ones, and show that the newly proposed semide finite relaxations could be strictly tighter than the previous ones. Numerical results indicate that the proposed semidefi nite relaxations not only provide tighter relaxation bounds but also improve some existing approximation algorithms by finding better sub-optimal solutions.
  • Classification : 90C20, 90C22, 90C35
  • Author(s) :
    • Zhibin Deng (University of Chinese Academy of Sciences)
    • Yinzhe Xu (North China Electric Power University)
    • Cheng Lu (North China Electric Power University)
    • Yafeng Liu (Chinese Academy of Sciences)

[00007] Sufficient Conditions for SDP Representability of Non-compact Convex Sets

  • Session Date & Time : 3C (Aug.23, 13:20-15:00)
  • Type : Industrial Contributed Talk
  • Abstract : The paper discusses the classical optimization problem of semidefinite representable $(SDR)$ non-compact convex sets. We introduce new notions, compactly SDR set, SDR away from 0 and SDR at infinity. We prove: the cone of feasible directions of a compactly SDR set has semidefinite representation. In addition, we characterize the polar of compactly SDR sets. It also illustrates the smallest cone containing compactly SDR set and its’ projections, which leads to new sufficient conditions for semidefinite representation.
  • Classification : 90C22, 90C06, 90C90
  • Author(s) :
    • Anusuya Ghosh (Senior Software Engineer)
    • Vishnu Narayanan (Associate professor)