Papers
1. Convexification of multi-period quadratic programs with indicators [pdf]
Jisun Lee, Andrés Gómez, and Alper Atamtürk. (Preprint)
- We study multi-period MIQPs utilizing unique properties of (block) separable matrices. A convex hull representation in an extended space and a polynomial-time algorithm are proposed.
2. Cut generation for hybrid model predictive control through coupling periods
Jisun Lee and Alper Atamtürk. (working paper)
- Two types of nonlinear cuts are generated by considering the costs of consecutive periods in hybrid control optimization problems, leveraging matrix decomposition and SOCP reformulation.
3. Efficient sampling from ϵ-optimality solution set of mixed-integer programming problems [pdf]
Jisun Lee, Alper Atamtürk, and Ignacio Aravena Solís. (working paper)
- We propose a parallelizable sampling method that efficiently splits the ε-optimality solution set of MILPs into smaller polytopes and generates well-scattered samples.
4. Strong formulations for hybrid system control [pdf]
Jisun Lee, Hyungki Im, and Alper Atamtürk. (Preprint)
- We propose a strong formulation for MIQPs arising in hybrid model predictive control by generating cuts using disjunctive programming and perspective reformulation.
5. A fully polynomial time approximation scheme for the probability maximizing shortest path problem [pdf]
Jisun Lee, Seulgi Joung, and Kyungsik Lee. European Journal of Operational Research, 2022.
- The probability maximization of a shortest path problem on a directed graph with Gaussian random arc costs is studied. We prove the problem is NP-hard and propose an FPTAS that iteratively solves deterministic SPP.
Presentations
1. Convexification of multi-period mixed-integer quadratic optimization.
- 2024 INFORMS Annual Meeting, Seattle, USA. [slides]
2. Convexification of multi-period quadratic programs with indicators.
- 2024 KAIST ISE Winter Symposium, Daejeon, South Korea. [slides]
3. Strong formulation of hybrid control problem with tridiagonal inverse matrix.
- 2024 Mixed Integer Programming Workshop, Kentucky, USA. [poster]
4. A fully polynomial time approximation scheme for the probability maximizing shortest path problem.
- 2019 European Conference on Operational Research, Dublin, Ireland. [slides]
- 2019 Spring & Fall Conference of Korean Institute of Industrial Engineers, Seoul, South Korea.