As of August of 2018, I moved to Argonne National Laboratory. I work with Mihai Anitescu at the MCS division. I was a postdoctoral researcher at the Wisconsin Institute for Discovery, working with Prof. Michael Ferris. I'm broadly interested in the area of linear/nonlinear optimization and complementarity problems with an emphasis on equilibrium programming and variational inequalities.
I received my Ph.D. degree in Computer Sciences from the University of Wisconsin-Madison under the supervision of Prof. Michael Ferris. I received M.S. in Computer Science and Engineering and B.S. in Computer Science and Engineering and Mathematics, both from POSTECH in South Korea.Here is my CV.
Youngdae Kim, Sven Leyffer, Todd Munson: MPEC methods for bilevel optimization problems. 2020 (to appear in the book entitled Bilevel optimization: advances and next challenges)
Noam Goldberg, Steffen Rebennack, Youngdae Kim, Vitaliy Krasko, Sven Leyffer. MINLP formulations for continuous piecewise linear function fitting. Computational Optimization and Applications, 2021
Youngdae Kim, Mihai Anitescu. A real-time optimization with warm-start of multiperiod AC optimal power flows. Electric Power Systems Research, 189, 2020
Wonjun Chang, Michael C. Ferris, Youngdae Kim, Thomas F. Rutherford. Solving stochastic dynamic programming problems: a mixed complementarity approach. Computational Economics. 55, 925-955, 2020
Youngdae Kim, Michael C. Ferris. Solving equilibrium problems using extended mathematical programming. Mathematical Programming Computation. 11(3): 457-501, 2019 (project page)
Youngdae Kim, Olivier Huber, Michael C. Ferris. A structure-preserving pivotal method for affine variational inequalities. Mathematical Programming, Series B. 168(1-2): 93-121, 2018
Noam Goldberg, Youngdae Kim, Sven Leyffer, Thomas D. Veselka. Adaptively refined dynamic program for linear spline regression. Computational Optimization and Applications. 58(3): 523-541, 2014
Youngdae Kim, Ilhwan Ko, Wook-shin Han, Hwanjo Yu. iKernel: Exact indexing for support vector machines. Information Sciences. 257(1): 32-53, 2014
Hwanjo Yu, Jinha Kim, Youngdae Kim, Seung-won Hwang, Young Ho Lee. An efficient method for learning nonlinear ranking SVM functions. Information Sciences. 209(20): 37-48, 2012
Youngdae Kim, Gae-won You, Seung-won Hwang. Ranking strategies and threats: a cost-based pareto optimization approach. Distributed and Parallel Databases. 26(1): 127-150, 2009
Sihan Zeng, Alyssa Kody, Youngdae Kim, Kibaek Kim, Daniel K. Molzahn. A reinforcement learning approach to parameter selection for distributed optimization in power systems. (to appear in PSCC, 2022)
Youngdae Kim, Kibaek Kim. A mixed complementarity problem approach for steady-state voltage and frequency stability analysis. (to appear in PES-GM, 2022)
Youngdae Kim, Mihai Anitescu. A real-time optimization with warm-start of multiperiod AC optimal power flows. In the proceedings of the 21st Power Systems Computation Conference, 2020
Hwanjo Yu, Ilhwan Ko, Youngdae Kim, Seung-won Hwang, Wook-shin Han. Exact indexing for support vector machines. In Proc. of the ACM SIGMOD Conference on Data Management, June 12-16, 2011, Athens, Greece, pp. 709-720
Hwanjo Yu, Youngdae Kim, Seung-won Hwang. RV-SVM: An efficient method for learning ranking SVM. Pacific-Asia Conference on Knowledge Discovery and Data Mining, pp 426-438, 2009
Youngdae Kim, Gae-won You, Seung-won Hwang. Escaping a dominance region at minimum cost. International Conference on Database and Expert Systems Applications, pp 800-807, 2008
Youngdae Kim, Seung-won Hwang. Approximate boolean + ranking query answering using wavelets. International Conference on Web-Age Information Management, pp 17-24, 2008
Minseok Ryu, Youngdae Kim, Kibaek Kim, Ravi K. Madduri. APPFL: Open-Source Software Framework for Privacy-Preserving Federated Learning
Youngdae Kim, François Pacaud, Kibaek Kim, Mihai Anitescu. Leveraging GPU batching for scalable nonlinear programming through massive Lagrangian decomposition.
Youngdae Kim, Kibaek Kim. Accelerated computation and tracking of AC optimal power flow solutions using GPUs.
Anirudh Subramanyam, Youngdae Kim, Michel Schanen, François Pacaud, Mihai Anitescu. A globally convergent distributed Jacobi scheme for block-structured nonconvex constrained optimization problems.
Mihai Anitescu, Kibaek Kim, Youngdae Kim, Daniel Maldonado, François Pacaud, Vishwas Rao, Michel Schanen, Sungho Shin, Anirudh Subramanyam. Targeting exascale with Julia on GPU for multiperiod optimization with scenario constraints.
Michel Schanen, Daniel Adrian Maldonado, François Pacaud, Alexis Montoison, Mihai Anitescu, Kibaek Kim, Youngdae Kim, Vishwas Rao, Anirudh Subramanyam. Julia as portable high-level language for numerical solvers of power flow equations on GPU architectures. 2021
Work in progress
Youngdae Kim, Michael C. Ferris: SELKIE: a model transformation and distributed solver for structured equilibrium problems.
Youngdae Kim, Michael C. Ferris: An efficient model generation for decomposition methods in modeling languages.
This implements the real-time optimization algorithm described in our PSCC 2020 paper. It exploits warm-start and sensitivity analysis results.
SELKIE is an agent-based decomposition method for equilibrium problems. It exploits agents' structure to decompose a given model into smaller sub-models, possibly amenable to parallel computations, so that we can find a more robust and faster solution path.
It defines a new set of constructs that enable a natural translation of the algebraic formulation of equilibrium problems into modeling languages such as GAMS. It automatically reformulates the given equilibrium problem into a corresponding mixed complementarity problem (MCP). It also provides constructs to exploit the problem structure by the back-end solvers.
PATH VI is a Newton-based complementary pivotal method for variational inequalities. It can be used to solve some equilibrium problems such as generalized Nash equilibrium problems (GNEP) and multiple optimization problems with equilibrium constraints (MOPEC) in one-shot.
PATH VI will be integrated into GAMS (General Algebraic Modeling System) so that users do not have to create interfaces tailored to PATH VI.
Block LU update routine provides an efficient rank-1 update, especially for large-scale problems. It can be used with existing basis factorization routines such as LUSOL and UMFPACK, and shows significant performance improvement on large-scale equilibrium problems. It is being currently used as one of the linear algebra engines for PATH and PATH VI solvers.