Talk
Talk by Anshika Chuahdhary
Title: Optimization Methods: From Set-Valued Algorithms and Analysis of Interval-Valued Functions to Operator Splitting for Large-Scale Problems Google Meet Link
Abstract:
Optimization and variational analysis provide a fundamental mathematical framework for modeling and solving complex problems arising in engineering, economics, data science, and machine learning. Many such problems involve nonsmoothness, uncertainty, or large-scale structures, motivating the development of robust theoretical tools and efficient computational algorithms. In practical deployments, sensing errors, communication uncertainties, localization inaccuracies, and dynamic environments inevitably introduce bounded uncertainty into the underlying optimization problem.
The first part of the seminar discusses the generalized optimization, encompassing interval-valued and set-valued optimization, followed by generalized subdifferential theory, optimality conditions, and derivative-based characterizations for nonsmooth optimization problems. Next, the development of Newton and quasi-Newton type algorithms together with their convergence properties will be illustrated for solving structured optimization problems. The other half of the seminar introduces the recent work on monotone inclusion problems and operator splitting methods. In particular, the fixed-point formulations based on Douglas-Rachford splitting and multi-block
ADMM schemes, together with the analysis of averagedness and convergence properties of the proposed algorithms.
Keywords: Interval-valued optimization; Multi-objective optimization; Operator splitting; Uncertainty quantification; Multi-agent systems.
Speaker Bio:Dr. Ansihka is an institute postdoctoral fellow in the Department of Electrical Engineering. She completed her Ph.D. at the Indian Institute of Technology (BHU), Varanasi, under the supervision of Dr. Debdas Ghosh. Her doctoral research focused on the nonsmooth analysis of interval-valued functions and the development of optimization methods for set-valued maps. During her Ph.D., she received the SERB Overseas Visiting Doctoral Fellowship, which supported a year-long research visit to Purdue University, USA, where she worked with Prof. Xiangxiong Zhang. Her research interests include interval-valued and set-valued optimization, nonsmooth analysis, and operator-splitting methods.
- 17 - 17 September,2026
- 02:00 pm - 03:00 pm
- EE Committee Room