[Коллоквиум]: Beyond Worst Case Analysis of Graph Partitioning Algorithms

Speaker - Konstantin Makarychev, Microsoft Research Many combinatorial optimization problems are much simpler in practice than in the worst-case. One of the challenges in the area of approximation algorithms is to explain this phenomenon and to design algorithms that work well in real-life. In this talk, I will first discuss different ways of modelling “real-life” instances. Then, I will present a new semi-random semi-adversarial model for graph partitioning problems, a planted model with permutation-invar