instituto de matemáticas universidad de sevilla
Antonio de Castro Brzezicki
Partial smoothness: a powerful paradigm for algorithm designing
Seminario Doctorado
PhD Program Activity
Date: 14.12.2016 From 10.30 till 12.00
Venue: Seminario I (IMUS), Edificio Celestino Mutis
 Over the past years, partial smoothness has proven to be a powerful tool for algorithmic and sensitivity analysis. In this talk, we will discuss the strength of partial smoothness on the designing of fast algorithms, which is one of the key topics of modern optimization. More precisely, given a optimization method, we will discuss how partial smoothness enables us to analyze the advantage and more importantly the bottle-neck of the method, which in turn allows us to design proper schemes to accelerate the method.