09:30
11:00

Regarding the manuscript, a preliminary version (without the suggestions by the jury) is available here: https://drive.google.com/file/d/14knVFx0qd18HhEMwORbVkihO0xje7DZg/view?usp=sharing

Summary: This document represents my habilitation thesis, which is required for obtaining the accreditation to supervise research (also known as habilitation). In it, I summarize parts of my research activities on the topic of scheduling for high-performance computing (HPC) systems. In order to make this knowledge more accessible, especially for students and young researchers, I use a less formal and more visual representation of the problems, models, and algorithms from our publications. 
This document covers topics related to the meaning of scheduling, its importance, and the way we approach scheduling problems. It also includes detailed discussions on three research topics in which I co-supervised young researchers: scheduling for federated learning systems, scheduling for software-defined radio, and scheduling for green cloud and edge systems. For every scheduling problem we solved, it presents how we combined ideas from basic algorithmic blocks — coming from classic problems and techniques — in order to build a solution. It also presents perspectives related to each of the three research topics, and to my future research endeavors.

Keywords: scheduling - resource allocation - high performance computing - parallel computing - distributed computing - algorithms

 

English
Salle Ada Lovelace (Inria)