PhD-TA Load Balancing Dynamics in Complex Networks

Are you eager to use your mathematical skills to explore the dynamics of randomized load balancing policies in complex compatibility graphs? Do you want to develop probabilistic methods to analyze and improve the performance of large-scale systems such as data centers and cloud networks? We invite applications for a PhD-TA position at the intersection of Applied Probability, Networking and Stochastic Operations Research. In this role you will specifically apply local-convergence techniques to gain valuable insight in the evolution of congestion phenomena in sparse and possibly dynamic random graphs.

Information
Randomized load balancing provides a key mechanism for distributing tasks among servers in resource-sharing networks. The analysis and design of load balancing policies has attracted immense attention in recent years, mainly spurred by crucial scalability challenges arising in parallel-processing systems with massive numbers of servers such as data centers and cloud computing platforms.

The performance of load balancing policies is well-understood in settings where any server can handle any task or where the compatibility graph is suitably dense. In real-life deployment scenarios, however, tasks have highly heterogeneous processing requirements, and servers have increasingly specialized features (e.g. data locality), giving rise to fairly sparse and possibly even dynamic compatibility graphs.

The asymmetry induced by the compatibility graph breaks the fundamental underpinnings of existing scaling limits and raises major methodological challenges. These issues manifest themselves particularly in sparse regimes where the network structure plays a crucial role, creating a strong need for innovative approaches. This PhD project will specifically leverage local-convergence techniques in order to get a grip on the dynamics of randomized load balancing policies. Local convergence has evolved into a key methodology to deal with sparse random graphs. In particular, the aim is to obtain insight how the local network structure impacts the overall performance, and use that to derive guidelines for optimal network design.

As a PhD-TA, you will also be crucially involved in educational activities in the area of stochastic processes. Depending on your interests, you can contribute as an instructor, lecturer or (co-)supervisor of BSc and MSc final projects. You will have the unique opportunity to spark students’ interest in this area, and help students develop the skills and obtain the knowledge needed to excel in their future career.

This position is in the Statistics, Probability and Operations Research (SPOR) cluster in the Mathematics and Computer Science (M&CS) Department at Eindhoven University of Technology (TU/e). The SPOR cluster has a vibrant research environment, accommodating strong scientists with diverse backgrounds from Applied Probability to Discrete Optimization. In particular, you will be a member of the Stochastic Operations Research group, which focuses on evaluating and optimizing the performance and reliability of large-scale

systems that operate in the presence of randomness and uncertainty. We develop mathematical models and methods for the design, analysis, optimization and control of such systems with techniques at the crossroads of Applied Probability and Operations Research. At the heart of the SPOR cluster is the workshop institute EURANDOM, offering major opportunities for you to learn through various international workshops and even co-organize workshops.

We are looking for a strong PhD candidate who is excited to perform cutting-edge research in the area of complex load balancing networks. Key characteristics of an ideal candidate include:

• A MSc degree in mathematics, preferably with specialization in Applied Probability or Stochastic Operations Research.
• Ideally some familiarity with large-deviations theory, asymptotic analysis, random graph theory, queueing networks and resource allocation algorithms.
• Strong independency in analytical thinking, and intrinsic drive to deep thinking and complex problem solving.
• Enthusiasm to contribute to the educational activities in the area of stochastic processes.
• Great communication skills, excellent team-working capabilities, strong academic writing competences, and fluency in English (CEFR level C1 or above).