Theses

Design and development of parallel algorithms for graphcore architectures

Bachelor’s or Master’s Degree

The project focuses on the development of parallel algorithms optimized for graphcore architectures.

The algorithms considered are various and are not limited to the following list, which is reported as an example:

  • Parallel algorithms for graphs
  • Foundational primitives
  • Algorithms for solving numerical systems

    • Download the full thesis description here.

    Randomised algorithms for centrality measurement on very large graphs

    Bachelor’s or Master’s Degree

    Betweenness Centrality (BC) is a metric used to identify the most important nodes in a general graph. The BC extraction revolves around the computation of shortest paths. Assigning an exact score to each node/vertex in a graph is a challenge in computational terms, since the most interesting graphs are made up of millions or even billions of vertices and edges.

    In this thesis the candidate will be asked to investigate randomised algorithmic techniques for approximate BC computation.

    The activities include:

  • Surveying the most relevant techniques already employed in the state of the art.
  • Evaluation of some previously implemented heuristics in source code used to perform exact and approxiamate BC computation on extremely large graphs.
  • Desing of new heuristics and integration in the existing source code.

    • Download the full thesis description here.