Efficient Spectral Algorithms for Massive and Dynamic Graphs is an EPSRC Fellowship awarded to He Sun (EP/T00729X/1). The fellowship studies advances of spectral graph theory, designs efficient spectral algorithms for massive and dynamic graphs, and develops an open-source library of spectral algorithms for graphs. With a total award of 1,507,133 pounds, the Fellowship runs from 2020 to 2024, and the current team consists of the PI, 2 Postdocs, and 2 PhD students. This presentation summaries the recent activities of the research group.
He Sun, Principal Investigator
Aparajita Haldar, Postdoctoral Researcher
Peter Macgregor, Postdoctoral Researcher
Ben Jourdan, PhD Student
Steinar Laenen, PhD Student
As part of the Fellowship, we are developing STAG, an open-source library of efficient spectral algorithms for graphs. STAG is the first such algorithmic library mainly written in C++, with a python wrapper around the underlying C++ library for python users. The source code of STAG can be found from our GitHub page, and will be updated actively. More information can be also found from the STAG website.
Fast Approximation of Similarity Graphs with Kernel Density Estimation.
with P. Macgregor (NeurIPS'23, Spotlight).
Fast and Simple Spectral Clustering in Theory and Practice. P. Macgregor (NeurIPS'23).
Three Hardness Results for Graph Similarity Problems. with D. Vagnozzi. submitted
The Support of Open versus Closed Random Walks.
with T. Sauerwald, and D. Vagnozzi (ICALP'23).
Measuring the Impact of a Database, with P. Buneman, D. Dosso, M. Lissandrini, and G. Silvello.
Is the Algorithmic Kadison-Singer Problem Hard? with B. Jourdan, and P. Macgregor (ISAAC'23).
Fully-Dynamic Graph Sparsifiers Against an Adaptive Adversary.
with A. Bernstein, J. van den Brand, M. Gutenberg, D. Nanongkai, T. Saranurak, and A. Sidford (ICALP'22).
Conference version | arXiv version