Tensors, or multidimensional arrays, have many use in data analysis applications. The additional dimensions over matrices (or two dimensional arrays) enable gleaning information that is otherwise unreachable. A remarkable example comes from the Netflix Challenge. The aim of the challenge was to improve the company's algorithm for predicting user ratings on movies using a dataset containing a set of ratings of users on movies. The winning algorithm, when the challenge was concluded, had to use the time dimension on top of user x movie rating, during the analysis. Tensors from many applications, such as the mentioned one, are sparse, which means that not all entries of the tensor are relevant or known. The PeachTree project investigates the building blocks of numerical parallel tensor computation algorithms on high end systems, and designs a set of scheduling and combinatorial tools for achieving efficiency.
PeachTree project is a collaborative research project, called Associated Team, formed by
the members of the Inria Team ROMA and TDAlab of Georgia Institute of Technology (an agreement is being drafted to be signed).
Updated technical report and codes:
Jules Bertrand, Fanny Dufossé, Somesh Singh, and Bora Uçar.
Algorithms and data structures for hyperedge queries, Research Report, RR-9390, Inria Grenoble - Rhone Alpes (report, codes), April 2022.
New paper (March 2022):
Somesh Singh and Bora Uçar,
An Efficient parallel implementation of a perfect hashing method for hypergraphs,
to appear in the IPDPSW proceedings GrAPL 2022: Workshop on Graphs, Architectures, Programming, and Learning, Lyon, France. Local
Somesh has presented parallel hashing algorithms for tensors at SIAM PP 2022 (virtual; Feb 2022).
Funding period is adjusted as 2022--2024 due to COVID19 pandemic.
We are now using PIGO for all our I/O operations with sparse tensors and matrices (Feb 2022). Thanks Kasimir and Umit.
BU and UVC meeting on I/O with tensors 27 May 2021, (online).
L3 Internship, June-July 2021. Elodie Bernard worked on the hypergraph strong connectivity problem.
M2 Internship, Feb-July 2021. Jules Bertrand continues his work on hyperedge queries.
Technical report and codes:
Jules Bertrand, Fanny Dufossé, and Bora Uçar.
Algorithms and data structures for hyperedge queries, Research Report, RR-9390, Inria Grenoble - Rhone Alpes (report, codes), February 2021.
M1 Internship, May-July 2020. Jules Bertrand was an M1 intern at ENS de Lyon. He worked on hyperedge queries.
BU and UVC attended SIAM PP and SIAM CSC 11--15 2020, Seattle.
Kick-off meeting, Feb 15, 2020, Seattle (BU and UVC).