Tessellated Distributed Computing
Abstract
The work considers the -server distributed computing scenario with users requesting functions that are linearly-decomposable over an arbitrary basis of real (potentially non-linear) subfunctions. In our problem, the aim is for each user to receive their function outputs, allowing for reduced reconstruction error (distortion) , reduced computing cost (; the fraction of subfunctions each server must compute), and reduced communication cost (; the fraction of users each server must connect to). For any given set of requested functions -- which is here represented by a coefficient matrix -- our problem is made equivalent to the open problem of sparse matrix factorization that seeks -- for a given parameter , representing the number of shots for each server -- to minimize the reconstruction distortion overall -sparse and -sparse matrices and . With these matrices respectively defining which servers compute each subfunction, and which users connect to each server, we here design our by designing tessellated-based and SVD-based fixed support matrix factorization methods that first split into properly sized and carefully positioned submatrices, which we then approximate and then decompose into properly designed submatrices of and .
Cite
@article{arxiv.2404.14203,
title = {Tessellated Distributed Computing},
author = {Ali Khalesi and Petros Elia},
journal= {arXiv preprint arXiv:2404.14203},
year = {2024}
}
Comments
65 Pages, 16 figure. The manuscript is submitted to IEEE Transactions on Information Theory