Asynchronous Distributed Matrix Balancing and Application to Suppressing Epidemic

This paper presents an efficient asynchronous distributed algorithm for the problem of balancing a nonnegative matrix using a network of processors, each of which has access to a portion of the global matrix. The goal of the algorithm is for the processors to collaborate through local information exchange so that each processor can determine its local weighting coefficients that balance the matrix. Our algorithm is of Gauss-Seidel type with strict relaxation and converges geometrically under mild assumptions on the communication model between neighboring processors. The analysis of our algorithm is based on a novel reformulation of matrix balancing as a network consensus problem, from which an upper bound on the convergence rate can be derived. Finally, we demonstrate the applicability of the algorithm to a problem of optimally allocating curing resources for suppressing epidemic spread in a directed weighted network, where the spreading dynamic is captured by a susceptible-infected-susceptible model.

[1]  Aleksander Madry,et al.  Matrix Scaling and Balancing via Box Constrained Newton's Method and Interior Point Methods , 2017, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[2]  Martin Idel.  A review of matrix scaling and Sinkhorn's normal form for matrices and positive maps , 2016, 1609.06349.

[3]  Chinwendu Enyioha,et al.  Optimal Resource Allocation for Network Protection Against Spreading Processes , 2013, IEEE Transactions on Control of Network Systems.

[4]  Alex Samorodnitsky,et al.  A Deterministic Strongly Polynomial Algorithm for Matrix Scaling and Approximate Permanents , 1998, STOC '98.

[5]  C. Scoglio,et al.  An individual-based approach to SIR epidemics in contact networks. , 2011, Journal of theoretical biology.

[6]  T. Elfving.  On some methods for entropy maximization and matrix scaling , 1980 .

[7]  E. E. Osborne.  On Pre-Conditioning of Matrices , 1960, JACM.

[8]  Piet Van Mieghem,et al.  Epidemic processes in complex networks , 2014, ArXiv.

[9]  I. Olkin,et al.  Scaling of matrices to achieve specified row and column sums , 1968 .

[10]  M. Newman.  Spread of epidemic disease on networks. , 2002, Physical review. E, Statistical, nonlinear, and soft matter physics.

[11]  Amin Saberi,et al.  How to distribute antidote to control epidemics , 2010 .

[12]  L. Khachiyan,et al.  ON THE COMPLEXITY OF NONNEGATIVE-MATRIX SCALING , 1996 .

[13]  Piet Van Mieghem,et al.  Virus Spread in Networks , 2009, IEEE/ACM Transactions on Networking.

[14]  Sherman Robinson,et al.  Updating and Estimating a Social Accounting Matrix Using Cross Entropy Methods , 2001 .

[15]  U. Rothblum,et al.  On complexity of matrix scaling , 1999 .

[16]  John N. Tsitsiklis,et al.  Parallel and distributed computation , 1989 .

[17]  L. Khachiyan,et al.  On the Complexity of Matrix Balancing , 1997 .

[18]  Christos Faloutsos,et al.  Epidemic spreading in real networks: an eigenvalue viewpoint , 2003, 22nd International Symposium on Reliable Distributed Systems, 2003. Proceedings..

[19]  Stavros A. Zenios,et al.  A Comparative Study of Algorithms for Matrix Balancing , 1990, Oper. Res..

[20]  Christoforos N. Hadjicostis,et al.  Distributed Weight Balancing Over Digraphs , 2014, IEEE Transactions on Control of Network Systems.

[21]  Piet Van Mieghem,et al.  In-homogeneous Virus Spread in Networks , 2013, ArXiv.

[22]  Van Sy Mai,et al.  Distributed Algorithm for Suppressing Epidemic Spread in Networks , 2018 .

[23]  Asuman E. Ozdaglar,et al.  Convergence rate for consensus with delays , 2010, J. Glob. Optim..

[24]  Leonard J. Schulman,et al.  Analysis of a Classical Matrix Preconditioning Algorithm , 2015, J. ACM.

[25]  Richard Sinkhorn.  A Relationship Between Arbitrary Positive Matrices and Doubly Stochastic Matrices , 1964 .

[26]  Rafail Ostrovsky,et al.  Matrix Balancing in Lp Norms: Bounding the Convergence Rate of Osborne's Iteration , 2017, SODA.

[27]  C. Reinsch,et al.  Balancing a matrix for calculation of eigenvalues and eigenvectors , 1969 .

[28]  Christoforos N. Hadjicostis,et al.  Distributed Matrix Scaling and Application to Average Consensus in Directed Graphs , 2013, IEEE Transactions on Automatic Control.

[29]  Eduardo Ramirez-Llanos,et al.  Distributed discrete‐time optimization algorithms with applications to resource allocation in epidemics control , 2018 .