Re: Sparsity estimates for Matrix Multiplication

Hi Nakul, The paper I referenced uses the term "Boolean matrix", but really it applies to any matrix product in which two properties hold: 1. Zero product property. a * b = 0 ==> a = 0 or b = 0 2. Few, if any, "sums to zero". There should be few cases of nonzero partial products summin

Re: Sparsity estimates for Matrix Multiplication

Thank you Dylan. The paper seems interesting. The abstract reads as follows : "We consider the problem of doing fast and reliable estimation of the number z of non-zero entries in a sparse boolean matrix product"... I think they maybe doing this for boolean matrices. The code that I pointed to in S

Re: Sparsity estimates for Matrix Multiplication

In case it is helpful, this 2010 paper is my favorite for estimating the nnz of a matrix product with high confidence. The algorithm is much more invo

Mirror download links

Please note that the Apache mirror download links on the project website ( http://systemml.apache.org/download) currently do not work (this relates to our move to an Apache TLP). The project artifacts are available at: https://dist.apache.org/repos/dist/release/systemml/0.14.0-incubating/ This is

Sparsity estimates for Matrix Multiplication

Hi, There is code in systemml to “estimate” the output sparsity of a matrix multiplication operation between two sparse matrices. Is there a citation for this? If not, have we done any tests to figure out how good these estimates are? https://github.com/nakul02/systemml/blob/df8d4a63d8d09cae94b6ca

Re: On the need for Parameter Server. ( A Model Parallel Construct )

Waiting for the response from Niketan Pansare,and other committers. [SYSTEMML-739] ​ Parameter Server: a model parallel construct ​ Thanks, Janardhan > On Sun, Jun 18, 2017 at 10:16 PM, Janardh