Index of /afs/cs.cmu.edu/academic/class/15440-f11/code/class17/matrixmult

Icon  Name                    Last modified      Size  Description
[DIR] Parent Directory - [TXT] BadGraphException.java 19-Jul-2007 09:51 191 [TXT] Graph.java 30-Jul-2007 11:19 1.8K [TXT] GraphEdge.java 30-Jul-2007 10:54 2.2K [TXT] MatrixMultDriver.java 30-Jul-2007 11:06 2.6K [TXT] P1Mapper.java 20-Jul-2007 18:32 1.0K [TXT] P1Reducer.java 20-Jul-2007 18:35 1.7K [TXT] P2Mapper.java 20-Jul-2007 15:05 893 [TXT] P2Reducer.java 30-Jul-2007 11:23 1.1K
This directory contains a Hadoop implementation of sparse matrix multiply.

Matrices are described in textual form as a list of nonzero entries,
each of the form:

row col value ID

where row & col are the row & column indices (integers), value is the
matrix entry, and ID is the name of the array.

The main program, MatrixMultDriver, takes as arguments the file names of
the two input matrices, the pathname for the output, and optionally
the pathname for the directory in which to store intermediate files.

The code was written for Hadoop version 0.13.0.  It is not guaranteed
to work with more recent versions of Hadoop.

Files:
	MatrixMultDriver.java:
		Implements the main function

	BadGraphException.java:
	Graph.java:
	GraphEdge.java:
		Implement a graph suitable for representing sparse matrices

	P1Mapper.java:
	P1Reducer.java:
	P2Mapper.java:
	P2Reducer.java:
		Implement map and reduce operations for the two phases
		of matrix multiplication