skuntz / minitri

miniTri data analytics miniapp

miniTri is a simple, triangle-based data analytics code. miniTri is a miniapp in the Mantevo project (http://www.mantevo.org) at Sandia National Laboratories The primary authors of miniTri are Jon Berry and Michael Wolf ([email protected]).

miniTri v. 1.0. Copyright (2016) Sandia Corporation.

For questions, contact Jon Berry ([email protected]) or Michael Wolf ([email protected]).

Please read the accompanying README and LICENSE files.

---

Description:

miniTri is a proxy for a class of triangle based data analytics (Mantevo). This simple code is a self-contained piece of C++ software that uses triangle enumeration with a calculation of specific vertex and edge properties. Key uses related to miniTri include dense subgraph detection, characterizing graphs, improving community detection, and generating graphs. Related applications exist in cyber security, intelligence, and functional biology. miniTri attempts to be more application relevant than standard data analytics benchmarks such as Graph 500.

Authors: Jon Berry ([email protected]), Michael Wolf ([email protected]) and Dylan Stark

The objective of the miniTri miniapp is to calculate a specific number (k) for all triangles in the graph. miniTri has the following basic steps (some of which can be combined):

1. Find all triangles in the graph;
2. For all triangles in the graph, calculate the vertex triangle degrees;
3. For all triangles in the graph, calculate the edge triangle degrees; and
4. For all each triangle, calculate integer k given triangle degree info.

From these k values, an upper bound on the largest clique in the graph can be calculated.

Citing miniTri

For citing miniTri or the linear algebra based formulation of miniTri use the following citation:

M.M. Wolf, J.W. Berry, and D.T. Stark, “A Task-Based Linear Algebra Building Blocks Approach for Scalable Graph Analytics,” Proc. of 19th Annual IEEE High Performance Extreme Computing Conference, 2015.

For citing the task-parallel linear algebra based miniTri work use the following citation:

M.M. Wolf, H.C. Edwards, and S.L. Olivier, “Kokkos/Qthreads Task-Parallel Approach to Linear Algebra Based Graph Analytics,” Proc. of 20th Annual IEEE High Performance Extreme Computing Conference, 2016.