The ML package in Trilinos provides algebraic multigrid methods based upon smoothed aggregation. ML contains algorithm variants appropriate for Poisson, elasticity, convection-diffusion, H(curl), and H(div) problems. The successor to ML is MueLu, which supports a variety of scalar types via templating, including extended precision types. Optimal algorithmic scaling in MueLu has been demonstrated on 1 million cores of Sequoia BG/Q.

Usage and applications: ML is distributed in Trilinos (> 4800 downloads/year). ML is used throughout the DOE complex in ASC applications, in Department of Defense applications, and at universities worldwide. For example, ML is used within Sandia’s Sierra engineering mechanics simulation suite.


Karen Devine

Sandia National Laboratories, P.O. Box 5800, MS 1318, Albuquerque, NM 87185-1318