FMS Features

FMS includes the following:

• Matrix Assembly
  Efficiently initializes matrix data. Options for finite-element assembly. Full matrixes specified by rows, columns or blocks.

• Triangular Factoring
  [L][D][L]t for symmetric cases, [L][U] for nonsymmetric. Nonlinear restart and substructuring capability. Pivot precision monitoring or full column partial pivoting: sign monitoring to check positive definiteness or generate Sturm sequences. Factorization reusable for multiple solution vectors.

• Vector Solution
  Forward reduction, diagonal scaling, and back-substitution. Parallel processing for multiple solution vectors.

• Array Multiplication
  Includes matrix-vector and vector-vector cases. Useful for eigenvalue extraction and quadratic form evaluation. Also includes weighted matrix triple-product for finite element integral evaluation.

• Macrotasking Library
  Includes task initiation, task synchronization and critical section locking for running your code in parallel.

• Error Reporting
  Fatal errors invoke an intelligent display of data including traceback.

• Performance Reporting
  Optional reports include CPU and elapsed times, actual megaflops, disk transfer rates and profile of all FMS subroutines called.

• Documentation
  Complete set including a step-by-step guide for incorporating FMS into your application with several examples.