9. Matrix Data Structures

The SUNDIALS library comes packaged with a variety of SUNMatrix implementations, designed for simulations requiring direct linear solvers for problems in serial or shared-memory parallel environments. SUNDIALS additionally provides a simple interface for generic matrices (akin to a C++ abstract base class). All of the major SUNDIALS packages (CVODE(s), IDA(s), KINSOL, ARKODE), are constructed to only depend on these generic matrix operations, making them immediately extensible to new user-defined matrix objects. For each of the SUNDIALS-provided matrix types, SUNDIALS also provides SUNLinearSolver implementations that factor these matrix objects and use them in the solution of linear systems.

• 9.1. Description of the SUNMATRIX Modules
• 9.2. Description of the SUNMATRIX operations
• 9.3. SUNMATRIX functions used by ARKODE
• 9.4. SUNMatrix functions used by CVODE
• 9.5. SUNMatrix functions used by CVODES
• 9.6. SUNMatrix functions used by IDA
• 9.7. SUNMatrix functions used by IDAS
• 9.8. SUNMatrix functions used by KINSOL
• 9.9. The SUNMATRIX_DENSE Module
• 9.10. The SUNMATRIX_MAGMADENSE Module
• 9.11. The SUNMATRIX_ONEMKLDENSE Module
• 9.12. The SUNMATRIX_BAND Module
• 9.13. The SUNMATRIX_CUSPARSE Module
• 9.14. The SUNMATRIX_SPARSE Module
• 9.15. The SUNMATRIX_SLUNRLOC Module
• 9.16. The SUNMATRIX_GINKGO Module
• 9.17. The SUNMATRIX_KOKKOSDENSE Module
• 9.18. SUNMATRIX Examples