2. Using SUNDIALS

As discussed in §7.3, the six solvers packages (CVODE(S), IDA(S), ARKODE, KINSOL) that make up SUNDIALS are built upon common classes/modules for vectors, matrices, and algebraic solvers. In addition, the six packages all leverage some other common infrastructure, which we discuss in this section.

• 2.1. The SUNContext Type
  • 2.1.1. Implications for task-based programming and multi-threading
  • 2.1.2. Convenience class for C++ Users
• 2.2. Performance Profiling
  • 2.2.1. Enabling Profiling
  • 2.2.2. Profiler API
  • 2.2.3. Example Usage
  • 2.2.4. Other Considerations
• 2.3. SUNDIALS Status Logging
  • 2.3.1. Enabling Logging
  • 2.3.2. Logger API
  • 2.3.3. Example Usage
• 2.4. SUNDIALS Version Information
• 2.5. SUNDIALS Fortran Interface
  • 2.5.1. Data Types
  • 2.5.2. Notable Fortran/C usage differences
    • 2.5.2.1. Creating generic SUNDIALS objects
    • 2.5.2.2. Arrays and pointers
    • 2.5.2.3. Passing procedure pointers and user data
    • 2.5.2.4. Passing NULL to optional parameters
    • 2.5.2.5. Working with N_Vector arrays
    • 2.5.2.6. Providing file pointers
  • 2.5.3. Important notes on portability
  • 2.5.4. Common Issues
• 2.6. Features for GPU Accelerated Computing
  • 2.6.1. SUNDIALS GPU Programming Model
  • 2.6.2. Steps for Using GPU Accelerated SUNDIALS