Getting Started
This section introduces the standard execution path, Config -> Engine -> run(). In the current implementation, initial conditions are not injected by Engine itself; they are applied inside each stepper builder, which is an important detail when tracing the runtime logic.
- Installation: requirements, dependencies, and JIT compilation of CUDA/CPU extensions
- Engine Overview: config normalization, model completion, and stepper construction
- Examples: representative explicit, implicit, holo, and CNN warm-start runs