Recipe: run tungsten with JSON (spectral session)

Goal: run the shipped tungsten application with a JSON config from the repository (production-style spectral PFC path).

Prerequisites

• Built with OpenPFC_BUILD_APPS=ON (default). CPU binary: <build>/apps/tungsten/tungsten.

• HeFFTe and MPI as in INSTALL.md.

• Inputs documented under apps/tungsten/inputs_json/README.md.

Steps

1. Build OpenPFC (Release recommended).

2. Follow the first tungsten run to copy the small single-seed input into results/first-tungsten/ and replace its site-specific output paths. From the checkout root, run:

   mpirun -n 1 ./builds/release/apps/tungsten/tungsten results/first-tungsten/input.json
   

   Increase -n only within the available allocation. On a cluster use its supported launcher; see the HPC operator guide.

3. Inspect the config if you change ranks or paths — writers and fields entries must stay consistent (spectral_app_config_reference.md, io_results.md).

Expected result

• Clean exit (0) if the run completes.

• Binary field output if configured under fields / data in the JSON (see binary_field_io_spec.md).

Next steps

• Other sample inputs: tungsten_fixed_bc.json, tungsten_moving_bc.json, tungsten_performance.json in the same folder (applications.md).

• GPU variants: gpu_path_decision.md, tutorials/gpu_app_quickstart.md.