# Test: [Name]
Describe what this test checks. One markdown file per test; the runner discovers all `*.md` in this folder (except TEMPLATE.md) and runs the test defined in the code block below.
## Fixed values (test config)
Put a single fenced code block with language **`test`** containing a JSON object. Required keys:
- **test_name**: string, short id (used in results).
- **script**: string, one of `hpc_verification`, `graph`, `baseline` (must match a runner script).
- **params**: object, script-specific parameters. Omitted keys use script defaults.
### Example (hpc_verification)
```test
{
"test_name": "solar_system",
"script": "hpc_verification",
"params": {
"steps": 45,
"seed": 123,
"expansion_per_tick": 1.0,
"deficit_S": 0.05,
"deep_deficit_S": 1e-4,
"cost_diagnostic_step": 0,
"initial_H": 2.5e-8
}
}
```
### Param reference (hpc_verification)
| Param | Type | Default | Description |
|-------|------|---------|-------------|
| steps | int | 45 | Ledger ticks per run. |
| seed | int | 123 | RNG seed. |
| expansion_per_tick | float | 1.0 | 1.0 = no Hubble. |
| deficit_S | float | 0.05 | Signal at C for “deficit” run. |
| deep_deficit_S | float | 1e-4 | Signal at C for deep deficit / geodesic. |
| cost_diagnostic_step | int \| null | null | If set, print base vs A5 cost on this step. |
| initial_H | float | 2.5e-8 | So a0 is comparable to S. |
Other scripts (e.g. `graph`, `baseline`) define their own params; see their docstrings or this folder’s examples.
## Implementation (optional, for transparency)
Add an **Implementation** section to show which Python runs the test and a short code snippet. For run_tests.py-driven tests: document that the runner calls `run_verification(params, output_dir, verbose)` from `rst_reality_engine_hpc_verification`. For suite-driven tests (e.g. EFE, log_radial): reference the function in `rst_reality_engine_test_suite.py` and show the core logic.
## Running tests
From the Reality Engine directory (parent of `tests/`):
```bash
python run_tests.py
```
- Discovers all `*.md` in `tests/` except `TEMPLATE.md`.
- Parses the `test` code block and runs the script with `params`.
- Writes outputs to `tests/results/<test_name>/` (plot, report).
- Use `python run_tests.py --quiet` for less per-test output, or `python run_tests.py solar_system` to run one test.