1 reporting.surface_reference

reporting.surface_reference

Scene-aware reporting for a completed nested surface-reference extension.

The Oracle owner validates the raw evidence and returns one paired candidate table. This module reduces that table with pandas and draws publication diagnostics with Seaborn; it never changes the experiment or its labels.

1.1 Functions

Name	Description
export_surface_extension_shape	Write hash-bound scene summary and Seaborn figures from canonical readers.
shape_statistics	Separate within-state score-shape agreement from candidate-choice regret.

1.1.1 export_surface_extension_shape

reporting.surface_reference.export_surface_extension_shape(
    assignment_path,
    run_dir,
    output_dir,
)

Write hash-bound scene summary and Seaborn figures from canonical readers.

1.1.2 shape_statistics

reporting.surface_reference.shape_statistics(scores, comparison)

Separate within-state score-shape agreement from candidate-choice regret.

Spearman rho measures rank agreement; affine R² measures whether an offset and scale explain dense scores. They are undefined for constant shells. Regret is imported from the Oracle’s predeclared top-set comparison, not recomputed from a second tie policy. Scene means give each physical scene one vote; candidates and repeated seeds are not independent replicates.

# reporting.surface_reference { #aria_nbv.reporting.surface_reference }

`reporting.surface_reference`

Scene-aware reporting for a completed nested surface-reference extension.

The Oracle owner validates the raw evidence and returns one paired candidate
table. This module reduces that table with pandas and draws publication
diagnostics with Seaborn; it never changes the experiment or its labels.

## Functions

| Name | Description |
| --- | --- |
| [export_surface_extension_shape](#aria_nbv.reporting.surface_reference.export_surface_extension_shape) | Write hash-bound scene summary and Seaborn figures from canonical readers. |
| [shape_statistics](#aria_nbv.reporting.surface_reference.shape_statistics) | Separate within-state score-shape agreement from candidate-choice regret. |

### export_surface_extension_shape { #aria_nbv.reporting.surface_reference.export_surface_extension_shape }

```python
reporting.surface_reference.export_surface_extension_shape(
    assignment_path,
    run_dir,
    output_dir,
)
```

Write hash-bound scene summary and Seaborn figures from canonical readers.

### shape_statistics { #aria_nbv.reporting.surface_reference.shape_statistics }

```python
reporting.surface_reference.shape_statistics(scores, comparison)
```

Separate within-state score-shape agreement from candidate-choice regret.

Spearman rho measures rank agreement; affine R² measures whether an offset
and scale explain dense scores. They are undefined for constant shells.
Regret is imported from the Oracle's predeclared top-set comparison, not
recomputed from a second tie policy. Scene means give each physical scene
one vote; candidates and repeated seeds are not independent replicates.