# Multi-Pixel

CHAI-class receivers have multiple pixels per array (e.g., 7 HFA pixels)
and multiple arrays (HFA, LFA, 4G1--4G4). The engine handles this via
a single vectorized path.

## Vectorized Path

`resolve_calibration_load_full()` computes gamma, T_rec, and bad
channels for **all pixels simultaneously** as `[C, R, A]` arrays.
The calibration loop (`calibrate_full`) broadcasts these across the
dump and subscan axes -- no per-pixel loop needed.

This is the **only** calibration path used by `cal-cli`,
`cal-io::pipeline`, and `cal-py`.

### Common PWV Strategy

By default (`pwv_per_pixel = false`), PWV is fitted from pixel 0 only
and propagated to all other pixels via `PwvSource::External`. This
ensures consistent atmospheric correction across the focal plane and
reduces computation by $\frac{1}{N_{pixels}}$.

Set `pwv_per_pixel = true` (`--pwv-per-pixel` on the CLI) to fit
each pixel independently (useful for diagnosing receiver-dependent
systematics).

### Per-Pair Atmosphere Fitting

When `pwv_per_pixel` is enabled, the engine can also perform per-pair
atmosphere fitting (`determine_atmosphere_per_pair`), where each
ON-OFF pair gets its own atmospheric transmission computed from its
associated OFF subscan's airmass and sky signal.

### Multi-Frontend Linked PWV

For multi-frontend observations (e.g., HFA + LFA + 4G bands observed
simultaneously), the engine pools PWV estimates across all frontends to
produce a single linked PWV value. This is handled by
`cal-io/src/pipeline/linked_pwv.rs` and ensures physically consistent
atmospheric correction across all bands in a multi-frontend session.

**Implementation:**

- `cal-core/src/scan/cal_load.rs` -- `CalibrationLoadFull` with
  `[C, R, A]` arrays
- `cal-core/src/calibrate.rs` -- `calibrate_full()` broadcasts
  across all pixels
- `cal-io/src/pipeline/wrappers.rs` -- `common_pwv_source()`
- `cal-io/src/pipeline/linked_pwv.rs` -- multi-frontend PWV pooling