prepare_observation

observations.prepare_observation(monitor, grid, sample, variable_names)

Prepare a streaming monitor for block-wise execution.

Calls monitor.prepare(grid, sample, variable_names), which receives a SimulationGrid, a jax.ShapeDtypeStruct describing one [channels, nodes] input value, and optional channel names, and returns a PreparedObservation. Subclasses select their preparation through ordinary method resolution. To stream a type you cannot modify, wrap it in a StreamingMonitor subclass that implements prepare.

The returned recipe is validated abstractly for cadence, callback return structure, state invariance, rank, node count, dtype, and channel metadata before execution.

# prepare_observation { #tvboptim.observations.prepare_observation }

```python
observations.prepare_observation(monitor, grid, sample, variable_names)
```

Prepare a streaming monitor for block-wise execution.

Calls ``monitor.prepare(grid, sample, variable_names)``, which receives a
``SimulationGrid``, a ``jax.ShapeDtypeStruct`` describing one
``[channels, nodes]`` input value, and optional channel names, and returns
a ``PreparedObservation``. Subclasses select their preparation through
ordinary method resolution. To stream a type you cannot modify, wrap it in
a ``StreamingMonitor`` subclass that implements ``prepare``.

The returned recipe is validated abstractly for cadence, callback return
structure, state invariance, rank, node count, dtype, and channel metadata
before execution.