`virgil.simulate`

Simulated observations of a scene with another observation's sampling, errors and times, and bias tests that fit a model to many noise draws.

Simulated observations of a scene, and bias tests across instruments.

simulate observes a scene with the sampling, errors and times of a template: real data from one instrument, or the synthetic coverage of virgil.coverage. It answers "what would instrument B measure for the scene instrument A sees?", and, with shift_days, "what would the same coverage give at other epochs?". bias_test fits a model to many noise draws and reports the spread of the fitted parameters, which shows biases from a model that is simpler than the scene (e.g. two point sources fitted to a star with extended emission) and how well the coverage constrains a parameter.

`simulate(scene, template, key=None, noise_scale=1.0, shift_days=None)`

Observe scene with the sampling, errors and times of template.

Parameters:

Name	Type	Description	Default
`scene`	`SourceModel`	The scene. A time-dependent scene (one with `Attached` components) is evaluated at each sample's own time, so the template needs times.	required
`template`	`OIData`	The observation to copy: its uv samples, wavelengths, observables, closure-phase correlations and projections, and errors.	required
`key`	`PRNGKey`	For Gaussian noise drawn with the template's errors (closure phases correlated as in the template); noiseless without it.	`None`
`noise_scale`	`float`	Multiplies the noise (not the stored errors).	`1.0`
`shift_days`	`float`	Move every sample's time by this many days, e.g. to see what the same coverage would give at a later epoch of an orbit.	`None`

Returns:

Type	Description
`OIData`	The template with its observables replaced by the scene's.

`bias_test(scene, template, model, priors, n, key, **fit_kwargs)`

Fit model to n noisy simulations of scene.

Parameters:

Name	Type	Description	Default
`scene`	`SourceModel`	The truth, simulated with `simulate`.	required
`template`	`OIData`	The observation to simulate.	required
`model`		The model fitted to each simulation and its priors, as in `fit`. The model may differ from the scene (that is the point of a bias test).	required
`priors`		The model fitted to each simulation and its priors, as in `fit`. The model may differ from the scene (that is the point of a bias test).	required
`n`	`int`	Number of noise draws.	required
`key`	`PRNGKey`	Seed for the draws.	required
`**fit_kwargs`		Passed to `fit` (e.g. `init`, `method`).	`{}`

Returns:

Type	Description
`dict`	For each fitted path, the `n` fitted values as an array; and `chi2_red`, the reduced χ² of each fit. Compare their means and spreads with the truth.