check_realizations

check_realizations(
    model,
    realizations,
    data,
    values,
    *,
    weights=None,
    variogram=None,
    lag=None,
    max_lag=None,
    directions=None,
    tolerance=22.5,
    normal_scores=True,
) -> RealizationCheck

How well realizations on model reproduce the histogram, variogram and correlations of their data.

Each realization's quantiles are compared with the declustered quantiles of the data, in data units and in the data's normal scores; its experimental variogram along directions with the data's and with the model it was drawn from; with several variables, its correlation matrix with the data's. Categorical realizations compare category proportions and indicator variograms instead. Realization variograms on a regular or masked BlockModel pair cells by index shifts, in parallel over realizations, identically on any number of threads.

Parameters:

Name	Type	Description	Default
model	BlockModel	The nodes the realizations were simulated at.	required
realizations	SimulationSummary, CategoricalSummary, list of SimulationSummary or array_like	Summaries simulated with keep= (a list from MultivariateSimulation), or (n, targets) arrays, one per variable; integer arrays are categories 0..k.	required
data	(PointSet or array_like, shape(m, 2) or (m, 3))	Conditioning data, or their coordinates.	required
values	str, array_like or list of them	Data values, one per variable, or the columns of data holding them; NaN is dropped.	required
weights	str or array_like	Declustering weights of the data.	None
variogram	Variogram or list of Variogram	Model of each variable (of each category's indicator for categories), in normal scores when normal_scores.	None
lag	float	Lag width and largest lag of the variograms; without them no variograms are computed.	None
max_lag	float	Lag width and largest lag of the variograms; without them no variograms are computed.	None
directions	list of (float, float)	(azimuth, dip) of each direction. Default: along the first variogram's azimuth and dip, across it horizontally and, on a 3D grid, vertically; omnidirectional without variogram.	None
tolerance	float	Direction half-angle in degrees.	22.5
normal_scores	bool	Variograms of continuous variables in the normal scores of the declustered data, the units of Gaussian simulation models; else in data units.	True

Returns:

Type	Description
RealizationCheck

used in

• From drill holes to a 3D model
• Simulation
• target distribution correction
• local proportions
• SGS along a shared path
• collocated cosimulation
• realization checks