DSS

Direct sequential simulation: SGS in data units, with no normal-score transform of the kriged field.

Each node is simple kriged from the raw values of its neighbors, with its domain's declustered mean as the known mean. variogram is the variogram of the values: within each domain its nugget and sills are rescaled so that the total sill equals the domain's declustered variance, keeping its shape and relative nugget. The node is then drawn from the domain's declustered histogram, as the back-transform of a Gaussian m + s·z whose back-transformed mean and variance are the kriged ones. A kriged pair that no such draw reaches, such as a mean beyond the data or a large variance near the tails, takes the nearest reachable pair; simulate warns with the fraction of such nodes. Each draw matches the kriged mean and variance, not the shape of the histogram: on skewed grades with a large nugget, low means take wide draws and the realizations hold too many extreme grades, which correct_distribution maps back to the histogram.

search is a Search, or a sequence of them as passes, as in SGS. With domains at fit, Search.soft lets the raw grades of other domains inform a node within the soft distance, and high_grade compares grades with its threshold, kriging a clamped neighbor as the threshold.

correlation property

correlation

Correlation of the standardized primary and secondary used in collocated co-DSS; None when fitted without secondary.

fit method descriptor

fit(
    coords,
    values,
    *,
    weights=None,
    holes=None,
    domains=None,
    domain_column=None,
    secondary=None,
    correlation=None,
)

Takes the conditioning data. Samples sharing a location keep the first, with a warning naming their holes.

Parameters:

Name	Type	Description	Default
coords	(array_like, shape(n, 2) or (n, 3), PointSet or BlockModel)		required
values	array_like, shape (n,), or str	Values, or the column of coords holding them; so for weights and holes.	required
weights	array_like or str	Declustering weights, for each domain's histogram, mean and variance.	None
holes	array_like or str	Drill-hole ids or names, for max_per_hole.	None
domains	array_like or label	Domain label of each sample, or one label for all: strings, numbers or booleans. Each domain has its own histogram, mean and rescaled variogram.	None
domain_column	str	The column of coords holding the domains, instead of domains.	None
secondary	array_like or str	A secondary variable at the data, for collocated co-DSS: simulate then needs it at every target. Each node's kriged mean and variance become those of the collocated simple cokriging, in standardized units, of its neighbors and the secondary at the node, under the Markov model (the cross-covariance is correlation times the primary covariance), and the node is drawn from the histogram with that mean and variance. The primary is standardized by its domain's declustered mean and variance, the secondary by its own at the data.	None
correlation	float	Correlation of the standardized primary and secondary, in [-1, 1]; by default their weighted correlation at the data. 0 simulates as without secondary.	None

Raises:

Type	Description
InvalidInput	If the values of a domain do not vary, a Search.soft names a domain without samples or has no domains to work on, both domains and domain_column are given, or correlation comes without secondary or outside [-1, 1].

from_parquet staticmethod

from_parquet(path)

Reads to_parquet output; raises InvalidInput on another class's file or a newer format.

passes method descriptor

passes(
    targets,
    *,
    anisotropy=None,
    domains=None,
    domain_column=None,
)

The search pass of every target, as in SGS.passes.

Parameters:

Name	Type	Description	Default
targets	array_like or BlockModel		required
anisotropy	LocalAnisotropy	As in simulate.	None
domains	array_like or label	As in simulate.	None
domain_column	str	As in simulate.	None

Returns:

Type	Description
ndarray	The pass, 1-based; NaN where no search finds enough data, and simulate uses the last.

simulate method descriptor

simulate(
    targets,
    *,
    n=100,
    seed=0,
    cutoffs=None,
    quantiles=Ellipsis,
    tolerances=Ellipsis,
    keep=None,
    anisotropy=None,
    blocks=None,
    window=None,
    groups=None,
    domains=None,
    domain_column=None,
    secondary=None,
    grade_tonnage_cutoffs=None,
    density=None,
    tonnage=None,
    categories=None,
)

Summary of n realizations at targets, each seeded from seed and its index.

Parameters:

Name	Type	Description	Default
targets	array_like or BlockModel		required
n	int		100
seed	int		0
cutoffs	sequence of float or str	Grades for the probability and mean above, in the unit of the values or as text such as "0.5 g/t".	None
quantiles	sequence of float	Probabilities for the values at quantiles.	Ellipsis
tolerances	sequence of float	Relative tolerances for SimulationSummary.precision, as in SGS.simulate.	Ellipsis
keep	bool or sequence of int	Keep all realizations, or those indices, as an (n, targets) array.	False
anisotropy	LocalAnisotropy	Orients each node's variogram and search.	None
blocks	BlockModel	Coarser blocks each realization is averaged to, weighted by node volume, and summarized at; nodes outside every block are ignored and a block holding no node is an error.	None
window	tuple of float	(dx, dy, dz) box centered on each row (the blocks when given, else the targets), summarized at the average of the rows whose center it holds, as in SGS.simulate.	None
groups	array_like or str	Label of each row, or a column name: one summary row per distinct label, in numpy.unique order, as in SGS.simulate.	None
domains	array_like or label	Needed when fitted with domains: labels of the targets, one label for all, or an (n, targets) array of simulated domains, realization k of the grades simulated within row k. A target in a domain without samples raises InvalidInput.	None
domain_column	str	The column of PointSet or BlockModel targets holding the domains.	None
secondary	array_like or str	Needed when fitted with one: the secondary variable at the targets, the name of a column of PointSet or BlockModel targets, or an (n, targets) array such as the realizations of a simulation of the secondary, realization k then cosimulated with row k.	None
grade_tonnage_cutoffs	optional	Grade–tonnage curves per realization, as in SGS.simulate.	None
density	optional	Grade–tonnage curves per realization, as in SGS.simulate.	None
tonnage	optional	Grade–tonnage curves per realization, as in SGS.simulate.	None
categories	optional	Grade–tonnage curves per realization, as in SGS.simulate.	None

Returns:

Type	Description
SimulationSummary

Warns:

Type	Description
UserWarning	With the fraction of nodes whose kriged mean and variance no draw from the histogram reaches, drawn from the nearest reachable pair.

to_parquet method descriptor

to_parquet(path)

Writes arrays as Parquet columns and parameters as JSON in the file metadata; from_parquet reads it back.

used in

• collocated cosimulation
• direct sequential simulation