def sis_simulation(prop, grid, data, seed, marginal_probs, use_correlogram=True, mask=None, force_single_thread=False, force_parallel=False, use_harddata=True, use_regions=False, region_size = (), min_neighbours = 0): out_prop, is_lvm, marginal_probs, mask = __prepare_sis(prop, data, marginal_probs, mask, use_harddata) prop_2 = _create_hpgl_ind_masked_array(out_prop, grid) ikps = __create_hpgl_ik_params(data, len(data), is_lvm, marginal_probs) means = [] if is_lvm: for i in range(len(data)): means.append(_create_hpgl_float_array(marginal_probs[i], grid)) if not is_lvm: _hpgl_so.hpgl_sis_simulation( prop_2, ikps, len(data), seed, _create_hpgl_ubyte_array(mask, grid) if mask != None else None) # hpgl.sis_simulation(_prop_to_tuple(out_prop), grid.grid, data, seed, False, use_correlogram, mask) else: _hpgl_so.hpgl_sis_simulation_lvm( prop_2, ikps, _c_array(_HPGL_FLOAT_ARRAY, len(data), means), len(data), seed, _create_hpgl_ubyte_array(mask, grid) if mask != None else None, use_correlogram ) #hpgl.sis_simulation_lvm(_prop_to_tuple(out_prop), grid.grid, data, seed, marginal_probs, use_correlogram, mask) return out_prop
def sis_simulation(prop,
grid,
data,
seed,
marginal_probs,
use_correlogram=True,
mask=None,
force_single_thread=False,
force_parallel=False,
use_harddata=True,
use_regions=False,
region_size=(),
min_neighbours=0):
out_prop, is_lvm, marginal_probs, mask = __prepare_sis(
prop, data, marginal_probs, mask, use_harddata)
prop_2 = _create_hpgl_ind_masked_array(out_prop, grid)
ikps = __create_hpgl_ik_params(data, len(data), is_lvm, marginal_probs)
means = []
if is_lvm:
for i in range(len(data)):
means.append(_create_hpgl_float_array(marginal_probs[i], grid))
if not is_lvm:
_hpgl_so.hpgl_sis_simulation(
prop_2, ikps, len(data), seed,
_create_hpgl_ubyte_array(mask, grid) if mask != None else None)
# hpgl.sis_simulation(_prop_to_tuple(out_prop), grid.grid, data, seed, False, use_correlogram, mask)
else:
_hpgl_so.hpgl_sis_simulation_lvm(
prop_2, ikps, _c_array(_HPGL_FLOAT_ARRAY, len(data), means),
len(data), seed,
_create_hpgl_ubyte_array(mask, grid) if mask != None else None,
use_correlogram)
#hpgl.sis_simulation_lvm(_prop_to_tuple(out_prop), grid.grid, data, seed, marginal_probs, use_correlogram, mask)
return out_prop
def sgs_simulation(prop, grid, cdf_data, radiuses, max_neighbours, cov_model, seed, kriging_type="sk", mean=None, use_harddata = True, use_regions=False, region_size = None, mask=None, force_single_thread=False, force_parallel=False, min_neighbours = 0, **params):
"""Performs Sequential Gaussian Simulation
Parameters:
-----------
cdf_data: None or array_like or CdfData:
Cumulative distribution data or data to use for calulating it.
If None - no cdf transformation is performed.
radiuses : tuple of 3 integers
Search radiuses by X, Y and Z axes.
max_neighbours: integer
Maximum number of neighbour points to use.
cov_model: CovarianceModel
Model of covariance.
seed: integer
Seed for random number generator.
kriging_type:"sk" or "ok", optional
Selects either simple or ordinary kriging type. Defaults to Simple Kriging.
mean:None or integer or array_like, optional
Stationary mean value or varying mean array. None - stationary mean value
will be calculated from source data. Default: None
use_harddata: bool, optional
True - to use source data values for simulation. False - to ignore source data
values. Default: True.
mask: None or array_like, optional:
Array containing 1 in cell that need to be simulated and 0 in cell that aren't.
If None simulate all cells. Defualt: None.
"""
prop.fix_shape(grid)
cov_model = normed_cov_model(cov_model)
out_prop, mean, mask = __prepare_sgs(
prop=prop,
mean=mean,
use_harddata=use_harddata,
mask=mask)
sgsp = _HPGL_SGS_PARAMS(
covariance_type = cov_model.type,
ranges = cov_model.ranges,
angles = cov_model.angles,
sill = cov_model.sill,
nugget = cov_model.nugget,
radiuses = radiuses,
max_neighbours = max_neighbours,
kriging_kind = {"sk" : _HPGL_KRIGING_KIND.simple, "ok" : _HPGL_KRIGING_KIND.ordinary}[kriging_type],
seed = seed,
min_neighbours = min_neighbours)
if (cdf_data is None):
hpgl_cdf = None
else:
C.byref(_create_hpgl_nonparam_cdf(cdf_data))
hpgl_mask = C.byref(_create_hpgl_ubyte_array(mask, grid)) if mask != None else None
hpgl_cdf = _create_hpgl_nonparam_cdf(cdf_data)
if mean is None or numpy.isscalar(mean):
_hpgl_so.hpgl_sgs_simulation(
C.byref(_create_hpgl_cont_masked_array(out_prop, grid)),
C.byref(sgsp),
hpgl_cdf,
C.byref(C.c_double(mean)) if mean != None else None,
hpgl_mask
)
else:
_hpgl_so.hpgl_sgs_lvm_simulation(
C.byref(_create_hpgl_cont_masked_array(out_prop, grid)),
C.byref(sgsp),
hpgl_cdf,
C.byref(_create_hpgl_float_array(mean, grid)),
hpgl_mask)
return out_prop
def sgs_simulation(prop,
grid,
cdf_data,
radiuses,
max_neighbours,
cov_model,
seed,
kriging_type="sk",
mean=None,
use_harddata=True,
use_regions=False,
region_size=None,
mask=None,
force_single_thread=False,
force_parallel=False,
min_neighbours=0,
**params):
"""Performs Sequential Gaussian Simulation
Parameters:
-----------
cdf_data: None or array_like or CdfData:
Cumulative distribution data or data to use for calulating it.
If None - no cdf transformation is performed.
radiuses : tuple of 3 integers
Search radiuses by X, Y and Z axes.
max_neighbours: integer
Maximum number of neighbour points to use.
cov_model: CovarianceModel
Model of covariance.
seed: integer
Seed for random number generator.
kriging_type:"sk" or "ok", optional
Selects either simple or ordinary kriging type. Defaults to Simple Kriging.
mean:None or integer or array_like, optional
Stationary mean value or varying mean array. None - stationary mean value
will be calculated from source data. Default: None
use_harddata: bool, optional
True - to use source data values for simulation. False - to ignore source data
values. Default: True.
mask: None or array_like, optional:
Array containing 1 in cell that need to be simulated and 0 in cell that aren't.
If None simulate all cells. Defualt: None.
"""
prop.fix_shape(grid)
cov_model = normed_cov_model(cov_model)
out_prop, mean, mask = __prepare_sgs(prop=prop,
mean=mean,
use_harddata=use_harddata,
mask=mask)
sgsp = _HPGL_SGS_PARAMS(covariance_type=cov_model.type,
ranges=cov_model.ranges,
angles=cov_model.angles,
sill=cov_model.sill,
nugget=cov_model.nugget,
radiuses=radiuses,
max_neighbours=max_neighbours,
kriging_kind={
"sk": _HPGL_KRIGING_KIND.simple,
"ok": _HPGL_KRIGING_KIND.ordinary
}[kriging_type],
seed=seed,
min_neighbours=min_neighbours)
if (cdf_data is None):
hpgl_cdf = None
else:
C.byref(_create_hpgl_nonparam_cdf(cdf_data))
hpgl_mask = C.byref(_create_hpgl_ubyte_array(
mask, grid)) if mask != None else None
hpgl_cdf = _create_hpgl_nonparam_cdf(cdf_data)
if mean is None or numpy.isscalar(mean):
_hpgl_so.hpgl_sgs_simulation(
C.byref(_create_hpgl_cont_masked_array(out_prop, grid)),
C.byref(sgsp), hpgl_cdf,
C.byref(C.c_double(mean)) if mean != None else None, hpgl_mask)
else:
_hpgl_so.hpgl_sgs_lvm_simulation(
C.byref(_create_hpgl_cont_masked_array(out_prop, grid)),
C.byref(sgsp), hpgl_cdf,
C.byref(_create_hpgl_float_array(mean, grid)), hpgl_mask)
return out_prop
