# -*- coding: utf-8 -*-
# pylint: disable=missing-module-docstring
# import...
# ...from site-packages
import numpy
# ...from HydPy
from hydpy.core import parametertools
from hydpy.models.conv import conv_control
from hydpy.models.conv import conv_fluxes
[docs]
class NmbOutputs(parametertools.Parameter):
"""The number of outlet nodes [-]"""
NDIM, TYPE, TIME, SPAN = 0, int, None, (1, None)
CONTROLPARAMETERS = (conv_control.OutputCoordinates,)
_DEPENDENT_SEQUENCES = (
conv_fluxes.Outputs,
conv_fluxes.OutputPredictions,
conv_fluxes.OutputResiduals,
)
def __call__(self, *args, **kwargs):
super().__call__(*args, **kwargs)
for sequence in self.subpars.pars.model.sequences.fluxes:
if isinstance(sequence, self._DEPENDENT_SEQUENCES):
sequence.shape = self.value
[docs]
def update(self) -> None:
"""Determine the number of inlet nodes via inspecting control parameter
|OutputCoordinates|.
Note that invoking method |NmbOutputs.update| like calling the parameter
directly also sets the shape of flux sequence |conv_fluxes.Outputs|:
>>> from hydpy.models.conv import *
>>> parameterstep()
>>> outputcoordinates(
... out1=(0.0, 3.0),
... out2=(2.0, -1.0))
>>> derived.nmboutputs.update()
>>> derived.nmboutputs
nmboutputs(2)
>>> fluxes.outputs.shape
(2,)
>>> derived.nmboutputs(3)
>>> derived.nmboutputs
nmboutputs(3)
>>> fluxes.outputs.shape
(3,)
"""
self(self.subpars.pars.control.outputcoordinates.shape[0])
[docs]
class Distances(parametertools.Parameter):
"""Distances of the inlet nodes to each outlet node [?]."""
NDIM, TYPE, TIME, SPAN = 2, float, None, (None, None)
CONTROLPARAMETERS = (
conv_control.InputCoordinates,
conv_control.OutputCoordinates,
)
[docs]
def update(self) -> None:
"""Determine the distances.
The individual rows of parameter |Distances| correspond to the
outlet nodes; the columns contain the inlet nodes' indices:
>>> from hydpy.models.conv import *
>>> parameterstep()
>>> inputcoordinates(
... in1=(0.0, 3.0),
... in2=(2.0, -1.0))
>>> outputcoordinates(
... out1=(0.0, 3.0),
... out2=(3.0, -2.0),
... out3=(1.0, 2.0))
>>> derived.distances.update()
>>> derived.distances
distances([[0.0, 4.472136],
[5.830952, 1.414214],
[1.414214, 3.162278]])
"""
control = self.subpars.pars.control
incoords = control.inputcoordinates.__hydpy__get_value__()
outcoords = control.outputcoordinates.__hydpy__get_value__()
distances = numpy.empty((len(outcoords), len(incoords)), dtype=float)
for idx, outcoord in enumerate(outcoords):
distances[idx, :] = numpy.sqrt(
numpy.sum((outcoord - incoords) ** 2, axis=1)
)
self.__hydpy__set_shape__(distances.shape)
self.__hydpy__set_value__(distances)
[docs]
class ProximityOrder(parametertools.Parameter):
"""Indices of the inlet nodes in the order of their proximity to each
outlet node [-]."""
NDIM, TYPE, TIME, SPAN = 2, int, None, (None, None)
CONTROLPARAMETERS = (
conv_control.MaxNmbInputs,
conv_control.InputCoordinates,
conv_control.OutputCoordinates,
)
DERIVEDPARAMETERS = (Distances,)
[docs]
def update(self) -> None:
"""Determine the proximity-order of the inlet and outlet nodes.
The individual rows of parameter |ProximityOrder| correspond to the
outlet nodes; the columns contain the inlet nodes' indices:
>>> from hydpy.models.conv import *
>>> parameterstep()
>>> inputcoordinates(
... in1=(0.0, 3.0),
... in2=(2.0, -1.0))
>>> outputcoordinates(
... out1=(0.0, 3.0),
... out2=(3.0, -2.0),
... out3=(1.0, 2.0))
>>> maxnmbinputs()
>>> derived.distances.update()
>>> derived.proximityorder.update()
>>> derived.proximityorder
proximityorder([[0, 1],
[1, 0],
[0, 1]])
Set the value of parameter |MaxNmbInputs| to one,if you want to
consider the respective nearest input node only:
>>> maxnmbinputs(1)
>>> derived.proximityorder.update()
>>> derived.proximityorder
proximityorder([[0],
[1],
[0]])
"""
control = self.subpars.pars.control
nmbinputs = control.maxnmbinputs.__hydpy__get_value__()
distances = self.subpars.distances.__hydpy__get_value__()
idxs = numpy.empty((len(distances), nmbinputs), dtype=int)
for idx, distances_ in enumerate(distances):
idxs[idx, :] = numpy.argsort(distances_)[:nmbinputs]
self.__hydpy__set_shape__(idxs.shape)
self.__hydpy__set_value__(idxs)
[docs]
class Weights(parametertools.Parameter):
"""Weighting coefficients of the inlet nodes corresponding to their
proximity to each outlet node and parameter |Power| [-]."""
NDIM, TYPE, TIME, SPAN = 2, float, None, (None, None)
CONTROLPARAMETERS = (
conv_control.MaxNmbInputs,
conv_control.InputCoordinates,
conv_control.OutputCoordinates,
conv_control.Power,
)
DERIVEDPARAMETERS = (
Distances,
ProximityOrder,
)
[docs]
def update(self) -> None:
"""Determine the weighting coefficients.
The individual rows of parameter |Weights| correspond to the
outlet nodes; the rows contain the weights of the inlet nodes:
>>> from hydpy.models.conv import *
>>> parameterstep()
>>> inputcoordinates(
... in1=(0.0, 3.0),
... in2=(2.0, -1.0),
... in3=(0.0, 3.0),
... in4=(99.0, 99.0))
>>> outputcoordinates(
... out1=(0.0, 3.0),
... out2=(3.0, -2.0),
... out3=(1.0, 2.0))
>>> maxnmbinputs()
>>> power(2.0)
>>> derived.distances.update()
>>> derived.proximityorder.update()
>>> derived.weights.update()
>>> derived.weights
weights([[inf, inf, 0.05, 0.000053],
[0.5, 0.029412, 0.029412, 0.000052],
[0.5, 0.5, 0.1, 0.000053]])
You can restrict the number of inlet nodes used for each outlet
node via parameter |MaxNmbInputs|. In the following example, it
seems reasonable to set its value to three to ignore the far-distant
inlet node `in4`:
>>> maxnmbinputs(3)
>>> derived.distances.update()
>>> derived.proximityorder.update()
>>> derived.weights.update()
>>> derived.weights
weights([[inf, inf, 0.05],
[0.5, 0.029412, 0.029412],
[0.5, 0.5, 0.1]])
"""
control = self.subpars.pars.control
nmbinputs = control.maxnmbinputs.__hydpy__get_value__()
power = control.power.__hydpy__get_value__()
distances = self.subpars.distances.__hydpy__get_value__()
proximityorder = self.subpars.proximityorder.__hydpy__get_value__()
weights = numpy.empty((len(distances), nmbinputs), dtype=float)
for idx, distances_ in enumerate(distances):
sorteddistances = distances_[proximityorder[idx, :]]
jdxs = sorteddistances > 0.0
weights[idx, jdxs] = 1.0 / sorteddistances[jdxs] ** power
weights[idx, ~jdxs] = numpy.inf
self.__hydpy__set_shape__(weights.shape)
self.__hydpy__set_value__(weights)
