# -*- coding: utf-8 -*-
# pylint: disable=line-too-long, unused-wildcard-import
"""
|conv_nn| performs simple nearest-neighbour interpolations between an arbitrary number
of models (or data files) providing output and an arbitrary number of models requiring
input.
Integration tests
=================
.. how_to_understand_integration_tests::
We perform the following examples over a simulation period of 3 days:
>>> from hydpy import Element, Node, pub
>>> pub.timegrids = "2000-01-01", "2000-01-04", "1d"
|conv_nn| implements no parameter with values depending on the simulation step size,
which is why we can pass anything (or nothing) to function |parameterstep| without
changing the following results:
>>> from hydpy.models.conv_nn import *
>>> parameterstep()
Due to the following configuration, |conv_nn| queries its input from the inlet nodes
`in1` and `in2` and passes the interpolation results to the outlet nodes `out1`,
`out2`, `out3`, and `out4`:
>>> in1, in2 = Node("in1"), Node("in2")
>>> element = Element("conv",
... inlets=(in1, in2),
... outlets=["out1", "out2", "out3", "out4"])
The following coordinate definitions contain the particular case of outlet node `out4`,
being in the middle of inlet nodes `in1` and `in2` exactly:
>>> 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),
... out4=(1.0, 1.0))
In the first example, we perform a strict nearest-neighbour interpolation, where we
always take only one input location into account, even in case of missing data:
>>> maxnmbinputs(1)
|conv_nn| does not implement any state or log sequences and thus has no memory at all,
making finalising the test setup quite easy. We only need to define time series for
both inlet nodes. Note that we set some |numpy| |numpy.nan| values to demonstrate how
|conv_nn| deals with missing values:
>>> element.model = model
>>> from hydpy.core.testtools import IntegrationTest
>>> test = IntegrationTest(element)
>>> test.dateformat = "%Y-%m-%d"
>>> with pub.options.checkseries(False):
... in1.sequences.sim.series = 1.0, 2.0, nan
... in2.sequences.sim.series = 4.0, nan, nan
The calculated results provide no surprises. However, outlet node `out4` receives the
output of node `in1` instead of the equidistant node `in2`, which is due to their
definition order when preparing parameter |InputCoordinates| above:
>>> test()
| date | inputs | outputs | in1 | in2 | out1 | out2 | out3 | out4 |
---------------------------------------------------------------------------------------------
| 2000-01-01 | 1.0 4.0 | 1.0 4.0 1.0 1.0 | 1.0 | 4.0 | 1.0 | 4.0 | 1.0 | 1.0 |
| 2000-01-02 | 2.0 nan | 2.0 nan 2.0 2.0 | 2.0 | nan | 2.0 | nan | 2.0 | 2.0 |
| 2000-01-03 | nan nan | nan nan nan nan | nan | nan | nan | nan | nan | nan |
For the last timestep, where no input values are available at all, there is nothing we
can do. However, by using the second nearest location as an alternative, we at least
achieve complete output for the second timestep:
>>> maxnmbinputs(2)
>>> test()
| date | inputs | outputs | in1 | in2 | out1 | out2 | out3 | out4 |
---------------------------------------------------------------------------------------------
| 2000-01-01 | 1.0 4.0 | 1.0 4.0 1.0 1.0 | 1.0 | 4.0 | 1.0 | 4.0 | 1.0 | 1.0 |
| 2000-01-02 | 2.0 nan | 2.0 2.0 2.0 2.0 | 2.0 | nan | 2.0 | 2.0 | 2.0 | 2.0 |
| 2000-01-03 | nan nan | nan nan nan nan | nan | nan | nan | nan | nan | nan |
"""
# import...
# ...from HydPy
from hydpy.core import modeltools
from hydpy.exe.modelimports import *
from hydpy.models.conv import conv_model
[docs]
class Model(conv_model.BaseModel):
"""|conv_nn.DOCNAME.complete|."""
DOCNAME = modeltools.DocName(
short="Conv-NN", description="nearest neighbour interpolation"
)
__HYDPY_ROOTMODEL__ = True
INLET_METHODS = (conv_model.Pick_Inputs_V1,)
RECEIVER_METHODS = ()
RUN_METHODS = (conv_model.Calc_Outputs_V1,)
ADD_METHODS = ()
OUTLET_METHODS = (conv_model.Pass_Outputs_V1,)
SENDER_METHODS = ()
SUBMODELINTERFACES = ()
SUBMODELS = ()
tester = Tester()
cythonizer = Cythonizer()
