Merge commit '3c04cc03f17ff545a0a566bb4fcbb69dac4cd4fc' into gh-pages…

…-custom

* commit '3c04cc03f17ff545a0a566bb4fcbb69dac4cd4fc':
  Add release notes and contributor entry
  Pin Dockerfile image to `linux/x86_64`
  Update design text and image
  Updated forecasts documentation
  Clean docs
  Reduce size of reference result files
  Updated release notes
  Added unit tests for simulation and forecast across the year
  Extend data_manager get_data reading beyond one year

Dockerfile

@@ -1,4 +1,4 @@
-FROM ubuntu:20.04
+FROM --platform=linux/x86_64 ubuntu:20.04
 
 # Install required packages
 RUN 	apt-get update && \

contributors.md

@@ -19,6 +19,7 @@ Thank you to all who have provided guidance on the development of this software.
 - Robert Lutes, Pacific Northwest National Laboratory
 - Kefei Mo, Pacific Northwest National Laboratory
 - Erik Paulson, Independent
+- Matt Robinson, University of Colorado - Boulder
 - Jermy Thomas, National Renewable Energy Laboratory
 - Christian Veje, University of Southern Denmark
 - Draguna Vrabie, Pacific Northwest National Laboratory

data/data_manager.py

@@ -333,20 +333,36 @@ def get_data(self, horizon=24*3600, interval=None, index=None,
             # the closest possible point under stop will be the end
             # point in order to keep interval unchanged among index.
             index = np.arange(start,stop+0.1,interval).astype(int)
+        else:
+            if not isinstance(index, np.ndarray):
+                index = np.asarray(index)
 
         # Reindex to the desired index
-        data_slice_reindexed = data_slice.reindex(index)
-
-        for key in data_slice_reindexed.keys():
-            # Use linear interpolation for continuous variables
-            if key in self.categories['weather']:
-                f = interpolate.interp1d(self.case.data.index,
-                    self.case.data[key], kind='linear')
-            # Use forward fill for discrete variables
-            else:
-                f = interpolate.interp1d(self.case.data.index,
-                    self.case.data[key], kind='zero')
-            data_slice_reindexed.loc[:,key] = f(index)
+        start = index[0]
+        # 1 year in (s)
+        year = 31536000
+        # Starting year
+        year_start = int(np.floor(start/year))*year
+        # Normalizing index with respect to starting year
+        index_norm = index - year_start
+        stop_norm = index_norm[-1]
+        # If stop happens across the year divide df and interpolate separately
+        if stop_norm > data_slice.index[-1]:
+            idx_year = (np.abs(index_norm - year)).argmin() + 1
+            # Take previous index value if index at idx_year > year
+            if index_norm[idx_year - 1] - year > np.finfo(float).eps:
+                idx_year = idx_year -1
+            df_slice1 = data_slice.reindex(index_norm[:idx_year])
+            df_slice1 = self.interpolate_data(df_slice1,index_norm[:idx_year])
+            df_slice2 = data_slice.reindex(index_norm[idx_year:] - year)
+            df_slice2 = self.interpolate_data(df_slice2,index_norm[idx_year:] - year)
+            df_slice2.index = df_slice2.index + year
+            data_slice_reindexed = pd.concat([df_slice1,df_slice2])
+        else:
+            data_slice_reindexed = data_slice.reindex(index_norm)
+            data_slice_reindexed = self.interpolate_data(data_slice_reindexed,index_norm)
+        # Add starting year back to index desired by user
+        data_slice_reindexed.index = data_slice_reindexed.index + year_start
 
         if plot:
             if category is None:
@@ -504,6 +520,33 @@ def get_data_metadata(self):
 
         return data_metadata
 
+    def interpolate_data(self,df,index):
+        '''Interpolate testcase data.
+
+        Parameters
+        ----------
+        df: pandas.DataFrame
+            Dataframe that needs to be interpolated
+        index: np.array()
+            Index to use to get interpolated data
+
+        Returns
+        -------
+        df: pandas.DataFrame
+            Interpolated dataframe
+
+        '''
+        for key in df.keys():
+            # Use linear interpolation for continuous variables
+            if key in self.categories['weather']:
+                f = interpolate.interp1d(self.case.data.index,
+                    self.case.data[key], kind='linear')
+            # Use forward fill for discrete variables
+            else:
+                f = interpolate.interp1d(self.case.data.index,
+                    self.case.data[key], kind='zero')
+            df.loc[:,key] = f(index)
+        return df
 
 if __name__ == "__main__":
     import sys

docs/design/source/forecasts.rst

@@ -3,10 +3,13 @@
 Forecast Generation
 ===================
 
+Forecaster Module
+-----------------
+
 A forecaster module is developed to retrieve forecast from a BOPTEST
 testcase, which is needed for predictive controllers. This module uses the
-data_manager object of a test case to read the deterministic forecast from
-the testcase wrapped.fmu. In future developments it will be possible to
+:code:`data_manager` object of a test case to read the deterministic forecast from
+the testcase :code:`wrapped.fmu`. In future developments it will be possible to
 request stochastic forecast with a predefined distribution over the
 deterministic forecast for research purposes. This distribution will be
 added on the top of the deterministic forecast mentioned before.
@@ -15,3 +18,30 @@ The controller developer can choose the prediction horizon and interval of
 the forecast from the actual simulation time. The controller developer may
 also filter the forecast for a specific data category or request all data
 variables and filter it afterwards.
+
+Getting Weather Forecasts Across Year-End
+-----------------------------------------
+
+The data in TMY weather files used in BOPTEST test cases are discontinuous
+at the end of the year relative to the start of the year.
+Therefore, so is the weather data in the .csv files supplied for weather
+forecast generation.  As an example, see the relative humidity in the
+figure below (orange line). If weather forecasts are asked for that cross
+year-end, the :code:`data_manager` object used by the forcaster module splits the
+data at year-end into one portion that is inclusive of the last data point
+at the end of the year (midnight), and one portion after the end of the
+year that is not inclusive of the first data point at the start of the
+year (midnight). The implementation is done this way so that the forecast
+is more consistent for any interval through the full first year if a user
+only intends to simulate one year. The relative humidity plot below shows
+the interpolation behavior of the implementation graphically for forecast
+intervals of 1800s and 123s (intervals used in the unit tests), compared
+to the reference weather boundary condition data at 1800s intervals
+in the .csv file.
+
+.. figure:: images/relative_humidity_over_year_sim.png
+    :scale: 50 %
+
+    Forecast retrieved across one year for intervals of 1800s (blue) and
+    123s (yellow) compared to the reference data in the boundary condition
+    .csv file (orange), shown for relative humidity.

releasenotes.md

@@ -9,6 +9,8 @@ Released on xx/xx/xxxx.
 - Add materials for RLEM23 workshop at ``docs/workshops/RlemWorkshop_20231112``. This is for [#585](https://github.com/ibpsa/project1-boptest/issues/585).
 - Change JModelica docker container address  in ``testing/Dockerfile``. This is for [#590](https://github.com/ibpsa/project1-boptest/issues/590).
 - Specify the Python version (3.7) used for building the wrapper to execute the example JavaScript controllers in the unit test. This is for [#594](https://github.com/ibpsa/project1-boptest/issues/594).
+- Allow simulations and forecast to work across the end of the year to the next year. This is for [#239](https://github.com/ibpsa/project1-boptest/issues/239).
+- Pin base Docker image to ``linux/x86_64`` platform. This is for [#608](https://github.com/ibpsa/project1-boptest/issues/608).
 
 ## BOPTEST v0.5.0