Telework and other code changes

.dockerignore

@@ -1,3 +1,6 @@
 */data/*
 */output/*
 .git
+.omx
+.mmap
+.log

Dockerfile

@@ -6,20 +6,28 @@ ENV FULL_CONDA_PATH $CONDA_DIR/envs/$CONDA_ENV
 
 
 ENV ASIM_PATH /activitysim
-ENV ASIM_SUBDIR example
+ENV ASIM_SUBDIR examples
 ENV EXEC_NAME simulation.py
 
 RUN apt-get --allow-releaseinfo-change update \
 	&& apt-get install -y build-essential zip unzip
 RUN conda update conda --yes
 
-RUN git clone https://github.com/ual/activitysim.git
+RUN git clone -b telework https://github.com/LBNL-UCB-STI/activitysim.git
 
-RUN conda env create --quiet -p $FULL_CONDA_PATH --file activitysim/environment.yml
-RUN cd activitysim && $FULL_CONDA_PATH/bin/python setup.py install
+RUN conda install -n base conda-libmamba-solver
+
+RUN conda env create --quiet -p $FULL_CONDA_PATH --file activitysim/environment.yml --solver=libmamba
+
+RUN cd activitysim && git pull && $FULL_CONDA_PATH/bin/python setup.py install
 
 ENV PATH $FULL_CONDA_PATH/bin:$PATH
 ENV CONDA_DEFAULT_ENV $CONDA_ENV
 
-WORKDIR $ASIM_PATH/$ASIM_SUBDIR
+ENV EXAMPLE bay_area
+
+WORKDIR $ASIM_PATH/$EXAMPLE
+
+RUN git pull
+
 ENTRYPOINT ["python", "-u", "simulation.py"]

activitysim/abm/models/atwork_subtour_destination.py

@@ -136,7 +136,11 @@ def atwork_subtour_destination_logsums(
         chunk_size,
         trace_label)
 
-    destination_sample['mode_choice_logsum'] = logsums
+    if logsums.isna().sum() > 0:
+        logger.warning("Finding {0} logsums that are NaN. Filling with {0}".format(logsums.isna().sum(),
+                                                                                   logsums.loc[~logsums.isna()].min()))
+
+    destination_sample['mode_choice_logsum'] = logsums.fillna(logsums.loc[~logsums.isna()].min())
 
     return destination_sample
 
@@ -297,6 +301,11 @@ def atwork_subtour_destination(
         chunk_size=chunk_size,
         trace_label=tracing.extend_trace_label(trace_label, 'simulate'))
 
+    bad_choices = ~(choices_df.choice >= 0).values
+    if bad_choices.sum() > 0:
+        logger.warning("Somehow we still have {0} missing choices in non-mandatory, WTF?".format(bad_choices.sum()))
+        choices_df.loc[bad_choices, :] = choices_df.loc[~bad_choices, :].sample(bad_choices.sum(), replace=True)
+
     if estimator:
         estimator.write_choices(choices_df['choice'])
         choices_df['choice'] = estimator.get_survey_values(choices_df['choice'], 'tours', 'destination')

activitysim/abm/models/generate_beam_plans.py

@@ -7,7 +7,7 @@
 import logging
 import warnings
 
-from activitysim.core import pipeline
+from activitysim.core import pipeline, orca
 from activitysim.core import inject
 
 logger = logging.getLogger('activitysim')
@@ -141,12 +141,19 @@ def generate_beam_plans():
     tours = pipeline.get_table('tours')
     persons = pipeline.get_table('persons')
     households = pipeline.get_table('households')
-    land_use = pipeline.get_table('land_use').reset_index()
 
-    # re-create zones shapefile
-    land_use['geometry'] = land_use['geometry'].apply(wkt.loads)
-    zones = gpd.GeoDataFrame(land_use, geometry='geometry', crs='EPSG:4326')
-    zones.geometry = zones.geometry.buffer(0)
+    if orca.is_table('beam_geoms'):
+        beam_geoms = pipeline.get_table('beam_geoms')
+        beam_geoms['geometry'] = gpd.GeoSeries.from_wkt(beam_geoms['geometry'])
+        zones = gpd.GeoDataFrame(beam_geoms, geometry='geometry', crs='EPSG:4326')
+        zones.geometry = zones.geometry.buffer(0)
+    else:
+        land_use = pipeline.get_table('land_use').reset_index()
+
+        # re-create zones shapefile
+        land_use['geometry'] = land_use['geometry'].apply(wkt.loads)
+        zones = gpd.GeoDataFrame(land_use, geometry='geometry', crs='EPSG:4326')
+        zones.geometry = zones.geometry.buffer(0)
 
     # augment trips table with attrs we need to generate plans
     trips = get_trip_coords(trips, zones, persons)

activitysim/abm/models/initialize.py

@@ -86,6 +86,12 @@ def initialize_landuse():
 
     model_settings = config.read_model_settings('initialize_landuse.yaml', mandatory=True)
 
+    beam_geometries_path = config.setting('beam_geometries')
+    data_file_path = config.data_file_path(beam_geometries_path, mandatory=True)
+
+    beam_geom_dataframe = pd.read_csv(data_file_path)
+    pipeline.rewrap("beam_geoms", beam_geom_dataframe)
+
     annotate_tables(model_settings, trace_label)
 
     # create accessibility (only required if multiprocessing wants to slice accessibility)

activitysim/abm/models/joint_tour_destination.py

@@ -22,7 +22,6 @@
 from .util import logsums as logsum
 from .util import estimation
 
-
 logger = logging.getLogger(__name__)
 
 
@@ -35,7 +34,6 @@ def run_destination_sample(
         destination_size_terms,
         estimator,
         chunk_size, trace_label):
-
     spec = simulate.spec_for_segment(model_settings, spec_id='SAMPLE_SPEC',
                                      segment_name=spec_segment_name, estimator=estimator)
 
@@ -127,7 +125,11 @@ def run_destination_logsums(
         chunk_size,
         trace_label)
 
-    destination_sample['mode_choice_logsum'] = logsums
+    if logsums.isna().sum() > 0:
+        logger.warning("Finding {0} logsums that are NaN. Filling with {0}".format(logsums.isna().sum(),
+                                                                                   logsums.loc[~logsums.isna()].min()))
+
+    destination_sample['mode_choice_logsum'] = logsums.fillna(logsums.loc[~logsums.isna()].min())
 
     return destination_sample
 
@@ -197,6 +199,14 @@ def run_destination_simulate(
         trace_choice_name='destination',
         estimator=estimator)
 
+    n_bad_dests = choices.isna().sum().sum()
+
+    if n_bad_dests > 0:
+        logger.warning("Found {0} nan destination choices. Filling with random".format(n_bad_dests))
+        choices.loc[choices.isna().any(axis=1), :] = choices.loc[~choices.isna().any(axis=1), :].sample(
+            n_bad_dests,
+            replace=True).values
+
     return choices
 
 
@@ -211,7 +221,6 @@ def run_joint_tour_destination(
         skim_stack,
         estimator,
         chunk_size, trace_hh_id, trace_label):
-
     size_term_calculator = tour_destination.SizeTermCalculator(model_settings['SIZE_TERM_SELECTOR'])
 
     chooser_segment_column = model_settings['CHOOSER_SEGMENT_COLUMN_NAME']
@@ -319,7 +328,6 @@ def joint_tour_destination(
         skim_dict, skim_stack,
         chunk_size,
         trace_hh_id):
-
     """
     Given the tour generation from the above, each tour needs to have a
     destination, so in this case tours are the choosers (with the associated
@@ -380,6 +388,11 @@ def joint_tour_destination(
         estimator.end_estimation()
 
     # add column as we want joint_tours table for tracing.
+    bad_choices = ~(choices_df.choice >= 0).values
+    if bad_choices.sum() > 0:
+        logger.warning("Somehow we still have {0} missing choices, WTF?".format(bad_choices.sum()))
+        choices_df.loc[bad_choices, :] = choices_df.loc[~bad_choices, :].sample(bad_choices.sum(), replace=True)
+
     joint_tours['destination'] = choices_df.choice
     assign_in_place(tours, joint_tours[['destination']])
     pipeline.replace_table("tours", tours)

activitysim/abm/models/location_choice.py

@@ -221,7 +221,11 @@ def run_location_logsums(
     # when the index has duplicates, however, in the special case that the series index exactly
     # matches the table index, then the series value order is preserved
     # logsums now does, since workplace_location_sample was on left side of merge de-dup merge
-    location_sample_df['mode_choice_logsum'] = logsums
+    if logsums.isna().sum() > 0:
+        logger.warning("Finding {0} logsums that are NaN. Filling with {0}".format(logsums.isna().sum(),
+                                                                                   logsums.loc[~logsums.isna()].min()))
+
+    location_sample_df['mode_choice_logsum'] = logsums.fillna(logsums.loc[~logsums.isna()].min())
 
     return location_sample_df
 

activitysim/abm/models/non_mandatory_destination.py

@@ -76,6 +76,11 @@ def non_mandatory_tour_destination(
         estimator,
         chunk_size, trace_hh_id, trace_label)
 
+    bad_choices = ~(choices_df.choice >= 0).values
+    if bad_choices.sum() > 0:
+        logger.warning("Somehow we still have {0} missing choices in non-mandatory, WTF?".format(bad_choices.sum()))
+        choices_df.loc[bad_choices, :] = choices_df.loc[~bad_choices, :].sample(bad_choices.sum(), replace=True)
+
     if estimator:
         estimator.write_choices(choices_df.choice)
         choices_df.choice = estimator.get_survey_values(choices_df.choice, 'tours', 'destination')

activitysim/abm/models/util/tour_destination.py

@@ -17,7 +17,6 @@
 
 from activitysim.core.mem import force_garbage_collect
 
-
 from . import logsums as logsum
 from activitysim.abm.tables.size_terms import tour_destination_size_terms
 
@@ -33,7 +32,6 @@ class SizeTermCalculator(object):
     """
 
     def __init__(self, size_term_selector):
-
         # do this once so they can request siae_terms for various segments (tour_type or purpose)
         land_use = inject.get_table('land_use')
         size_terms = inject.get_injectable('size_terms')
@@ -66,7 +64,6 @@ def run_destination_sample(
         destination_size_terms,
         estimator,
         chunk_size, trace_label):
-
     model_spec = simulate.spec_for_segment(model_settings, spec_id='SAMPLE_SPEC',
                                            segment_name=spec_segment_name, estimator=estimator)
 
@@ -160,7 +157,11 @@ def run_destination_logsums(
         chunk_size,
         trace_label)
 
-    destination_sample['mode_choice_logsum'] = logsums
+    if logsums.isna().sum() > 0:
+        logger.warning("Finding {0} logsums that are NaN. Filling with {0}".format(logsums.isna().sum(),
+                                                                                   logsums.loc[~logsums.isna()].min()))
+
+    destination_sample['mode_choice_logsum'] = logsums.fillna(logsums.loc[~logsums.isna()].min())
 
     return destination_sample
 
@@ -252,7 +253,6 @@ def run_tour_destination(
         skim_stack,
         estimator,
         chunk_size, trace_hh_id, trace_label):
-
     size_term_calculator = SizeTermCalculator(model_settings['SIZE_TERM_SELECTOR'])
 
     chooser_segment_column = model_settings['CHOOSER_SEGMENT_COLUMN_NAME']

activitysim/abm/models/util/tour_frequency.py

@@ -54,7 +54,7 @@ def canonical_tours():
     atwork_subtour_flavors = {'eat': 1, 'business': 2, 'maint': 1}
     atwork_subtour_channels = enumerate_tour_types(atwork_subtour_flavors)
     max_work_tours = mandatory_tour_flavors['work']
-    atwork_subtour_channels = ['%s_%s' % (c, i+1)
+    atwork_subtour_channels = ['%s_%s' % (c, i + 1)
                                for c in atwork_subtour_channels
                                for i in range(max_work_tours)]
 
@@ -341,6 +341,24 @@ def process_mandatory_tours(persons, mandatory_tour_frequency_alts):
 
     tours['household_id'] = tours_merged.household_id
 
+    bad_work_dest = ~(tours.loc[tours_merged.tour_type == 'work', "destination"] >= 0)
+
+    if bad_work_dest.sum() > 0:
+        logger.warning(
+            "At the start we have {0} bad work tour destinations, replacing them by sampling from {1} good ones".format(
+                bad_work_dest.sum(), (~bad_work_dest).sum()))
+        tours.loc[tours_merged.tour_type == 'work', "destination"].loc[bad_work_dest] = \
+            tours.loc[tours_merged.tour_type == 'work', "destination"].loc[~bad_work_dest].sample(
+                bad_work_dest.sum(), replace=True).values
+
+    bad_school_dest = ~(tours.loc[tours_merged.tour_type == 'school', "destination"] >= 0)
+
+    if bad_school_dest.sum() > 0:
+        logger.warning("At the start we have {0} bad school tour destinations".format(bad_school_dest.sum()))
+        tours.loc[tours_merged.tour_type == 'school', "destination"].loc[bad_school_dest] = \
+            tours.loc[tours_merged.tour_type == 'school', "destination"].loc[~bad_school_dest].sample(
+                bad_school_dest.sum(), replace=True).values
+
     # assign stable (predictable) tour_id
     set_tour_index(tours)
 
@@ -415,7 +433,6 @@ def process_non_mandatory_tours(persons, tour_counts):
 
 
 def process_atwork_subtours(work_tours, atwork_subtour_frequency_alts):
-
     """
     This method processes the atwork_subtour_frequency column that comes
     out of the model of the same name and turns into a DataFrame that

activitysim/abm/models/util/vectorize_tour_scheduling.py

@@ -371,6 +371,11 @@ def _schedule_tours(
     )
 
     # - update previous_tour and timetable parameters
+    if choices.min() < 0:
+        logger.warning("INDEX BELOW 0")
+    choices[~(choices > 0)] = 0
+    if choices.min() < 0:
+        logger.warning("INDEX STIIIIILL BELOW 0")
 
     # update previous_tour (series with most recent previous tdd choices) with latest values
     previous_tour.loc[tours[tour_owner_id_col]] = choices.values

activitysim/cli/run.py

@@ -59,10 +59,10 @@ def add_run_args(parser, multiprocess=True):
     parser.add_argument("-u", "--skims_url", 
                         action="store", 
                         help="url of skims .csv")
-    parser.add_argument( "-x", "--path_to_remote_data", 
+    parser.add_argument("-x", "--path_to_remote_data",
                         action="store",
                         help="url of urbansim .h5 model data")
-    parser.add_argument( "t", "--write_to_s3", 
+    parser.add_argument("-t", "--write_to_s3",
                         action="store_true", 
                         help="write output to s3?")
 

activitysim/core/config.py

@@ -57,9 +57,9 @@ def pipeline_file_name(settings):
     return pipeline_file_name
 
 
-@inject.injectable()
-def rng_base_seed():
-    return 0
+@inject.injectable(cache=True)
+def rng_base_seed(settings):
+    return settings.get('rng_base_seed', None)
 
 
 @inject.injectable(cache=True)