Allow multiple Lawrencium runs

hpc/job.sh

@@ -5,5 +5,9 @@ module load python/3.8.8
 #pip install --user geopandas
 export PYTHONPATH=`python -m site --user-site`:$PYTHONPATH
 cd /global/scratch/users/$USER/sources/PILATES
-python run.py -v
+echo "$1"
+
+
+
+python run.py -c "$1" -S "$2"
 

hpc/job_runner.sh

@@ -1,13 +1,29 @@
 #!/bin/bash
 
+
 RANDOM_PART="$(tr -dc A-Z0-9 </dev/urandom | head -c 8)"
 DATETIME="$(date "+%Y.%m.%d-%H.%M.%S")"
 JOB_NAME="$RANDOM_PART.$DATETIME"
 
+settings_file="settings.yaml"
+stage_file="current_stage_${JOB_NAME}.yaml"
+while getopts :c:s: name
+do
+    case $name in
+    c)    cflag=1
+          settings_file="$OPTARG";;
+    s)    sflag=1
+          stage_file="$OPTARG";;
+    ?)   printf "Usage: %s: [-c settings/config file loc] [-s stage file loc] args\n" $0
+          exit 2;;
+    esac
+done
+
+
 PARTITION="lr7"
 QOS="lr_normal"
 NUM_CPUS=50
-MEMORY_LIMIT="240"
+MEMORY_LIMIT="250"
 ACCOUNT="pc_beamcore"
 JOB_LOG_FILE_PATH="/global/scratch/users/$USER/pilates_logs/log_${DATETIME}_${RANDOM_PART}.log"
 EXPECTED_EXECUTION_DURATION="2-00:00:00"
@@ -23,6 +39,6 @@ sbatch --partition="$PARTITION" \
     --job-name="$JOB_NAME" \
     --output="$JOB_LOG_FILE_PATH" \
     --time="$EXPECTED_EXECUTION_DURATION" \
-    job.sh
+    job.sh "$settings_file" "$stage_file"
 
 echo $JOB_LOG_FILE_PATH

lawrencium-setup.md

@@ -4,16 +4,21 @@
 cd /global/scratch/users/$USER
 mkdir sources
 cd sources
-git clone --branch atlas-v2 https://github.com/LBNL-UCB-STI/PILATES.git
+git clone --branch copy-woorking-directory https://github.com/LBNL-UCB-STI/PILATES.git
 cd PILATES
 ```
 
 ## Setup Python
 
 ```commandline
-module load python/3.8.8
-pip install --user openmatrix
-pip install --user geopandas
+module load python/3.10.12
+python -m pip uninstall --user shapely
+python -m pip install --user shapely
+python -m pip install --user openmatrix
+python -m pip install --user pygeos
+python -m pip install --user geopandas
+python -m pip install --user table
+python -m pip install --user PyYAML
 export PYTHONPATH=`python -m site --user-site`:$PYTHONPATH
 ```
 
@@ -85,4 +90,7 @@ cd ../../../
 cd hpc
 ./job_runner.sh
 ```
+#### Optional tags:
+`-c`: Define a new settings file, defaults to `settings.yaml`
 
+`-s`: Current stage file name of run to restart in the middle of. If none, starts a new run with a new current stage file. Defaults to none.

pilates/activitysim/postprocessor.py

@@ -4,19 +4,19 @@
 import os
 
 from pilates.utils.io import read_datastore
+from workflow_state import WorkflowState
 
 logger = logging.getLogger(__name__)
 
 
-def _load_asim_outputs(settings):
+def _load_asim_outputs(settings, output_path):
     output_tables_settings = settings['asim_output_tables']
     prefix = output_tables_settings['prefix']
     output_tables = output_tables_settings['tables']
     asim_output_dict = {}
     for table_name in output_tables:
         file_name = "%s%s.csv" % (prefix, table_name)
-        file_path = os.path.join(
-            settings['asim_local_output_folder'], file_name)
+        file_path = os.path.join(output_path, settings['asim_local_output_folder'], file_name)
         if table_name == 'persons':
             index_col = 'person_id'
         elif table_name == 'households':
@@ -50,18 +50,18 @@ def get_usim_datastore_fname(settings, io, year=None):
 
 
 def _prepare_updated_tables(
-        settings, forecast_year, asim_output_dict, tables_updated_by_asim,
+        settings, state: WorkflowState, asim_output_dict, tables_updated_by_asim,
         prefix=None):
     """
     Combines ActivitySim and UrbanSim outputs for tables updated by
     ActivitySim (e.g. households and persons)
     """
 
-    data_dir = settings['usim_local_data_folder']
+    data_dir = os.path.join(state.full_path, settings['usim_local_mutable_data_folder'])
 
     # e.g. model_data_2012.h5
     usim_output_store_name = get_usim_datastore_fname(
-        settings, io='output', year=forecast_year)
+        settings, io='output', year=state.forecast_year)
     usim_output_store_path = os.path.join(data_dir, usim_output_store_name)
     if not os.path.exists(usim_output_store_path):
         raise ValueError('No output data store found at {0}'.format(
@@ -172,7 +172,7 @@ def create_beam_input_data(settings, forecast_year, asim_output_dict):
 
 def create_usim_input_data(
         settings, input_year, forecast_year, asim_output_dict,
-        tables_updated_by_asim):
+        tables_updated_by_asim, full_path=None):
     """
     Creates UrbanSim input data for the next iteration.
 
@@ -186,7 +186,9 @@ def create_usim_input_data(
     """
 
     # parse settings
-    data_dir = settings['usim_local_data_folder']
+    data_dir = settings['usim_local_mutable_data_folder']
+    if full_path is not None:
+        data_dir = os.path.join(full_path, data_dir)
 
     # Move UrbanSim input store (e.g. custom_mpo_193482435_model_data.h5)
     # to archive (e.g. input_data_for_2015_outputs.h5) because otherwise
@@ -215,7 +217,7 @@ def create_usim_input_data(
     if not os.path.exists(usim_output_store_path):
         raise ValueError('No output data found at {0}.'.format(
             usim_output_store_path))
-    usim_output_store, table_prefix_year = read_datastore(settings, forecast_year)
+    usim_output_store, table_prefix_year = read_datastore(settings, forecast_year, mutable_data_dir=full_path)
 
     logger.info(
         'Merging results back into UrbanSim format and storing as .h5!')
@@ -270,18 +272,19 @@ def create_usim_input_data(
     return
 
 
-def create_next_iter_inputs(settings, year, forecast_year):
+def create_next_iter_inputs(settings, year, state: WorkflowState):
+    forecast_year = state.forecast_year
     tables_updated_by_asim = ['households', 'persons']
-    asim_output_dict = _load_asim_outputs(settings)
+    asim_output_dict = _load_asim_outputs(settings, state.full_path)
     asim_output_dict = _prepare_updated_tables(
-        settings, forecast_year, asim_output_dict, tables_updated_by_asim,
+        settings, state, asim_output_dict, tables_updated_by_asim,
         prefix=forecast_year)
 
     # if settings['traffic_assignment_enabled']:
     #     create_beam_input_data(settings, forecast_year, asim_output_dict)
     create_usim_input_data(
         settings, year, forecast_year, asim_output_dict,
-        tables_updated_by_asim)
+        tables_updated_by_asim, state.full_path)
 
     return
 
@@ -294,7 +297,7 @@ def update_usim_inputs_after_warm_start(
 
     # load usim data
     if not usim_data_dir:
-        usim_data_dir = settings['usim_local_data_folder']
+        usim_data_dir = settings['usim_local_mutable_data_folder']
     datastore_name = get_usim_datastore_fname(settings, io='input')
     input_store_path = os.path.join(usim_data_dir, datastore_name)
     if not os.path.exists(input_store_path):