Add example file

examples/example_run.sh

@@ -0,0 +1,65 @@
+#!/bin/bash
+
+#1) In the first run we do not have a model
+if [ $# -ne 2 ]; then
+  echo "Wrong command line. Please provide:"
+  echo "$0 'TEMP DIR' 'DB DIR'"
+  exit
+fi
+
+SRC=$1
+db_path=$2
+
+#We start clean
+mkdir -p ${SRC}
+rm -r ${db_path}/*
+
+# Create an AMS database and the respective file-system support
+python -m ams_wf.AMSStore create --path ${db_path} --name example
+
+for i in {1..3}; do
+
+  model=$(python -m ams_wf.AMSStore query --path ${db_path} -e models -f uri --version latest)
+
+  echo "Current Model file is : ${model}"
+
+  # We need to clean the temp files produced by the previous step
+  rm -rf ${SRC}/*
+
+  app_args=(
+    -dt hdf5
+    -db ${SRC}
+    -d cpu
+  )
+
+  if [ ! -z ${model} ]; then
+    app_args+=(-S)
+    app_args+=(${model})
+  fi
+
+  echo "Arguments are: ${app_args[@]}"
+  @CMAKE_CURRENT_BINARY_DIR@/ams_example ${app_args[@]}
+
+  #2) We need to run the DBStage mechanism to pick up the intermediate data and move them into the kosh-store
+  args=(
+    -db ${db_path}
+  #user options to use random pruner at the staging level
+    --load @CMAKE_CURRENT_BINARY_DIR@/prune.py --class RandomPruneAction -f 0.001
+  #Options to store the results in the candidates directory using a file format of h5
+    --policy process -m fs  --dest ${db_path}/candidates --db-type hdf5
+  #Where to pick data from
+    --src ${SRC} --pattern "*.h5"
+  #Make results public to store
+  --store
+  )
+
+# Copy data from local storage, apply user defined trimming and move to candidates
+  python -m ams_wf.AMSDBStage ${args[@]}
+
+# Sub-select candidate data, store them into storage.
+  python @CMAKE_CURRENT_BINARY_DIR@/sub_selection.py -db ${db_path}
+
+#Train a new model and push it into the store
+CUBLAS_WORKSPACE_CONFIG=:4096:8 python @CMAKE_CURRENT_BINARY_DIR@/train.py -db ${db_path}
+
+done