Bash script

There is a .sh script to perform several runs at the same time with the same simulation configuration. It is based on a script developed by Pedro Arce (author of GAMOS) [1] but with modifications based on our simulation.

Ther are some important notes about this script.

• The variable NCORES must be changed based on your computer. It is the number of cores used on the simulation, if your computer have 30 cores, NCORES must not be 30 or more.

• The path where the .out files will be saved must be rewritten on each run usign the variable OUTPUT_PATH.

• Once the script has been modified, you can run it as follows:

bash sendrun.sh NRUNS SEED

Where NRUNS is the number of runs we want to perform and SEED will be the initial seed of the main_0 generated file.

Note

Important!!! Modify the output path and make sure the folder exists.

#!/bin/bash

# ----------- Argument validation ----------- #
NRUNS=$1
SEED=$2
NCORES=20

if [[ -z "$NRUNS" || ! "$NRUNS" =~ ^[0-9]+$ ]]; then
    echo "Usage: $0 <NRUNS> [SEED]"
    echo "NRUNS must be a positive integer."
    exit 1
fi

# ----------- Generate random seed if not provided ----------- #
if [[ -z "$SEED" ]]; then
    SEED=$(date +%s%N | cut -b1-9)  # Nano timestamp, first 9 digits
    echo "[INFO] No seed provided. Generated random seed: $SEED"
fi

# ----------- Define output folder using timestamp and seed ----------- #
TIMESTAMP=$(date +"%Y%m%d_%H%M%S")
OUTPUT_PATH="output/run_${TIMESTAMP}_seed${SEED}/"

echo "[INFO] Creating output directories..."
mkdir -p "$OUTPUT_PATH" "$OUTPUT_PATH/main_files" "$OUTPUT_PATH/inputs_tmp" "$OUTPUT_PATH/logs"

inputs_tmp="$OUTPUT_PATH/inputs_tmp"
main_files="$OUTPUT_PATH/main_files"
logs="$OUTPUT_PATH/logs"

# ----------- Save execution parameters ----------- #
{
    echo "NRUNS=$NRUNS"
    echo "SEED=$SEED"
    echo "TIMESTAMP=$TIMESTAMP"
    echo "NCORES=$NCORES"
} > "$OUTPUT_PATH/run_info.txt"

# ----------- Determine max concurrent jobs ----------- #
MAX_JOBS=$(( NRUNS < NCORES ? NRUNS : NCORES ))

echo "[INFO] Running $NRUNS simulations with up to $MAX_JOBS concurrent jobs..."
echo "[INFO] Output will be saved in: $OUTPUT_PATH"

# ----------- Launch simulation jobs ----------- #
for ((nj=0; nj<NRUNS; nj++)); do
    SUFFIX=$nj
    echo "[INFO] Launching job $nj with suffix $SUFFIX..."

    # Create customized GetImage_$SUFFIX.in
    awk -v SUFFIX=$SUFFIX -v OUTPUT_PATH=$OUTPUT_PATH '{
        if ($2=="GetImage:ResultsFolder") {
            printf("/gamos/setParam GetImage:ResultsFolder %s\n", OUTPUT_PATH)
        } else if ($2=="GetImage:OutputFilename") {
            printf("/gamos/setParam GetImage:OutputFilename image_%s\n", SUFFIX)
        } else {
            print $0
        }
    }' inputs/GetImage.in > "$inputs_tmp/GetImage_$SUFFIX.in"

    # Compute unique seeds
    seed1=$((SEED + 2 * nj))
    seed2=$((SEED + 2 * nj + 1))

    # Create customized main_$SUFFIX.inn
    awk -v SUFFIX=$SUFFIX -v s1=$seed1 -v s2=$seed2 -v TMP_INPUTS=$inputs_tmp '{
        if ($2=="inputs/GetImage.in") {
            printf("/control/execute %s/GetImage_%s.in\n", TMP_INPUTS, SUFFIX)
        } else if ($1=="/gamos/random/setSeeds") {
            printf("/gamos/random/setSeeds %s %s\n", s1, s2)
        } else {
            print $0
        }
    }' main.in > "$main_files/main_$SUFFIX.inn"

    # Run simulation and log output
    gamos "$main_files/main_$SUFFIX.inn" 2>&1 | tee "$logs/log_main_$SUFFIX.log" &

    # Control number of concurrent jobs
    if [[ $(jobs -r -p | wc -l) -ge $MAX_JOBS ]]; then
        wait -n
    fi
done

wait

echo "[INFO] All simulations have completed."
echo "[INFO] Output directory: $OUTPUT_PATH"

# ----------- Prompt for cleanup ----------- #
echo ""
read -p "Do you want to delete the temporary files ($inputs_tmp and $main_files)? [y/N]: " answer

if [[ "$answer" == "y" || "$answer" == "Y" ]]; then
    echo "[INFO] Deleting temporary files..."
    rm -rf "$inputs_tmp" "$main_files"
    echo "[INFO] Temporary files deleted."
else
    echo "[INFO] Temporary files kept."
fi