grading-on-a-curve/code/learning_curves/grid_search.sh

#!/bin/bash

CODE_DIR="$1"
DATA_DIR="$2"
ANNOTATED_DIR="$3"
REPO_DIR="$4"
EXP_DIR="$5"
PLOT_DIR="$6"
RUST_PROJECTS="$7"
GRAD_DESCENT_DIR="$DATA_DIR/gradient_descent/"
mkdir -p "$GRAD_DESCENT_DIR"

# assumes:
# "$ANNOTATED_DIR/rust-blame.csv"
# "$ANNOTATED_DIR/relevant-dirs.csv"
# "$REPO_DIR/firefox"
# "$EXP_DIR/firefox"
# "$REPO_DIR/{$RUST_PROJECTS}"
# "$EXP_DIR/{$RUST_PROJECTS}"

t1_guesses="0.000001:0.00001:0.0001:0.001:0.01:0.1:0.2:0.4:0.8:0.99"
l_guesses_1="0.0:0.1:0.2:0.3:0.4:0.5"
l_guesses_2="-0.1:-0.2:-0.3:-0.4:-0.5"

rust_repos=$(for project in $RUST_PROJECTS ; do
                 echo -n "$REPO_DIR/$project/:"
             done | head -c -1)
rust_exps=$(for project in $RUST_PROJECTS ; do
                echo -n "$EXP_DIR/$project/:"
            done | head -c -1)
rust_relevant=$(for project in $RUST_PROJECTS ; do
                    echo -n "$(grep "^$project," "$ANNOTATED_DIR/relevant-dirs.csv" | cut -d, -f2):"
                done | head -c -1)
# mozilla is kind enough to have no files with spaces in names in gecko-dev
c_relevant="$(tr '\n' ' ' <"$ANNOTATED_DIR/relevant-c++")"

echo "running gradient descent (this will take a while)..."

(
    python3 "$CODE_DIR/learning_curves/learningCurve_gradientDescent.py" \
            "$ANNOTATED_DIR/rust-blame.csv" \
            "$rust_repos" \
            "$rust_relevant" \
            "$rust_exps" \
            "$t1_guesses" \
            "$l_guesses_1" > "$GRAD_DESCENT_DIR/grid_search.rust.1.cuml.txt" &&
        echo "thread1 rust complete"
) &

(
    python3 "$CODE_DIR/learning_curves/learningCurve_gradientDescent.py" \
            "$ANNOTATED_DIR/rust-blame.csv" \
            "$rust_repos" \
            "$rust_relevant" \
            "$rust_exps" \
            "$t1_guesses" \
            "$l_guesses_2" > "$GRAD_DESCENT_DIR/grid_search.rust.2.cuml.txt" &&
        echo "thread2 rust complete"
) &

if [ $(nproc) -lt 4 ] ; then wait ; fi

(
    python3 "$CODE_DIR/learning_curves/learningCurve_gradientDescent.py" \
            "$ANNOTATED_DIR/c++-blame.csv" \
            "$REPO_DIR/firefox/" \
            "$c_relevant" \
            "$EXP_DIR/firefox/" \
            "$t1_guesses" \
            "$l_guesses_1" > "$GRAD_DESCENT_DIR/grid_search.c.1.cuml.txt" &&
        echo "thread1 c complete"
) &

(
    python3 "$CODE_DIR/learning_curves/learningCurve_gradientDescent.py" \
            "$ANNOTATED_DIR/c++-blame.csv" \
            "$REPO_DIR/firefox/" \
            "$c_relevant" \
            "$EXP_DIR/firefox/" \
            "$t1_guesses" \
            "$l_guesses_2" > "$GRAD_DESCENT_DIR/grid_search.c.2.cuml.txt" &&
        echo "thread2 c complete"
) &

wait
echo "processing complete"

echo
cat "$GRAD_DESCENT_DIR/grid_search.rust.1.cuml.txt" "$GRAD_DESCENT_DIR/grid_search.rust.2.cuml.txt" > "$GRAD_DESCENT_DIR/grid_search.rust.cuml.txt"
cat "$GRAD_DESCENT_DIR/grid_search.c.1.cuml.txt" "$GRAD_DESCENT_DIR/grid_search.c.2.cuml.txt" > "$GRAD_DESCENT_DIR/grid_search.c.cuml.txt"
rm "$GRAD_DESCENT_DIR/grid_search.rust.1.cuml.txt" "$GRAD_DESCENT_DIR/grid_search.rust.2.cuml.txt" "$GRAD_DESCENT_DIR/grid_search.c.1.cuml.txt" "$GRAD_DESCENT_DIR/grid_search.c.2.cuml.txt"

function get_t1_l() {
    bias="$1"
    file="$2"
    best=$(grep "^$bias" "$file" | sort -nk 8 | head -1)
    t1=$(echo $best | cut -f4 -d' ')
    l=$(echo $best | cut -f5 -d' ')
    echo $t1 $l
}

c_t1_l="$(get_t1_l 0 "$GRAD_DESCENT_DIR/grid_search.c.cuml.txt")"
c_t1_l_err_low="$(get_t1_l -1 "$GRAD_DESCENT_DIR/grid_search.c.cuml.txt")"
c_t1_l_err_up="$(get_t1_l 1 "$GRAD_DESCENT_DIR/grid_search.c.cuml.txt")"

c_t1="$(echo $c_t1_l | cut -f1 -d' ')"
c_l="$(echo $c_t1_l | cut -f2 -d' ')"
c_t1_err_low="$(echo $c_t1_l_err_low | cut -f1 -d' ')"
c_l_err_low="$(echo $c_t1_l_err_low | cut -f2 -d' ')"
c_t1_err_up="$(echo $c_t1_l_err_up | cut -f1 -d' ')"
c_l_err_up="$(echo $c_t1_l_err_up | cut -f2 -d' ')"

echo "C++ T1: $c_t1 ($c_t1_err_low, $c_t1_err_up)"
echo "C++ l: $c_l ($c_l_err_low, $c_l_err_up)"

rust_t1_l="$(get_t1_l 0 "$GRAD_DESCENT_DIR/grid_search.rust.cuml.txt")"
rust_t1_l_err_low="$(get_t1_l -1 "$GRAD_DESCENT_DIR/grid_search.rust.cuml.txt")"
rust_t1_l_err_up="$(get_t1_l 1 "$GRAD_DESCENT_DIR/grid_search.rust.cuml.txt")"

rust_t1="$(echo $rust_t1_l | cut -f1 -d' ')"
rust_l="$(echo $rust_t1_l | cut -f2 -d' ')"
rust_t1_err_low="$(echo $rust_t1_l_err_low | cut -f1 -d' ')"
rust_l_err_low="$(echo $rust_t1_l_err_low | cut -f2 -d' ')"
rust_t1_err_up="$(echo $rust_t1_l_err_up | cut -f1 -d' ')"
rust_l_err_up="$(echo $rust_t1_l_err_up | cut -f2 -d' ')"

echo "Rust T1: $rust_t1 ($rust_t1_err_low, $rust_t1_err_up)"
echo "Rust l: $rust_l ($rust_l_err_low, $rust_l_err_up)"

echo
echo "plotting data..."
python3 "$CODE_DIR/learning_curves/model-vs-real.py" \
        "$ANNOTATED_DIR/c++-blame.csv" \
        "$REPO_DIR/firefox/" \
        "$c_relevant" \
        "$EXP_DIR/firefox/" \
        "$PLOT_DIR/C++.pdf" \
        $c_t1 $c_l $c_t1_err_low $c_l_err_low $c_t1_err_up $c_l_err_up &

python3 "$CODE_DIR/learning_curves/model-vs-real.py" \
        "$ANNOTATED_DIR/rust-blame.csv" \
        "$rust_repos" \
        "$rust_relevant" \
        "$rust_exps" \
        "$PLOT_DIR/Rust.pdf" \
        $rust_t1 $rust_l $rust_t1_err_low $rust_l_err_low \
        $rust_t1_err_up $rust_l_err_up &

cpp_tex='C\nolinebreak\hspace{-.05em}\raisebox{.4ex}{\relsize{-3}{\textbf{+}}}\nolinebreak\hspace{-.10em}\raisebox{.4ex}{\relsize{-3}{\textbf{+}}}\xspace'

python3 "$CODE_DIR/learning_curves/model-vs-model.py" \
        "$PLOT_DIR/C++-vs-Rust.pdf" \
        "$cpp_tex" \
        "$c_t1" "$c_l" "$c_t1_err_low" "$c_l_err_low" \
        "$c_t1_err_up" "$c_l_err_up" \
        "Rust" "$rust_t1" "$rust_l" "$rust_t1_err_low" "$rust_l_err_low" \
        "$rust_t1_err_up" "$rust_l_err_up" \
        200

python3 "$CODE_DIR/learning_curves/plot-experience.py" \
        "$REPO_DIR/firefox/" \
        "$c_relevant" \
        "$EXP_DIR/firefox/" \
        "$cpp_tex" \
        "$rust_repos" \
        "$rust_relevant" \
        "$rust_exps" \
        "Rust" \
        "$PLOT_DIR/experience.frac.pdf" \
        "$PLOT_DIR/experience.tot.pdf"

wait
echo "plotting complete"

echo "calculating number of 0-exp. VCCs to flip sign of Rust l..."
python3 "$CODE_DIR/learning_curves/learningCurve_gradientDescent.py" \
        "$ANNOTATED_DIR/rust-blame.csv" \
        "$rust_repos" \
        "$rust_relevant" \
        "$rust_exps" \
        "$t1_guesses" \
        "-0.3:-0.2:-0.1:0:0.1:0.2:0.3" \
        "search"