Refactoring the 'replicate-results' into a makefile

bin/measure-scaling-performance.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+
+# # # # #
+#
+# Function to call the requisite Haskell and Python binary/script (respectively)
+# which will generate the results data and figures presented in the paper.
+#
+# # # # #
+
+function using()
+{
+    ./bin/generate-timings $5 \
+        --data "data-sets/$1.afasta" \
+        --tree "data-sets/$1.tree" \
+        --tcm  "data-sets/tcm-${2}.tcm" \
+        --output "$1-$2" \
+        -n "$3" \
+        -k "$4"
+
+    python3 ./bin/plot-figure.py \
+        "replicate-results/csv/${1}-${2}.preorder.csv" \
+        "replicate-results/img/${1}-preorder.eps"
+
+    python3 ./bin/plot-figure.py \
+        "replicate-results/csv/${1}-${2}.postorder.csv" \
+        "replicate-results/img/${1}-postorder.eps"
+}
+
+"$@"

plot-figure.py → bin/plot-figure.py

@@ -1,3 +1,5 @@
+#!/bin/python3
+
 from mpl_toolkits import mplot3d
 import math
 import matplotlib.pyplot as plt

cabal.project

@@ -1,9 +1,8 @@
-allow-newer:
-    base,
-    Cabal,
-    ghc-prim,
-    template-haskell
+installdir: ./bin
+
+install-method: copy
 
 packages: .
 
 with-compiler: ghc-9.2.1
+

makefile

@@ -0,0 +1,69 @@
+# # # # #
+#
+# Script to reproduce the results of the paper from a clean environment.
+# Installs all required dependencies. Builds the required binaries.
+# Run the experimental algorithm on datasets to generate the empirical results.
+#
+# # # # #
+
+bin-dir          := ./bin
+data-dir         := replicate-results
+executable       := $(bin-dir)/generate-timings
+generate         := $(bin-dir)/measure-scaling-performance.sh using
+biological-scale := [1%1,1%2,1%4,1%8,1%16,1%32]
+customized-scale := [1%1,1%2,1%4,1%8,1%16,1%32,1%64]
+customized-nodes := [4,8,16,32,64,128,256]
+
+
+# All synonyms for replicating the paper's results.
+all: results
+
+replicate: results
+
+reproduce: results
+
+results: fungi metazoa pathological
+
+
+# Install dependencies required to replicate results:
+ensure-Haskell: $(bin-dir)/generate-timings
+	@command -v ghcup &> /dev/null || curl --proto '=https' --tlsv1.2 -sSf https://get-ghcup.haskell.org | sh
+
+ensure-Python:
+	@command -v python3    &> /dev/null || apt-get install python3     --yes
+	@command -v pip3       &> /dev/null || apt-get install python3-pip --yes
+	@pip3  show matplotlib &> /dev/null || pip3 install --upgrade matplotlib
+
+ensure-R:
+	command -v &> /dev/null || sudo apt install r-base
+
+compile-binaries: ensure-Haskell
+	@echo "Copiling binaries"
+	@ghcup run \
+	    --ghc 9.2.1 \
+	    --cabal 3.6.2.0 \
+	    -- cabal update && \
+	       cabal install \
+	        --installdir=$(bin-dir) \
+	        --install-method=copy 
+
+ensure-workspace:
+	@mkdir -p $(data-dir)/csv
+	@mkdir -p $(data-dir)/data
+	@mkdir -p $(data-dir)/img
+	@mkdir -p $(data-dir)/taxa
+	@mkdir -p $(data-dir)/tree
+
+fungi: compile-binaries ensure-Python ensure-R ensure-workspace
+	$(generate) \
+	    'fungi'        '11' '[25,50,100,200,400,800,1553]' $(biological-scale)
+
+metazoa: compile-binaries ensure-Python ensure-R ensure-workspace
+	$(generate) \
+	    'metazoa'      '11' '[25,50,100,200,400,800,1766]' $(biological-scale)
+
+pathological: compile-binaries ensure-Python ensure-R ensure-workspace
+	$(generate) \
+	    'pathological' '12' $(customized-nodes) $(customized-scale)
+	$(generate) \
+	    'pathological' '31' $(customized-nodes) $(customized-scale) '--no-generate'

replicate-results.sh

@@ -1,65 +1,67 @@
 #!/bin/bash
-#Script to run implied alingment data sets
 
-#load Haskell dependencies
-which stack || curl -sSL https://get.haskellstack.org/ | sh
 
-#load Python3 dependencies
-which python3        || apt-get install python3     --yes
-which pip3           || apt-get install python3-pip --yes
-pip3 show matplotlib || pip3 install --upgrade matplotlib
+# # # # #
+#
+# Script to reproduce the results of the paper from a clean environment.
+# Installs all required dependencies. Builds the required binaries.
+# Run the experimental algorithm on datasets to generate the empirical results.
+#
+# # # # #
 
-#load R dependencies
-which R || sudo apt install r-base
 
-#build binaries
-stack install
+# Install dependencies required to replicate results:
+#   * Load Python3 dependencies
+which python3         &> /dev/null || apt-get install python3     --yes
+which pip3            &> /dev/null || apt-get install python3-pip --yes
+pip3  show matplotlib &> /dev/null || pip3 install --upgrade matplotlib
 
-#ensure target directories exist
+#   * Load R dependencies
+which R &> /dev/null || sudo apt install r-base
+
+#   * Load Haskell dependencies
+which ghcup &> /dev/null || curl --proto '=https' --tlsv1.2 -sSf https://get-ghcup.haskell.org | sh
+
+
+# Build binaries
+ghcup run \
+    --ghc 9.2.1 \
+    --cabal 3.6.2.0 \
+    -- cabal update && \
+       cabal install \
+        --installdir=./bin \
+        --install-method=copy 
+
+
+# Ensure output directories exist to place results data
 mkdir -p replicate-results/csv
 mkdir -p replicate-results/data
 mkdir -p replicate-results/img
 mkdir -p replicate-results/taxa
 mkdir -p replicate-results/tree
 
-#run data sets
-./bin/generate-timings \
-    --data data-sets/fungi.afasta \
-    --tree data-sets/fungi.tree \
-    --tcm  data-sets/tcm-11.tcm \
-    --output fungi-11 \
-    -n [25,50,100,200,400,800,1553] \
-    -k [1%1,1%2,1%4,1%8,1%16,1%32]
-python3 plot-figure.py replicate-results/csv/fungi-11.preorder.csv  replicate-results/img/fungi-preorder.eps
-python3 plot-figure.py replicate-results/csv/fungi-11.postorder.csv replicate-results/img/fungi-postorder.eps
 
-./bin/generate-timings \
-    --data data-sets/metazoa.afasta \
-    --tree data-sets/metazoa.tree \
-    --tcm  data-sets/tcm-11.tcm \
-    --output metazoa-11 \
-    -n [25,50,100,200,400,800,1766] \
-    -k [1%1,1%2,1%4,1%8,1%16,1%32]
-python3 plot-figure.py replicate-results/csv/metazoa-11.preorder.csv  replicate-results/img/metazoa-preorder.eps
-python3 plot-figure.py replicate-results/csv/metazoa-11.postorder.csv replicate-results/img/metazoa-postorder.eps
+# Run data sets to produce the results data
+function generate()
+{
+    ./bin/generate-timings $5 \
+        --data "data-sets/$1.afasta" \
+        --tree "data-sets/$1.tree" \
+        --tcm  "data-sets/tcm-${2}.tcm" \
+        --output "$1-$2" \
+        -n "$3" \
+        -k "$4"
+
+    python3 plot-figure.py \
+        "replicate-results/csv/${1}-${2}.preorder.csv" \
+        "replicate-results/img/${1}-preorder.eps"
 
-./bin/generate-timings \
-    --data data-sets/pathological.afasta \
-    --tree data-sets/pathological.tree \
-    --tcm  data-sets/tcm-12.tcm \
-    --output pathological-12 \
-    -n [4,8,16,32,64,128,256] \
-    -k [1%1,1%2,1%4,1%8,1%16,1%32,1%64]
-python3 plot-figure.py replicate-results/csv/pathological-12.preorder.csv  replicate-results/img/pathological-12-preorder.eps
-python3 plot-figure.py replicate-results/csv/pathological-12.postorder.csv replicate-results/img/pathological-12-postorder.eps
+    python3 plot-figure.py \
+        "replicate-results/csv/${1}-${2}.postorder.csv" \
+        "replicate-results/img/${1}-postorder.eps"
+}
 
-./bin/generate-timings \
-    --no-generate \
-    --data data-sets/pathological.afasta \
-    --tree data-sets/pathological.tree \
-    --tcm  data-sets/tcm-31.tcm \
-    --output pathological-31 \
-    -n [4,8,16,32,64,128,256] \
-    -k [1%1,1%2,1%4,1%8,1%16,1%32,1%64]
-python3 plot-figure.py replicate-results/csv/pathological-31.preorder.csv  replicate-results/img/pathological-31-preorder.eps
-python3 plot-figure.py replicate-results/csv/pathological-31.postorder.csv replicate-results/img/pathological-31-postorder.eps
+generate 'fungi'        '11' '[25,50,100,200,400,800,1553]' '[1%1,1%2,1%4,1%8,1%16,1%32]'
+generate 'metazoa'      '11' '[25,50,100,200,400,800,1766]' '[1%1,1%2,1%4,1%8,1%16,1%32]'
+generate 'pathological' '12' '[4,8,16,32,64,128,256]'       '[1%1,1%2,1%4,1%8,1%16,1%32,1%64]'
+generate 'pathological' '31' '[4,8,16,32,64,128,256]'       '[1%1,1%2,1%4,1%8,1%16,1%32,1%64]' '--no-generate'