Numerical 3D Generator

Numerical-3-Dimensional-Matching/gen/README.md

@@ -0,0 +1,13 @@
+# Problem: Numerical 3-Dimensional Matching
+
+# Description
+The generator takes in 2 arguments: the desired set size (`s`) for each input set and the target value of the sum (`n`).
+
+# Algorithm
+For generating the sets, we borrow an approach from combinatorics called the [Stars & Bars](https://en.wikipedia.org/wiki/Stars_and_bars_(combinatorics)) method.
+We make a list with a number of stars equal to the target value and 3 bars (since there are 3 sets to generate). Then we shuffle the list and count how many stars 
+we encounter until the next bar. This is partitions the list into 3 portions and guarantees that we always sum to the appropriate value. Then we add one of each 
+value to each of the lists. 
+
+Due to the algorithm the solver uses, no set can contain repeat elements. So occainsonally, the sets may be a few shorter than the user 
+expects. Generally, larger target values and smaller set sizes produce the desired set size every time.

Numerical-3-Dimensional-Matching/gen/generate.py

@@ -0,0 +1,92 @@
+# File: generate.py
+# Author: Jackson Madsen
+
+# Description: Set generator for Numerical 3-Dimension problem solver
+# Use: python3 generate.py -s s -n n > filename.lp
+#	where: s is the maximum size of each set, inclusive
+#	       n is the target value for the triples
+#	       filename.lp is the location to write the instance to
+#	Each of s, n are optional (Defaults: s=5, n=10)
+#	The redirect (> filename.lp) can be omitted (to print to stdout)
+
+import random
+import argparse
+
+# Process arguments
+parser = argparse.ArgumentParser()
+parser.add_argument('-n', default=10, type=int,
+                    help="Target vale for the triples. (Default: 10)")
+parser.add_argument('-s', default=5, type=int,
+                    help="Set the size of each set. (Default: 5)")
+args = parser.parse_args()
+
+targetVal = args.n
+maxSize = args.s
+
+# Here we use the stars and bars method to generate a random permutation
+# The total number of stars is the target value.
+# Each bar will partition the list such that the number of stars before encountering a bar will be the value that gets appended.
+# When we shuffle, the list is partitioned and gives us one of the triples.
+# Read in the values from the shuffled list, then appened those to one of each of the lists.
+# Repeat n times until we have 3 sets of the requested size from the User.
+stars = ['*'] * targetVal + ['|'] * 2
+
+sets = []
+
+for _ in range(maxSize):
+  random.shuffle(stars)
+  partition = []
+
+  count = 0
+  for item in stars:
+    if item == '*':
+      count += 1
+    elif item == '|':
+      partition.append(count)
+      count = 0
+  partition.append(count)
+
+  seen = False
+  for index, part in enumerate(partition):
+    for item in sets:
+      if part == item[index]:
+        seen = True
+
+  if not seen:
+    sets.append(partition)
+
+  random.shuffle(stars)
+
+# Create lists
+x = []
+y = []
+z = []
+
+for i in range(len(sets)):
+  a, b, c = sets[i]
+  x.append(a)
+  y.append(b)
+  z.append(c)
+
+random.shuffle(x)
+random.shuffle(y)
+random.shuffle(z)
+
+# Output file
+print("% GENERATE")
+print("% Custom instance works with a parameter of n=", targetVal)
+
+print("x(", end="")
+for val in x[:-1]:
+  print(val, end=" ; ")
+print(x[-1], ").", sep="")
+
+print("y(", end="")
+for val in y[:-1]:
+  print(val, end=" ; ")
+print(y[-1], ").", sep="")
+
+print("z(", end="")
+for val in z[:-1]:
+  print(val, end=" ; ")
+print(z[-1], ").", sep="")