source: branches/HeuristicLab.BottomUpTreeDistance/HeuristicLab.Tests/BottomUpTreeSimilarityCalculatorTest.cs @ 11244

Last change on this file since 11244 was 11244, checked in by bburlacu, 7 years ago

#2215: Updated test traits in the bottom up and max common subtree similarity tests.

File size: 8.2 KB
Line 
1using System;
2using System.Collections.Generic;
3using System.Diagnostics;
4using System.Drawing;
5using System.Globalization;
6using System.Linq;
7using System.Text;
8using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
9using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding.Views;
10using HeuristicLab.Random;
11using Microsoft.VisualStudio.TestTools.UnitTesting;
12
13namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Tests {
14  [TestClass]
15  public class BottomUpSimilarityCalculatorTest {
16    private readonly BottomUpTreeSimilarityCalculator busCalculator;
17    private readonly SymbolicExpressionImporter importer;
18
19    private const int N = 100;
20    private const int Rows = 1;
21    private const int Columns = 10;
22
23    public BottomUpSimilarityCalculatorTest() {
24      busCalculator = new BottomUpTreeSimilarityCalculator(new List<string> { "Addition", "Multiplication", "And", "Or", "Xor" });
25      importer = new SymbolicExpressionImporter();
26    }
27
28    [TestMethod]
29    [TestCategory("Problems.DataAnalysis.Symbolic")]
30    [TestProperty("Time", "short")]
31    public void TestBottomUpMatching() {
32      TestMatchedNodes("(+ 1 2)", "(+ 2 1)", 5);
33      TestMatchedNodes("(- 2 1)", "(- 1 2)", 2);
34      TestMatchedNodes("(* (variable 1 X1) (variable 1 X2))", "(* (+ (variable 1 X1) 1) (+ (variable 1 X2) 1))", 2);
35
36      TestMatchedNodes("(* (variable 1 X1) (variable 1 X2))", "(* (+ (variable 1 X1) 1) (variable 1 X2))", 2);
37
38      TestMatchedNodes("(+ (variable 1 a) (variable 1 b))", "(+ (variable 1 a) (variable 1 a))", 1);
39      TestMatchedNodes("(+ (+ (variable 1 a) (variable 1 b)) (variable 1 b))", "(+ (* (+ (variable 1 a) (variable 1 b)) (variable 1 b)) (+ (+ (variable 1 a) (variable 1 b)) (variable 1 b)))", 5);
40
41      TestMatchedNodes(
42        "(* (+ 2.84 (exp (+ (log (/ (variable 2.0539 X5) (variable -9.2452e-1 X6))) (/ (variable 2.0539 X5) (variable -9.2452e-1 X6))))) 2.9081)",
43        "(* (- (variable 9.581e-1 X6) (+ (- (variable 5.1491e-1 X5) 1.614e+1) (+ (/ (variable 2.0539 X5) (variable -9.2452e-1 X6)) (log (/ (variable 2.0539 X5) (variable -9.2452e-1 X6)))))) 2.9081)",
44        9);
45
46      TestMatchedNodes("(+ (exp 2.1033) (/ -4.3072 (variable 2.4691 X7)))", "(/ 1 (+ (/ -4.3072 (variable 2.4691 X7)) (exp 2.1033)))", 6);
47      TestMatchedNodes("(+ (exp 2.1033) (/ -4.3072 (variable 2.4691 X7)))", "(/ 1 (+ (/ (variable 2.4691 X7) -4.3072) (exp 2.1033)))", 4);
48
49      const string expr1 = "(* (- 1.2175e+1 (+ (/ (exp -1.4134e+1) (exp 9.2013)) (exp (log (exp (/ (exp (- (* -4.2461 (variable 2.2634 X5)) (- -9.6267e-1 3.3243))) (- (/ (/ (variable 1.0883 X1) (variable 6.9620e-1 X2)) (log 1.3011e+1)) (variable -4.3098e-1 X7)))))))) (log 1.3011e+1))";
50      const string expr2 = "(* (- 1.2175e+1 (+ (/ (/ (+ (variable 3.0140 X9) (variable 1.3430 X8)) -1.0864e+1) (exp 9.2013)) (exp (log (exp (/ (exp (- (* -4.2461 (variable 2.2634 X5)) (- -9.6267e-1 3.3243))) (- (/ (/ (variable 1.0883 X1) (variable 6.9620e-1 X2)) (log 1.3011e+1)) (variable -4.3098e-1 X7)))))))) (exp (variable 4.0899e-1 X7)))";
51
52      TestMatchedNodes(expr1, expr2, 23);
53
54    }
55
56    [TestMethod]
57    [TestCategory("Problems.DataAnalysis.Symbolic")]
58    [TestProperty("Time", "short")]
59    private void TestMatchedNodes(string expr1, string expr2, int expected) {
60      var t1 = importer.Import(expr1);
61      var t2 = importer.Import(expr2);
62
63      var mapping = busCalculator.ComputeBottomUpMapping(t1.Root, t2.Root);
64      var c = mapping.Count;
65
66      if (c != expected) {
67        throw new Exception("Match count " + c + " is different than expected value " + expected);
68      }
69    }
70
71    [TestMethod]
72    [TestCategory("Problems.DataAnalysis.Symbolic")]
73    [TestProperty("Time", "long")]
74    public void TestBottomUpSimilarityCalculatorPerformance() {
75      var grammar = new TypeCoherentExpressionGrammar();
76      grammar.ConfigureAsDefaultRegressionGrammar();
77      var twister = new MersenneTwister(31415);
78      var ds = Util.CreateRandomDataset(twister, Rows, Columns);
79      var trees = Util.CreateRandomTrees(twister, ds, grammar, N, 1, 100, 0, 0);
80
81      double s = 0;
82      var sw = new Stopwatch();
83
84      sw.Start();
85      for (int i = 0; i < trees.Length - 1; ++i) {
86        for (int j = i + 1; j < trees.Length; ++j) {
87          s += busCalculator.CalculateSolutionSimilarity(trees[i], trees[j]);
88        }
89      }
90
91      sw.Stop();
92      Console.WriteLine("Elapsed time: " + sw.ElapsedMilliseconds / 1000.0 + ", Avg. similarity: " + s / (N * (N - 1) / 2));
93      Console.WriteLine(N * (N + 1) / (2 * sw.ElapsedMilliseconds / 1000.0) + " similarity calculations per second.");
94    }
95
96    private static string FormatMapping(ISymbolicExpressionTree t1, ISymbolicExpressionTree t2, Dictionary<ISymbolicExpressionTreeNode, ISymbolicExpressionTreeNode> map) {
97      var symbolNameMap = new Dictionary<string, string>
98    {
99      {"ProgramRootSymbol", "Prog"},
100      {"StartSymbol","RPB"},
101      {"Multiplication", "$\\times$"},
102      {"Division", "$\\div$"},
103      {"Addition", "$+$"},
104      {"Subtraction", "$-$"},
105      {"Exponential", "$\\exp$"},
106      {"Logarithm", "$\\log$"}
107    };
108
109      var sb = new StringBuilder();
110      var nodeIds = new Dictionary<ISymbolicExpressionTreeNode, string>();
111      int offset = 0;
112      var layoutEngine = new ReingoldTilfordLayoutEngine<ISymbolicExpressionTreeNode>(x => x.Subtrees);
113      var nodeCoordinates = layoutEngine.CalculateLayout(t1.Root).ToDictionary(n => n.Content, n => new PointF(n.X, n.Y));
114
115      double ws = 0.5;
116      double hs = 0.5;
117
118      var nl = Environment.NewLine;
119      sb.Append("\\documentclass[class=minimal,border=0pt]{standalone}" + nl +
120                 "\\usepackage{tikz}" + nl +
121                 "\\begin{document}" + nl +
122                 "\\begin{tikzpicture}" + nl +
123                 "\\def\\ws{1}" + nl +
124                 "\\def\\hs{0.7}" + nl +
125                 "\\def\\offs{" + offset + "}" + nl);
126
127      foreach (var node in t1.IterateNodesBreadth()) {
128        var id = Guid.NewGuid().ToString();
129        nodeIds[node] = id;
130        var coord = nodeCoordinates[node];
131        var nodeName = symbolNameMap.ContainsKey(node.Symbol.Name) ? symbolNameMap[node.Symbol.Name] : node.ToString();
132        sb.AppendLine(string.Format(CultureInfo.InvariantCulture, "\\node ({0}) at (\\ws*{1} + \\offs,\\hs*{2}) {{{3}}};", nodeIds[node], ws * coord.X, -hs * coord.Y, EscapeLatexString(nodeName)));
133      }
134
135      foreach (ISymbolicExpressionTreeNode t in t1.IterateNodesBreadth()) {
136        var n = t;
137        foreach (var s in t.Subtrees) {
138          sb.AppendLine(string.Format(CultureInfo.InvariantCulture, "\\draw ({0}) -- ({1});", nodeIds[n], nodeIds[s]));
139        }
140      }
141
142      nodeCoordinates = layoutEngine.CalculateLayout(t2.Root).ToDictionary(n => n.Content, n => new PointF(n.X, n.Y));
143
144      offset = 20;
145      sb.Append("\\def\\offs{" + offset + "}" + nl);
146      foreach (var node in t2.IterateNodesBreadth()) {
147        var id = Guid.NewGuid().ToString();
148        nodeIds[node] = id;
149        var coord = nodeCoordinates[node];
150        var nodeName = symbolNameMap.ContainsKey(node.Symbol.Name) ? symbolNameMap[node.Symbol.Name] : node.ToString();
151        sb.AppendLine(string.Format(CultureInfo.InvariantCulture, "\\node ({0}) at (\\ws*{1} + \\offs,\\hs*{2}) {{{3}}};", nodeIds[node], ws * coord.X, -hs * coord.Y, EscapeLatexString(nodeName)));
152      }
153
154      foreach (ISymbolicExpressionTreeNode t in t2.IterateNodesBreadth()) {
155        var n = t;
156        foreach (var s in t.Subtrees) {
157          sb.AppendLine(string.Format(CultureInfo.InvariantCulture, "\\draw ({0}) -- ({1});", nodeIds[n], nodeIds[s]));
158        }
159      }
160
161      foreach (var p in map) {
162        var id1 = nodeIds[p.Key];
163        var id2 = nodeIds[p.Value];
164
165        sb.Append(string.Format(CultureInfo.InvariantCulture, "\\path[draw,->,color=gray] ({0}) edge[bend left,dashed] ({1});" + Environment.NewLine, id1, id2));
166      }
167      sb.Append("\\end{tikzpicture}" + nl +
168                "\\end{document}" + nl);
169      return sb.ToString();
170    }
171
172    private static string EscapeLatexString(string s) {
173      return s.Replace("\\", "\\\\").Replace("{", "\\{").Replace("}", "\\}").Replace("_", "\\_");
174    }
175  }
176}
Note: See TracBrowser for help on using the repository browser.