[16674] | 1 | using System;
|
---|
| 2 | using System.Collections.Generic;
|
---|
| 3 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 4 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
| 5 | using Microsoft.VisualStudio.TestTools.UnitTesting;
|
---|
| 6 | using System.Linq;
|
---|
| 7 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
| 8 |
|
---|
| 9 | namespace Tests {
|
---|
| 10 | [TestClass]
|
---|
| 11 | public class AutoDiffTest {
|
---|
| 12 | [TestMethod]
|
---|
[16682] | 13 | public void Test() {
|
---|
[16674] | 14 | {
|
---|
[16682] | 15 | // eval
|
---|
[16674] | 16 | var parser = new InfixExpressionParser();
|
---|
| 17 | var t = parser.Parse("2.0*x+y");
|
---|
| 18 |
|
---|
[16682] | 19 | // interval eval
|
---|
[16674] | 20 | var evaluator = new IntervalEvaluator();
|
---|
[16682] | 21 | var intervals = new Dictionary<string, Interval>();
|
---|
| 22 | intervals.Add("x", new Interval(-1.0, 1.0));
|
---|
| 23 | intervals.Add("y", new Interval(2.0, 10.0));
|
---|
[16674] | 24 | var resultInterval = evaluator.Evaluate(t, intervals);
|
---|
| 25 | Assert.AreEqual(0, resultInterval.LowerBound);
|
---|
| 26 | Assert.AreEqual(12, resultInterval.UpperBound);
|
---|
| 27 | }
|
---|
| 28 |
|
---|
| 29 | {
|
---|
[16682] | 30 | // vector eval
|
---|
[16674] | 31 | var parser = new InfixExpressionParser();
|
---|
| 32 | var t = parser.Parse("2.0*x+y");
|
---|
| 33 |
|
---|
| 34 | var evaluator = new VectorEvaluator();
|
---|
| 35 | var vars = new string[] { "x", "y", "f(x)" };
|
---|
| 36 | var values = new double[,] {
|
---|
| 37 | { 1, 1, 0 },
|
---|
| 38 | { 2, 1, 0 },
|
---|
| 39 | { 3, -1, 0 },
|
---|
| 40 | { 4, -1, 0 },
|
---|
| 41 | { 5, -1, 0 },
|
---|
| 42 | };
|
---|
| 43 |
|
---|
| 44 | var ds = new Dataset(vars, values);
|
---|
| 45 | var problemData = new RegressionProblemData(ds, vars, "f(x)");
|
---|
| 46 | var train = evaluator.Evaluate(t, ds, problemData.TrainingIndices.ToArray());
|
---|
| 47 | Assert.AreEqual(2, train.Length);
|
---|
| 48 | Assert.AreEqual(3, train[0]);
|
---|
| 49 | Assert.AreEqual(5, train[1]);
|
---|
| 50 |
|
---|
| 51 | var test = evaluator.Evaluate(t, ds, problemData.TestIndices.ToArray());
|
---|
| 52 | Assert.AreEqual(3, test.Length);
|
---|
| 53 | Assert.AreEqual(5, test[0]);
|
---|
| 54 | Assert.AreEqual(7, test[1]);
|
---|
| 55 | Assert.AreEqual(9, test[2]);
|
---|
| 56 | }
|
---|
| 57 |
|
---|
| 58 | {
|
---|
[16682] | 59 | // vector eval and auto-diff
|
---|
[16674] | 60 | var parser = new InfixExpressionParser();
|
---|
| 61 | var t = parser.Parse("2.0*x+y");
|
---|
| 62 | var p0 = t.IterateNodesPostfix().First(n => n is ConstantTreeNode);
|
---|
| 63 | var p1 = t.IterateNodesPostfix().First(n => (n is VariableTreeNode var) && var.VariableName == "y");
|
---|
| 64 | var paramNodes = new ISymbolicExpressionTreeNode[] { p0, p1 };
|
---|
| 65 |
|
---|
| 66 | var evaluator = new VectorAutoDiffEvaluator();
|
---|
| 67 | var vars = new string[] { "x", "y", "f(x)" };
|
---|
| 68 | var values = new double[,] {
|
---|
| 69 | { 1, 1, 0 },
|
---|
| 70 | { 2, 1, 0 },
|
---|
| 71 | { 3, -1, 0 },
|
---|
| 72 | { 4, -1, 0 },
|
---|
| 73 | { 5, -1, 0 },
|
---|
| 74 | };
|
---|
| 75 |
|
---|
| 76 | var ds = new Dataset(vars, values);
|
---|
| 77 | var problemData = new RegressionProblemData(ds, vars, "f(x)");
|
---|
| 78 | evaluator.Evaluate(t, ds, problemData.TrainingIndices.ToArray(), paramNodes, out double[] train, out double[,] trainJac);
|
---|
| 79 | Assert.AreEqual(2, train.Length);
|
---|
| 80 | Assert.AreEqual(3, train[0]);
|
---|
| 81 | Assert.AreEqual(5, train[1]);
|
---|
| 82 | // check jac
|
---|
| 83 | Assert.AreEqual(1, trainJac[0, 0]);
|
---|
| 84 | Assert.AreEqual(1, trainJac[0, 1]);
|
---|
| 85 | Assert.AreEqual(2, trainJac[1, 0]);
|
---|
| 86 | Assert.AreEqual(1, trainJac[1, 1]);
|
---|
| 87 |
|
---|
| 88 | evaluator.Evaluate(t, ds, problemData.TestIndices.ToArray(), paramNodes, out double[] test, out double[,] testJac);
|
---|
| 89 | Assert.AreEqual(3, test.Length);
|
---|
| 90 | Assert.AreEqual(5, test[0]);
|
---|
| 91 | Assert.AreEqual(7, test[1]);
|
---|
| 92 | Assert.AreEqual(9, test[2]);
|
---|
| 93 |
|
---|
| 94 | // check jac
|
---|
| 95 | Assert.AreEqual(3, testJac[0, 0]);
|
---|
| 96 | Assert.AreEqual(-1, testJac[0, 1]);
|
---|
| 97 | Assert.AreEqual(4, testJac[1, 0]);
|
---|
| 98 | Assert.AreEqual(-1, testJac[1, 1]);
|
---|
| 99 | Assert.AreEqual(5, testJac[2, 0]);
|
---|
| 100 | Assert.AreEqual(-1, testJac[2, 1]);
|
---|
| 101 |
|
---|
| 102 | }
|
---|
| 103 |
|
---|
[16682] | 104 | {
|
---|
| 105 | // interval eval and auto-diff
|
---|
| 106 | var parser = new InfixExpressionParser();
|
---|
| 107 | var t = parser.Parse("2.0*x+y");
|
---|
| 108 | var p0 = t.IterateNodesPostfix().First(n => n is ConstantTreeNode);
|
---|
| 109 | var p1 = t.IterateNodesPostfix().First(n => (n is VariableTreeNode var) && var.VariableName == "y");
|
---|
| 110 | var paramNodes = new ISymbolicExpressionTreeNode[] { p0, p1 };
|
---|
| 111 |
|
---|
| 112 | var evaluator = new IntervalEvaluator();
|
---|
| 113 | var intervals = new Dictionary<string, Interval>();
|
---|
| 114 | intervals.Add("x", new Interval(-1.0, 1.0));
|
---|
| 115 | intervals.Add("y", new Interval(2.0, 10.0));
|
---|
| 116 | var resultInterval = evaluator.Evaluate(t, intervals, paramNodes, out double[] lowerGradient, out double[] upperGradient);
|
---|
| 117 | Assert.AreEqual(0, resultInterval.LowerBound);
|
---|
| 118 | Assert.AreEqual(12, resultInterval.UpperBound);
|
---|
| 119 |
|
---|
| 120 | Assert.AreEqual(-1, lowerGradient[0]);
|
---|
| 121 | Assert.AreEqual(2, lowerGradient[1]);
|
---|
| 122 | Assert.AreEqual(1, upperGradient[0]);
|
---|
| 123 | Assert.AreEqual(10, upperGradient[1]);
|
---|
| 124 | }
|
---|
| 125 |
|
---|
[16674] | 126 | }
|
---|
| 127 | }
|
---|
| 128 | }
|
---|