Optimization: interface segregation: don't ask for function if only evaluation is needed

src/Numerics/Optimization/BfgsBMinimizer.cs

@@ -158,9 +158,9 @@ namespace MathNet.Numerics.Optimization
 
         protected override Vector<double> CalculateSearchDirection(ref Matrix<double> pseudoHessian, 
             out double maxLineSearchStep,
-            out double startingStepSize, 
-            IObjectiveFunction previousPoint, 
-            IObjectiveFunction candidatePoint, 
+            out double startingStepSize,
+            IObjectiveFunction previousPoint,
+            IObjectiveFunction candidatePoint,
             Vector<double> step)
         {
             Vector<double> lineSearchDirection;
@@ -170,7 +170,7 @@ namespace MathNet.Numerics.Optimization
             if (sy > 0.0) // only do update if it will create a positive definite matrix
             {
                 double sts = step * step;
-                
+
                 var Hs = pseudoHessian * step;
                 var sHs = step * pseudoHessian * step;
                 pseudoHessian = pseudoHessian + y.OuterProduct(y) * (1.0 / sy) - Hs.OuterProduct(Hs) * (1.0 / sHs);
@@ -285,7 +285,7 @@ namespace MathNet.Numerics.Optimization
             }
         }
 
-        protected override double GetProjectedGradient(IObjectiveFunction candidatePoint, int ii)
+        protected override double GetProjectedGradient(IObjectiveFunctionEvaluation candidatePoint, int ii)
         {
             double projectedGradient;
             bool atLowerBound = candidatePoint.Point[ii] - _lowerBound[ii] < VerySmall;

src/Numerics/Optimization/BfgsMinimizerBase.cs

@@ -59,7 +59,7 @@ namespace MathNet.Numerics.Optimization
             MaximumIterations = maximumIterations;
         }
 
-        protected MinimizationResult.ExitCondition ExitCriteriaSatisfied(IObjectiveFunction candidatePoint, IObjectiveFunction lastPoint, int iterations)
+        protected MinimizationResult.ExitCondition ExitCriteriaSatisfied(IObjectiveFunctionEvaluation candidatePoint, IObjectiveFunctionEvaluation lastPoint, int iterations)
         {
             Vector<double> relGrad = new DenseVector(candidatePoint.Point.Count);
             double relativeGradient = 0.0;
@@ -99,12 +99,12 @@ namespace MathNet.Numerics.Optimization
             return MinimizationResult.ExitCondition.None;
         }
 
-        protected virtual double GetProjectedGradient(IObjectiveFunction candidatePoint, int ii)
+        protected virtual double GetProjectedGradient(IObjectiveFunctionEvaluation candidatePoint, int ii)
         {
             return candidatePoint.Gradient[ii];
         }
 
-        protected void ValidateGradientAndObjective(IObjectiveFunction eval)
+        protected void ValidateGradientAndObjective(IObjectiveFunctionEvaluation eval)
         {
             foreach (var x in eval.Gradient)
             {

src/Numerics/Optimization/ConjugateGradientMinimizer.cs

@@ -127,7 +127,7 @@ namespace MathNet.Numerics.Optimization
             return gradient.Norm(2.0) < GradientTolerance;
         }
 
-        void ValidateGradient(IObjectiveFunction objective)
+        void ValidateGradient(IObjectiveFunctionEvaluation objective)
         {
             foreach (var x in objective.Gradient)
             {
@@ -136,7 +136,7 @@ namespace MathNet.Numerics.Optimization
             }
         }
 
-        void ValidateObjective(IObjectiveFunction objective)
+        void ValidateObjective(IObjectiveFunctionEvaluation objective)
         {
             if (Double.IsNaN(objective.Value) || Double.IsInfinity(objective.Value))
                 throw new EvaluationException("Non-finite objective function returned.", objective);

src/Numerics/Optimization/Exceptions.cs

@@ -49,15 +49,15 @@ namespace MathNet.Numerics.Optimization
 
     public class EvaluationException : OptimizationException
     {
-        public IObjectiveFunction ObjectiveFunction { get; private set; }
+        public IObjectiveFunctionEvaluation ObjectiveFunction { get; private set; }
 
-        public EvaluationException(string message, IObjectiveFunction eval)
+        public EvaluationException(string message, IObjectiveFunctionEvaluation eval)
             : base(message)
         {
             ObjectiveFunction = eval;
         }
 
-        public EvaluationException(string message, IObjectiveFunction eval, Exception innerException)
+        public EvaluationException(string message, IObjectiveFunctionEvaluation eval, Exception innerException)
             : base(message, innerException)
         {
             ObjectiveFunction = eval;

src/Numerics/Optimization/LineSearch/StrongWolfeLineSearch.cs

@@ -40,7 +40,10 @@ namespace MathNet.Numerics.Optimization.LineSearch
             // Argument validation in base class
         }
 
-        protected override MinimizationResult.ExitCondition WolfeExitCondition { get { return MinimizationResult.ExitCondition.StrongWolfeCriteria; } }
+        protected override MinimizationResult.ExitCondition WolfeExitCondition
+        {
+            get { return MinimizationResult.ExitCondition.StrongWolfeCriteria; }
+        }
 
         protected override bool WolfeCondition(double stepDd, double initialDd)
         {

src/Numerics/Optimization/LineSearch/WeakWolfeLineSearch.cs

@@ -64,7 +64,7 @@ namespace MathNet.Numerics.Optimization.LineSearch
             return stepDd < C2 * initialDd;
         }
 
-        protected override void ValidateValue(IObjectiveFunction eval)
+        protected override void ValidateValue(IObjectiveFunctionEvaluation eval)
         {
             if (!IsFinite(eval.Value))
             {
@@ -78,7 +78,7 @@ namespace MathNet.Numerics.Optimization.LineSearch
                 throw new ArgumentException("objective function does not support gradient");
         }
 
-        protected override void ValidateGradient(IObjectiveFunction eval)
+        protected override void ValidateGradient(IObjectiveFunctionEvaluation eval)
         {
             foreach (double x in eval.Gradient)
             {

src/Numerics/Optimization/LineSearch/WolfeLineSearch.cs

@@ -87,8 +87,8 @@ namespace MathNet.Numerics.Optimization.LineSearch
             for (ii = 0; ii < MaximumIterations; ++ii)
             {
                 objective.EvaluateAt(startingPoint.Point + searchDirection * step);
-                ValidateGradient(objective);  
-                ValidateValue(objective);       
+                ValidateGradient(objective);
+                ValidateValue(objective);
 
                 double stepDd = searchDirection * objective.Gradient;
 
@@ -104,7 +104,7 @@ namespace MathNet.Numerics.Optimization.LineSearch
                 }
                 else
                 {
-                    reasonForExit = WolfeExitCondition; 
+                    reasonForExit = WolfeExitCondition;
                     break;
                 }
 
@@ -136,23 +136,21 @@ namespace MathNet.Numerics.Optimization.LineSearch
 
             return new LineSearchResult(objective, ii, step, reasonForExit);
         }
+
         protected abstract MinimizationResult.ExitCondition WolfeExitCondition { get; }
 
         protected abstract bool WolfeCondition(double stepDd, double initialDd);
 
-        protected virtual void ValidateGradient(IObjectiveFunction objective)
+        protected virtual void ValidateGradient(IObjectiveFunctionEvaluation objective)
         {
         }
-        protected virtual void ValidateValue(IObjectiveFunction objective)
+
+        protected virtual void ValidateValue(IObjectiveFunctionEvaluation objective)
         {
         }
 
         protected virtual void ValidateInputArguments(IObjectiveFunctionEvaluation startingPoint, Vector<double> searchDirection, double initialStep, double upperBound)
         {
-
         }
     }
-
-        
-    
 }

src/Numerics/Optimization/NelderMeadSimplex.cs

@@ -32,15 +32,12 @@
 
 using MathNet.Numerics.LinearAlgebra;
 using System;
-using System.Collections.Generic;
-using System.Linq;
-using System.Text;
 
 namespace MathNet.Numerics.Optimization
 {
     /// <summary>
-    /// Class implementing the Nelder-Mead simplex algorithm, used to find a minima when no gradient is available. 
-    /// Called fminsearch() in Matlab. A description of the algorithm can be found at 
+    /// Class implementing the Nelder-Mead simplex algorithm, used to find a minima when no gradient is available.
+    /// Called fminsearch() in Matlab. A description of the algorithm can be found at
     /// http://se.mathworks.com/help/matlab/math/optimizing-nonlinear-functions.html#bsgpq6p-11
     /// or
     /// https://en.wikipedia.org/wiki/Nelder%E2%80%93Mead_method
@@ -68,7 +65,7 @@ namespace MathNet.Numerics.Optimization
         /// <returns>The minimum point</returns>
         public MinimizationResult FindMinimum(IObjectiveFunction objectiveFunction, Vector<double> initialGuess)
         {
-            var initalPertubation = new MathNet.Numerics.LinearAlgebra.Double.DenseVector(initialGuess.Count);
+            var initalPertubation = new LinearAlgebra.Double.DenseVector(initialGuess.Count);
             for (int i = 0; i < initialGuess.Count; i++)
             {
                 initalPertubation[i] = initialGuess[i] == 0.0 ? 0.00025 : initialGuess[i] * 0.05;
@@ -249,7 +246,7 @@ namespace MathNet.Numerics.Optimization
             Vector<double>[] vertices = new Vector<double>[numDimensions + 1];
 
             // define one point of the simplex as the given initial guesses
-            var p0 = new MathNet.Numerics.LinearAlgebra.Double.DenseVector(numDimensions);
+            var p0 = new LinearAlgebra.Double.DenseVector(numDimensions);
             for (int i = 0; i < numDimensions; i++)
             {
                 p0[i] = simplexConstants[i].Value;
@@ -261,7 +258,7 @@ namespace MathNet.Numerics.Optimization
             for (int i = 0; i < numDimensions; i++)
             {
                 double scale = simplexConstants[i].InitialPerturbation;
-                Vector<double> unitVector = new MathNet.Numerics.LinearAlgebra.Double.DenseVector(numDimensions);
+                Vector<double> unitVector = new LinearAlgebra.Double.DenseVector(numDimensions);
                 unitVector[i] = 1;
                 vertices[i + 1] = p0.Add(unitVector.Multiply(scale));
             }
@@ -335,7 +332,7 @@ namespace MathNet.Numerics.Optimization
         {
             int numVertices = vertices.Length;
             // find the centroid of all points except the worst one
-            Vector<double> centroid = new MathNet.Numerics.LinearAlgebra.Double.DenseVector(numVertices - 1);
+            Vector<double> centroid = new LinearAlgebra.Double.DenseVector(numVertices - 1);
             for (int i = 0; i < numVertices; i++)
             {
                 if (i != errorProfile.HighestIndex)
@@ -409,7 +406,4 @@ namespace MathNet.Numerics.Optimization
             }
         }
     }
-
-    
-
 }

src/Numerics/Optimization/NewtonMinimizer.cs

@@ -126,7 +126,7 @@ namespace MathNet.Numerics.Optimization
             return gradient.Norm(2.0) < GradientTolerance;
         }
 
-        static void ValidateGradient(IObjectiveFunction eval)
+        static void ValidateGradient(IObjectiveFunctionEvaluation eval)
         {
             foreach (var x in eval.Gradient)
             {
@@ -137,13 +137,13 @@ namespace MathNet.Numerics.Optimization
             }
         }
 
-        private void ValidateObjective(IObjectiveFunction eval)
+        private void ValidateObjective(IObjectiveFunctionEvaluation eval)
         {
             if (Double.IsNaN(eval.Value) || Double.IsInfinity(eval.Value))
                 throw new EvaluationException("Non-finite objective function returned.", eval);
         }
 
-        private void ValidateHessian(IObjectiveFunction eval)
+        private void ValidateHessian(IObjectiveFunctionEvaluation eval)
         {
             for (int ii = 0; ii < eval.Hessian.RowCount; ++ii)
             {

src/Numerics/Optimization/ObjectiveFunctions/ForwardDifferenceGradientObjectiveFunction.cs

@@ -76,23 +76,23 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
             if (!ValueEvaluated)
                 EvaluateValue();
 
-            var tmp_point = Point.Clone();
-            var tmp_obj = InnerObjectiveFunction.CreateNew();
+            var tmpPoint = Point.Clone();
+            var tmpObj = InnerObjectiveFunction.CreateNew();
             for (int ii = 0; ii < _gradient.Count; ++ii)
             {
-                var orig_point = tmp_point[ii];
-                var rel_incr = orig_point * RelativeIncrement;
-                var h = Math.Max(rel_incr, MinimumIncrement);
+                var origPoint = tmpPoint[ii];
+                var relIncr = origPoint * RelativeIncrement;
+                var h = Math.Max(relIncr, MinimumIncrement);
                 var mult = 1;
-                if (orig_point + h > UpperBound[ii])
+                if (origPoint + h > UpperBound[ii])
                     mult = -1;
 
-                tmp_point[ii] = orig_point + mult*h;
-                tmp_obj.EvaluateAt(tmp_point);
-                double bumped_value = tmp_obj.Value;
-                _gradient[ii] = (mult * bumped_value - mult * InnerObjectiveFunction.Value) / h;
+                tmpPoint[ii] = origPoint + mult*h;
+                tmpObj.EvaluateAt(tmpPoint);
+                double bumpedValue = tmpObj.Value;
+                _gradient[ii] = (mult * bumpedValue - mult * InnerObjectiveFunction.Value) / h;
 
-                tmp_point[ii] = orig_point;
+                tmpPoint[ii] = origPoint;
             }
             GradientEvaluated = true;
         }

src/Numerics/Optimization/ObjectiveFunctions/GradientHessianObjectiveFunction.cs

@@ -84,4 +84,4 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
         public Vector<double> Gradient { get; private set; }
         public Matrix<double> Hessian { get; private set; }
     }
-}
+}

src/Numerics/Optimization/ObjectiveFunctions/GradientObjectiveFunction.cs

@@ -86,4 +86,4 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
             get { throw new NotSupportedException(); }
         }
     }
-}
+}

src/Numerics/Optimization/ObjectiveFunctions/HessianObjectiveFunction.cs

@@ -86,4 +86,4 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
             get { throw new NotSupportedException(); }
         }
     }
-}
+}

src/Numerics/Optimization/ObjectiveFunctions/LazyObjectiveFunction.cs

@@ -139,4 +139,4 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
             }
         }
     }
-}
+}

src/Numerics/Optimization/ObjectiveFunctions/ValueObjectiveFunction.cs

@@ -86,4 +86,4 @@ namespace MathNet.Numerics.Optimization.ObjectiveFunctions
             get { throw new NotSupportedException(); }
         }
     }
-}
+}