/// <summary>

/// AIProgrammer experiment, using artificial intelligence to generate a program that solves a solution.

/// This experiment uses a genetic algorithm to evolve a program in the programming language Brainfuck.

/// The resulting program will print a target string.

///

/// Created by Kory Becker 01-Jan-2013 http://www.primaryobjects.com/kory-becker

/// </summary>

class Program

{

#region Private Variables

private static GA _ga = null; // Our genetic algorithm instance.

private static GAStatus _bestStatus = new GAStatus(); // Holds values for displaying best generation statistics.

private static DateTime _startTime = DateTime.Now; // Time the program was started.

private static string _appendCode = null; // Program code, containing functions, that will be appended to main program code

private static TargetParams _targetParams = new TargetParams { TargetString = "hi" }; // Used for displaying the target fitness

#endregion

#region Genetic Algorithm Settings

private static double _crossoverRate = 0.70; // Percentage chance that a child genome will use crossover of two parents.

private static double _mutationRate = 0.01; // Percentage chance that a child genome will mutate a gene.

private static int _genomeSize = 150; // Number of programming instructions in generated program (size of genome array). loops).

private static int _maxGenomeSize = 150; // The max length a genome may grow to (only applicable if _expandAmount > 0).

private static int _maxIterationCount = 5000; // Max iterations a program may run before being killed (prevents infinite loops).

private static int _expandAmount = 0; // The max genome size will expand by this amount, every _expandRate iterations (may help learning). Set to 0 to disable.

private static int _expandRate = 5000; // The max genome size will expand by _expandAmount, at this interval of generations.

private static int _originalGenomeSize = _genomeSize;

#endregion

private static IFunction _functionGenerator = null; //new StringFunction(() => GetFitnessMethod(), (program, param) => OnFunctionStepComplete(program, param), _bestStatus, fitnessFunction, OnGeneration, _crossoverRate, _mutationRate, _genomeSize, _targetParams); /* Functions require setting BrainfuckVersion=2 in App.config */

/// <summary>

/// Selects the type of fitness algorithm to use (Hello World solutions, Calculation solutions, etc).

/// QUICK START GUIDE:

/// - Use the desired concrete Fitness class. For example: use StringOptimizedFitness() for a simple "Hello World" type of program.

///

/// return new StringOptimizedFitness(_ga, _maxIterationCount, _targetString)

/// return new AddFitness(_ga, _maxIterationCount)

/// return new SubtractFitness(_ga, _maxIterationCount)

/// return new ReverseStringFitness(_ga, _maxIterationCount)

/// return new HelloUserFitness(_ga, _maxIterationCount, _targetString)

///

/// </summary>

/// <returns>IFitness</returns>

private static IFitness GetFitnessMethod()

{

return new StringStrictFitness(_ga, _maxIterationCount, _targetParams.TargetString, _appendCode);

}

/// <summary>

/// Callback handler for each time a function completes solving a step of its process. For example, solving a word within a sentence, etc.

/// </summary>

/// <param name="program">Complete append code generated so far. This can be set as the value for _appendCode to generate programs.</param>

/// <param name="param">Optional parameter supplied by Function to indicate what has been solved (i.e., the term, sentence, numeric value, etc).</param>

private static void OnFunctionStepComplete(string program, object param)

{

// Reset genome size back to its original value for subsequent solving steps.

_genomeSize = _originalGenomeSize;

_ga.GAParams.GenomeSize = _genomeSize;

// Reset timer.

_startTime = DateTime.Now;

// Display generated code so far.

Console.WriteLine(param.ToString() + "\n" + program);

}

#region Worker Methods

/// <summary>

/// Event handler that is called upon each generation. We use this opportunity to display some status info and save the current genetic algorithm in case of crashes etc.

/// </summary>

private static void OnGeneration(GA ga)

{

if (_bestStatus.Iteration++ > 1000)

{

_bestStatus.Iteration = 0;

Console.WriteLine("Best Fitness: " + _bestStatus.TrueFitness + "/" + _targetParams.TargetFitness + " " + Math.Round(_bestStatus.TrueFitness / _targetParams.TargetFitness * 100, 2) + "%, Ticks: " + _bestStatus.Ticks + ", Total Ticks: " + _bestStatus.TotalTicks + ", Running: " + Math.Floor((DateTime.Now - _startTime).TotalSeconds / 60) + "m " + Math.Round(((DateTime.Now - _startTime).TotalSeconds % 60)) + "s, Size: " + _genomeSize + ", Best Output: " + _bestStatus.Output + ", Changed: " + _bestStatus.LastChangeDate.ToString() + ", Program: " + _bestStatus.Program);

ga.Save("my-genetic-algorithm.dat");

}

if (_expandAmount > 0 && ga.GAParams.CurrentGeneration > 0 && ga.GAParams.CurrentGeneration % _expandRate == 0 && _genomeSize < _maxGenomeSize)

{

_genomeSize += _expandAmount;

ga.GAParams.GenomeSize = _genomeSize;

_bestStatus.Fitness = 0; // Update display of best program, since genome has changed and we have a better/worse new best fitness.

}

}

/// <summary>

/// Fitness function to evaluate the current genetic algorithm. We decode the weights, run the resulting program, and score the output.

/// </summary>

/// <param name="weights">Array of double (genes), where each value cooresponds to a Brainfuck program command.</param>

/// <returns>double, indicating the score</returns>

private static double fitnessFunction(double[] weights)

{

// Get the selected fitness type.

IFitness myFitness = GetFitnessMethod();

// Get the fitness score.

double fitness = myFitness.GetFitness(weights);

// Is this a new best fitness?

if (fitness > _bestStatus.Fitness)

{

_bestStatus.Fitness = fitness;

_bestStatus.TrueFitness = myFitness.Fitness;

_bestStatus.Output = myFitness.Output;

_bestStatus.LastChangeDate = DateTime.Now;

_bestStatus.Program = myFitness.Program;

_bestStatus.Ticks = myFitness.Ticks;

_bestStatus.TotalTicks = myFitness.TotalTicks;

}

return fitness;

}

#endregion

#region Utility Methods

/// <summary>

/// Instantiates a fitness method.

/// This includes initializing any custom GA settings, as specified by the fitness.

/// This allows for reuse of the same GA settings when running a fitness method.

/// </summary>

/// <returns>IFitness</returns>

private static IFitness GetFitness()

{

IFitness fitness = GetFitnessMethod();

_appendCode += fitness.AppendCode;

_ga.GAParams.CrossoverRate = _crossoverRate = fitness.CrossoverRate.HasValue ? fitness.CrossoverRate.Value : _ga.GAParams.CrossoverRate;

_ga.GAParams.MutationRate = _mutationRate = fitness.MutationRate.HasValue ? fitness.MutationRate.Value : _ga.GAParams.MutationRate;

_ga.GAParams.GenomeSize = _genomeSize = fitness.GenomeSize.HasValue ? fitness.GenomeSize.Value : _ga.GAParams.GenomeSize;

_maxGenomeSize = fitness.MaxGenomeSize.HasValue ? fitness.MaxGenomeSize.Value : _maxGenomeSize;

_maxIterationCount = fitness.MaxIterationCount.HasValue ? fitness.MaxIterationCount.Value : _maxIterationCount;

_expandAmount = fitness.ExpandAmount.HasValue ? fitness.ExpandAmount.Value : _expandAmount;

_expandRate = fitness.ExpandRate.HasValue ? fitness.ExpandRate.Value : _expandRate;

// Re-initialize the fitness method for any changed internal values.

fitness = GetFitnessMethod();

return fitness;

}

#endregion

#region Main Program

/// <summary>

/// Main program.

/// </summary>

static void Main(string[] args)

{

// Genetic algorithm setup.

_ga = new GA(_crossoverRate, _mutationRate, 100, 10000000, _genomeSize);

if (_functionGenerator != null)

{

// Generate additional functions.

_appendCode += _functionGenerator.Generate(_ga);

}

// Generate main program. Instantiate the fitness method.

IFitness myFitness = GetFitness();

// Get the target fitness for this method.

_targetParams.TargetFitness = myFitness.TargetFitness;

// Run the genetic algorithm and get the best brain.

string program = GAManager.Run(_ga, fitnessFunction, OnGeneration);

// Append any functions.

if (!string.IsNullOrEmpty(_appendCode))

{

program += "@" + _appendCode;

}

// Display the final program.

Console.WriteLine(program);

Console.WriteLine();

// Compile to executable.

BrainPlus.Compile(program, "output.exe", myFitness);

// Run the result for the user.

string result = myFitness.RunProgram(program);

Console.WriteLine(result);

Console.ReadKey();

}

#endregion

}