static void Main(string[] args)
{
var env = new CliffWalkingEnvironment();
var sarsaAgent = new SarsaCliffWalker();
var values = sarsaAgent.ImproveEstimates(env, out var tdDiags, 10000);
System.Console.WriteLine("td 0 avg Values:");
StateActionValuesConsoleRenderer.RenderAverageValues(values);
System.Console.WriteLine("td 0 highest Values:");
StateActionValuesConsoleRenderer.RenderHighestValues(values);
System.Console.WriteLine("");
System.Console.WriteLine("td 0 Greedy path:");
ConsolePathRenderer.RenderPath(GreedyPath(env, values));
var qAgent = new QLearningCliffWalker();
var qValues = qAgent.ImproveEstimates(env, out var qDiags, 10000);
System.Console.WriteLine("");
System.Console.WriteLine("q learning avg Values:");
StateActionValuesConsoleRenderer.RenderAverageValues(qValues);
System.Console.WriteLine("q learning highest Values:");
StateActionValuesConsoleRenderer.RenderHighestValues(qValues);
System.Console.WriteLine("");
System.Console.WriteLine("q learning Greedy path:");
ConsolePathRenderer.RenderPath(GreedyPath(env, qValues));
}
public static void Run()
{
const int numEpisodes = 100;
var env = new CliffWalkingEnvironment();
var sarsaAgent = new SarsaCliffWalker(0.1, 0.1);
var qLearningAgent = new QLearningCliffWalker(0.1, 0.1);
var tdAverageRewards = CollectAverageRewardSums(sarsaAgent, env, numEpisodes);
var qLearningAverageRewards = CollectAverageRewardSums(qLearningAgent, env, numEpisodes);
var plotter = new Plotter();
var plt = plotter.Plt;
plt.Title("Average total reward per episode");
var dataX = Enumerable.Range(0, numEpisodes).Select(i => (double)i).ToArray();
plt.PlotScatter(dataX, tdAverageRewards, label: "TD 0 (Sarsa)");
plt.PlotScatter(dataX, qLearningAverageRewards, label: "Q learning");
plt.XLabel("Episodes");
plt.YLabel("Average total reward");
plt.Legend();
plotter.Show();
}
