indent

examples/example-004-001-cliff-onestep.cc

@@ -4,7 +4,7 @@
 
 #include <rl.hpp>
 #include <random>
- 
+
 
 #define NB_EPISODES    3000
 
@@ -30,90 +30,90 @@ int main(int argc, char* argv[]) {
     std::random_device rd;
     std::mt19937 gen(rd());
 
-  // 1) Instantiate the simulator
-  Param param;
-  Simulator simulator(param);
-
-  // 2) Instantiate the ActorCritic
-  auto action_begin = rl::enumerator<A>(rl::problem::cliff_walking::Action::actionNorth);
-  auto action_end = action_begin + rl::problem::cliff_walking::actionSize;
-  unsigned int nb_features = Cliff::size;
-  Architecture archi(nb_features,
-		     [](const S& s) { return s;},
-		     action_begin, action_end, gen);
-
-  // 3) Instantiate the learner
-  Learner learner(archi, paramGAMMA, paramALPHA_V, paramALPHA_P);
-
-  // 4) run NB_EPISODES episodes
-  unsigned int episode;
-  unsigned int step;
-  A action;
-  S state, next;
-  double rew;
-
-  std::cout << "Learning " << std::endl;
-  for(episode = 0 ;episode < NB_EPISODES; ++episode) {
-    simulator.restart();
-    learner.restart();
-    step = 0;
-    std::cout << '\r' << "Episode " << episode << std::flush;
-
-    state = simulator.sense();
-
-    while(true) {
-      action = archi.sample_action(state);
-      try {
-	// The following may raise a Terminal state exception
-	simulator.timeStep(action);
-
-	rew = simulator.reward();
-	next = simulator.sense();
-
-	learner.learn(state, action, rew, next);
-
-	state = next;
-	++step;
-      }
-      catch(rl::exception::Terminal& e) { 
-	learner.learn(state, action, simulator.reward());
-	break;
-      }
+    // 1) Instantiate the simulator
+    Param param;
+    Simulator simulator(param);
+
+    // 2) Instantiate the ActorCritic
+    auto action_begin = rl::enumerator<A>(rl::problem::cliff_walking::Action::actionNorth);
+    auto action_end = action_begin + rl::problem::cliff_walking::actionSize;
+    unsigned int nb_features = Cliff::size;
+    Architecture archi(nb_features,
+            [](const S& s) { return s;},
+            action_begin, action_end, gen);
+
+    // 3) Instantiate the learner
+    Learner learner(archi, paramGAMMA, paramALPHA_V, paramALPHA_P);
+
+    // 4) run NB_EPISODES episodes
+    unsigned int episode;
+    unsigned int step;
+    A action;
+    S state, next;
+    double rew;
+
+    std::cout << "Learning " << std::endl;
+    for(episode = 0 ;episode < NB_EPISODES; ++episode) {
+        simulator.restart();
+        learner.restart();
+        step = 0;
+        std::cout << '\r' << "Episode " << episode << std::flush;
+
+        state = simulator.sense();
+
+        while(true) {
+            action = archi.sample_action(state);
+            try {
+                // The following may raise a Terminal state exception
+                simulator.timeStep(action);
+
+                rew = simulator.reward();
+                next = simulator.sense();
+
+                learner.learn(state, action, rew, next);
+
+                state = next;
+                ++step;
+            }
+            catch(rl::exception::Terminal& e) { 
+                learner.learn(state, action, simulator.reward());
+                break;
+            }
+        }
     }
-  }
-  std::cout << std::endl;
-
-  std::cout << "Testing the learned policy" << std::endl;
-  // After this training phase, we test the policy
-  unsigned int nb_test_episodes = 1000;
-  double cum_length = 0.0;
-  for(unsigned int i = 0 ; i < nb_test_episodes; ++i) {
-    simulator.restart();
-    step = 0;
-    state = simulator.sense();
-    while(true) {
-      action = archi.sample_action(state);
-      try {
-	// The following may raise a Terminal state exception
-	simulator.timeStep(action);
-	state = simulator.sense();
-	++step;
-      }
-      catch(rl::exception::Terminal& e) { 
-	break;
-      }
+    std::cout << std::endl;
+
+    std::cout << "Testing the learned policy" << std::endl;
+    // After this training phase, we test the policy
+    unsigned int nb_test_episodes = 1000;
+    double cum_length = 0.0;
+    for(unsigned int i = 0 ; i < nb_test_episodes; ++i) {
+        simulator.restart();
+        step = 0;
+        state = simulator.sense();
+        while(true) {
+            action = archi.sample_action(state);
+            try {
+                // The following may raise a Terminal state exception
+                simulator.timeStep(action);
+                state = simulator.sense();
+                ++step;
+            }
+            catch(rl::exception::Terminal& e) { 
+                break;
+            }
+        }
+        cum_length += step;
     }
-    cum_length += step;
-  }
-  std::cout << "The mean length of "<< nb_test_episodes
-	    <<" testing episodes is " << cum_length / double(nb_test_episodes) << std::endl;
-
-  // And let us display the action probabilities for the first state :
-  std::cout << "The probabilities of the actions of the learned controller, in the start state are :" << std::endl;
-  auto proba = archi.get_action_probabilities(0);
-  std::cout << "P(North/s=start) = " << proba[rl::problem::cliff_walking::Action::actionNorth] << std::endl;
-  std::cout << "P(East/s=start) = " << proba[rl::problem::cliff_walking::Action::actionEast] << std::endl;
-  std::cout << "P(South/s=start) = " << proba[rl::problem::cliff_walking::Action::actionSouth] << std::endl;
-  std::cout << "P(West/s=start) = " << proba[rl::problem::cliff_walking::Action::actionWest] << std::endl;
-
+    std::cout << "The mean length of "<< nb_test_episodes
+        <<" testing episodes is " << cum_length / double(nb_test_episodes) << std::endl;
+
+    // And let us display the action probabilities for the first state :
+    std::cout << "The probabilities of the actions of the learned controller, in the start state are :" << std::endl;
+    auto proba = archi.get_action_probabilities(0);
+    std::cout << "P(North/s=start) = " << proba[rl::problem::cliff_walking::Action::actionNorth] << std::endl;
+    std::cout << "P(East/s=start) = " << proba[rl::problem::cliff_walking::Action::actionEast] << std::endl;
+    std::cout << "P(South/s=start) = " << proba[rl::problem::cliff_walking::Action::actionSouth] << std::endl;
+    std::cout << "P(West/s=start) = " << proba[rl::problem::cliff_walking::Action::actionWest] << std::endl;
+
 }