Merge pull request #2301 from joeljosephjin/patch-1

Some minor code improvements.

src/mlpack/methods/reinforcement_learning/environment/acrobot.hpp

@@ -169,10 +169,10 @@ class Acrobot
     nextState.Theta2() = Wrap(currentNextState[1], -M_PI, M_PI);
 
     //! The value of angular velocity is bounded in min and max value.
-    nextState.AngularVelocity1() = std::min(
-        std::max(currentNextState[2], -maxVel1), maxVel1);
-    nextState.AngularVelocity2() = std::min(
-        std::max(currentNextState[3], -maxVel2), maxVel2);
+    nextState.AngularVelocity1() = math::ClampRange(currentNextState[2],
+        -maxVel1, maxVel1);
+    nextState.AngularVelocity2() = math::ClampRange(currentNextState[3],
+        -maxVel2, maxVel2);
 
     // Check if the episode has terminated.
     bool done = IsTerminal(nextState);

src/mlpack/methods/reinforcement_learning/environment/continuous_mountain_car.hpp

@@ -18,6 +18,7 @@
 #define MLPACK_METHODS_RL_ENVIRONMENT_CONTINUOUS_MOUNTAIN_CAR_HPP
 
 #include <mlpack/prereqs.hpp>
+#include <mlpack/core/math/clamp.hpp>
 
 namespace mlpack {
 namespace rl {
@@ -96,7 +97,7 @@ class ContinuousMountainCar
    * @param positionGoal Final target position.
    * @param velocityMin Minimum legal velocity.
    * @param velocityMax Maximum legal velocity.
-   * @param power Power generated by car.
+   * @param duration Time Duration for which force is applied on the car.
    * @param doneReward Reward recieved by the agent on success.
    * @param maxSteps The number of steps after which the episode
    *    terminates. If the value is 0, there is no limit.
@@ -106,7 +107,7 @@ class ContinuousMountainCar
                         const double positionGoal = 0.45,
                         const double velocityMin = -0.07,
                         const double velocityMax = 0.07,
-                        const double power = 0.0015,
+                        const double duration = 0.0015,
                         const double doneReward = 100,
                         const size_t maxSteps = 0) :
       positionMin(positionMin),
@@ -114,7 +115,7 @@ class ContinuousMountainCar
       positionGoal(positionGoal),
       velocityMin(velocityMin),
       velocityMax(velocityMax),
-      power(power),
+      duration(duration),
       doneReward(doneReward),
       maxSteps(maxSteps),
       stepsPerformed(0)
@@ -136,16 +137,16 @@ class ContinuousMountainCar
     stepsPerformed++;
 
     // Calculate acceleration.
-    double force = std::min(std::max(action.action[0], -1.0), 1.0);
+    double force = math::ClampRange(action.action[0], -1.0, 1.0);
 
     // Update states.
-    nextState.Velocity() = state.Velocity() + force * power - 0.0025 *
+    nextState.Velocity() = state.Velocity() + force * duration - 0.0025 *
         std::cos(3 * state.Position());
-    nextState.Velocity() = std::min(
-        std::max(nextState.Velocity(), velocityMin), velocityMax);
+    nextState.Velocity() = math::ClampRange(nextState.Velocity(),
+      velocityMin, velocityMax);
     nextState.Position() = state.Position() + nextState.Velocity();
-    nextState.Position() = std::min(
-        std::max(nextState.Position(), positionMin), positionMax);
+    nextState.Position() = math::ClampRange(nextState.Position(),
+      positionMin, positionMax);
     if (nextState.Position() == positionMin && nextState.Velocity() < 0)
       nextState.Velocity() = 0.0;
 
@@ -236,8 +237,8 @@ class ContinuousMountainCar
   //! Locally-stored maximum legal velocity.
   double velocityMax;
 
-  //! Locally-stored power.
-  double power;
+  //! Locally-stored duration.
+  double duration;
 
   //! Locally-stored done reward.
   double doneReward;

src/mlpack/methods/reinforcement_learning/environment/mountain_car.hpp

@@ -17,6 +17,7 @@
 #define MLPACK_METHODS_RL_ENVIRONMENT_MOUNTAIN_CAR_HPP
 
 #include <mlpack/prereqs.hpp>
+#include <mlpack/core/math/clamp.hpp>
 
 namespace mlpack {
 namespace rl {
@@ -134,13 +135,13 @@ class MountainCar
     int direction = action - 1;
     nextState.Velocity() = state.Velocity() + 0.001 * direction - 0.0025 *
         std::cos(3 * state.Position());
-    nextState.Velocity() = std::min(
-        std::max(nextState.Velocity(), velocityMin), velocityMax);
+    nextState.Velocity() = math::ClampRange(nextState.Velocity(),
+        velocityMin, velocityMax);
 
     // Update states.
     nextState.Position() = state.Position() + nextState.Velocity();
-    nextState.Position() = std::min(
-        std::max(nextState.Position(), positionMin), positionMax);
+    nextState.Position() = math::ClampRange(nextState.Position(),
+        positionMin, positionMax);
 
     if (nextState.Position() == positionMin && nextState.Velocity() < 0)
       nextState.Velocity() = 0.0;

src/mlpack/methods/reinforcement_learning/environment/pendulum.hpp

@@ -17,6 +17,7 @@
 #define MLPACK_METHODS_RL_ENVIRONMENT_PENDULUM_HPP
 
 #include <mlpack/prereqs.hpp>
+#include <mlpack/core/math/clamp.hpp>
 
 namespace mlpack {
 namespace rl {
@@ -140,8 +141,7 @@ class Pendulum
     const double length = 1.0;
 
     // Get action and clip the values between max and min limits.
-    double torque = std::min(
-        std::max(action.action[0], -maxTorque), maxTorque);
+    double torque = math::ClampRange(action.action[0], -maxTorque, maxTorque);
 
     // Calculate costs of taking this action in the current state.
     double costs = std::pow(AngleNormalize(theta), 2) + 0.1 *
@@ -151,8 +151,8 @@ class Pendulum
     double newAngularVelocity = angularVelocity + (-3.0 * gravity / (2 *
         length) * std::sin(theta + M_PI) + 3.0 / std::pow(mass * length, 2) *
         torque) * dt;
-    nextState.AngularVelocity() = std::min(std::max(newAngularVelocity,
-        -maxAngularVelocity), maxAngularVelocity);
+    nextState.AngularVelocity() = math::ClampRange(newAngularVelocity,
+        -maxAngularVelocity, maxAngularVelocity);
     nextState.Theta() = theta + newAngularVelocity * dt;
 
     // Check if the episode has terminated

src/mlpack/methods/reinforcement_learning/environment/reward_clipping.hpp

@@ -13,6 +13,7 @@
 #define MLPACK_METHODS_RL_ENVIRONMENT_REWARD_CLIPPING_HPP
 
 #include <mlpack/prereqs.hpp>
+#include <mlpack/core/math/clamp.hpp>
 
 namespace mlpack {
 namespace rl {
@@ -91,7 +92,7 @@ class RewardClipping
     // Get original unclipped reward from base environment.
     double unclippedReward =  environment.Sample(state, action, nextState);
     // Clip rewards according to the min and max limit and return.
-    return std::min(std::max(unclippedReward, minReward), maxReward);
+    return math::ClampRange(unclippedReward, minReward, maxReward);
   }
 
   /**