PPO.py

############################### Import libraries ###############################


import os
import glob
import time
from datetime import datetime

import torch
import torch.nn as nn
from torch.distributions import MultivariateNormal
from torch.distributions import Categorical

import numpy as np

import gym



################################## PPO Policy ##################################


class RolloutBuffer:
    def __init__(self):
        self.actions = []
        self.states = []
        self.logprobs = []
        self.rewards = []
        self.is_terminals = []
    

    def clear(self):
        del self.actions[:]
        del self.states[:]
        del self.logprobs[:]
        del self.rewards[:]
        del self.is_terminals[:]


class ActorCritic(nn.Module):
    def __init__(self, actor,critic, has_continuous_action_space, action_std_init,device):
        super(ActorCritic, self).__init__()
        action_dim =  1
        self.has_continuous_action_space = has_continuous_action_space
        self.device = device
        if has_continuous_action_space:
            self.action_dim = action_dim
            self.action_var = torch.full((action_dim,), action_std_init * action_std_init).to(self.device)

        # actor
        
        self.actor = actor

        
        # critic
        self.critic = critic
        
    def set_action_std(self, new_action_std):
        action_dim = 1
        if self.has_continuous_action_space:
            self.action_var = torch.full((self.action_dim,), new_action_std * new_action_std).to(self.device)
        else:
            print("--------------------------------------------------------------------------------------------")
            print("WARNING : Calling ActorCritic::set_action_std() on discrete action space policy")
            print("--------------------------------------------------------------------------------------------")


    def forward(self):
        raise NotImplementedError
    

    def act(self, state):

        if self.has_continuous_action_space:
            action_mean = self.actor(state)
            cov_mat = torch.diag(self.action_var).unsqueeze(dim=0)
            dist = MultivariateNormal(action_mean, cov_mat)
        else:
            action_probs = self.actor(state)
            dist = Categorical(action_probs)
        action = dist.sample()
        action_logprob = dist.log_prob(action)
        
        return action.detach(), action_logprob.detach()
    

    def evaluate(self, state, action):

        if self.has_continuous_action_space:
            action_mean = self.actor(state)
            action_var = self.action_var.expand_as(action_mean)
            cov_mat = torch.diag_embed(action_var).to(self.device)
            dist = MultivariateNormal(action_mean, cov_mat)
            
            # for single action continuous environments
            if self.action_dim == 1:
                action = action.reshape(-1, self.action_dim)

        else:
            action_probs = self.actor(state)
            dist = Categorical(action_probs.detach())

        action_logprobs = dist.log_prob(action)
        dist_entropy = dist.entropy()
        state_values = self.critic(state)
        
        
        return action_logprobs, state_values, dist_entropy


class PPO:
    def __init__(self, actor,critic, lr_actor, lr_critic, gamma, K_epochs, eps_clip, has_continuous_action_space, device,action_std_init=0.6):

        self.has_continuous_action_space = has_continuous_action_space
        self.device = device
        if has_continuous_action_space:
            print(action_std_init)
            self.action_std = action_std_init

        self.gamma = gamma
        self.eps_clip = eps_clip
        self.K_epochs = K_epochs
        
        self.buffer = RolloutBuffer()

        self.policy = ActorCritic(actor,critic, has_continuous_action_space, action_std_init,device).to(self.device)
        self.optimizer = torch.optim.Adam([
                        {'params': self.policy.actor.parameters(), 'lr': lr_actor},
                        {'params': self.policy.critic.parameters(), 'lr': lr_critic}
                    ])

        self.policy_old = ActorCritic(actor,critic, has_continuous_action_space, action_std_init,device).to(self.device)
        self.policy_old.load_state_dict(self.policy.state_dict())
        
        self.MseLoss = nn.MSELoss()


    def set_action_std(self, new_action_std):
        
        if self.has_continuous_action_space:
            self.action_std = new_action_std
            self.policy.set_action_std(new_action_std)
            self.policy_old.set_action_std(new_action_std)
        
        else:
            print("--------------------------------------------------------------------------------------------")
            print("WARNING : Calling PPO::set_action_std() on discrete action space policy")
            print("--------------------------------------------------------------------------------------------")


    def decay_action_std(self, action_std_decay_rate, min_action_std):
        print("--------------------------------------------------------------------------------------------")

        if self.has_continuous_action_space:
            self.action_std = self.action_std - action_std_decay_rate
            self.action_std = round(self.action_std, 4)
            if (self.action_std <= min_action_std):
                self.action_std = min_action_std
                print("setting actor output action_std to min_action_std : ", self.action_std)
            else:
                print("setting actor output action_std to : ", self.action_std)
            self.set_action_std(self.action_std)

        else:
            print("WARNING : Calling PPO::decay_action_std() on discrete action space policy")

        print("--------------------------------------------------------------------------------------------")


    def select_action(self, state):

        if self.has_continuous_action_space:
            with torch.no_grad():
                state = torch.FloatTensor(state).to(self.device)
                action, action_logprob = self.policy_old.act(state)

            self.buffer.states.append(state)
            self.buffer.actions.append(action)
            self.buffer.logprobs.append(action_logprob)

            return action.detach().cpu().numpy().flatten()

        else:
            with torch.no_grad():
                state = torch.FloatTensor(state).to(self.device)
                action, action_logprob = self.policy_old.act(state)
            
            self.buffer.states.append(state)
            self.buffer.actions.append(action)
            self.buffer.logprobs.append(action_logprob)

            return action.item()


    def update(self):

        # Monte Carlo estimate of returns
        rewards = []
        discounted_reward = 0

        for reward, is_terminal in zip(reversed(self.buffer.rewards), reversed(self.buffer.is_terminals)):
            if is_terminal:
                discounted_reward = 0
            discounted_reward = reward + (self.gamma * discounted_reward)
            rewards.insert(0, discounted_reward)
 
        # Normalizing the rewards
        rewards = torch.tensor(rewards, dtype=torch.float32).to(self.device)

        if len(rewards) > 1:
            rewards = (rewards - rewards.mean()) / (rewards.std() + 1e-7)
        else:
            rewards = (rewards - rewards.mean()) 

        # convert list to tensor
        old_states = torch.squeeze(torch.stack(self.buffer.states, dim=0),dim=1).detach().to(self.device)
        old_actions = torch.squeeze(torch.stack(self.buffer.actions, dim=0),dim=1).detach().to(self.device)
        old_logprobs = torch.squeeze(torch.stack(self.buffer.logprobs, dim=0),dim=1).detach().to(self.device)

        
        # Optimize policy for K epochs
        for _ in range(self.K_epochs):

            # Evaluating old actions and values
            logprobs, state_values, dist_entropy = self.policy.evaluate(old_states, old_actions)
            
            # match state_values tensor dimensions with rewards tensor
            state_values = torch.squeeze(state_values,dim=0)

            # Finding the ratio (pi_theta / pi_theta__old)
            ratios = torch.exp(logprobs - old_logprobs.detach())

            # Finding Surrogate Loss
            advantages = rewards - state_values.detach()   
            surr1 = ratios * advantages
            surr2 = torch.clamp(ratios, 1-self.eps_clip, 1+self.eps_clip) * advantages
            # final loss of clipped objective PPO
            loss = -torch.min(surr1, surr2) + 0.5*self.MseLoss(state_values, rewards) - 0.01*dist_entropy
            
            # take gradient step
            self.optimizer.zero_grad()
            loss.mean().backward()
            self.optimizer.step()
            
        # Copy new weights into old policy
        self.policy_old.load_state_dict(self.policy.state_dict())

        # clear buffer
        self.buffer.clear()
    
    
    def save(self, checkpoint_path):
        torch.save(self.policy_old.state_dict(), checkpoint_path)
   

    def load(self, checkpoint_path):
        self.policy_old.load_state_dict(torch.load(checkpoint_path, map_location=lambda storage, loc: storage))
        self.policy.load_state_dict(torch.load(checkpoint_path, map_location=lambda storage, loc: storage))