pytorch架构的DDPG算法代码

以下是基于PyTorch的DDPG算法代码：

首先，需要导入相关的包：

import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
import numpy as np
import random

from collections import deque

接下来，我们需要定义Actor和Critic网络：

# 定义Actor网络
class Actor(nn.Module):
    def __init__(self, state_dim, action_dim, max_action):
        super(Actor, self).__init__()
        self.fc1 = nn.Linear(state_dim, 400)
        self.fc2 = nn.Linear(400, 300)
        self.fc3 = nn.Linear(300, action_dim)
        self.max_action = max_action
        
    def forward(self, state):
        x = F.relu(self.fc1(state))
        x = F.relu(self.fc2(x))
        x = self.max_action * torch.tanh(self.fc3(x))
        return x
    
# 定义Critic网络
class Critic(nn.Module):
    def __init__(self, state_dim, action_dim):
        super(Critic, self).__init__()
        self.fc1 = nn.Linear(state_dim + action_dim, 400)
        self.fc2 = nn.Linear(400 , 300)
        self.fc3 = nn.Linear(300, 1)
        
    def forward(self, state, action):
        x = torch.cat([state, action], 1)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x

接下来，我们定义DDPG算法的主要函数：

class DDPG(object):
    def __init__(self, state_dim, action_dim, max_action):
        # 定义Actor和Critic
        self.actor = Actor(state_dim, action_dim, max_action).to(device)
        self.actor_target = Actor(state_dim, action_dim, max_action).to(device)
        self.actor_target.load_state_dict(self.actor.state_dict())
        self.actor_optimizer = optim.Adam(self.actor.parameters(), lr=1e-4)
        
        self.critic = Critic(state_dim, action_dim).to(device)
        self.critic_target = Critic(state_dim, action_dim).to(device)
        self.critic_target.load_state_dict(self.critic.state_dict())
        self.critic_optimizer = optim.Adam(self.critic.parameters(), lr=1e-3)
        
        # 定义一些超参数
        self.state_dim = state_dim
        self.action_dim = action_dim
        self.max_action = max_action
        
        self.replay_buffer = deque(maxlen=1000000)
        self.batch_size = 64
        self.discount = 0.99
        self.tau = 0.005
        
    def select_action(self, state):
        state = torch.FloatTensor(state.reshape(1, -1)).to(device)
        return self.actor(state).cpu().data.numpy().flatten()
        
    def add_to_buffer(self, state, action, reward, next_state, done):
        self.replay_buffer.append((state, action, reward, next_state, done))
        
    def train(self):
        if len(self.replay_buffer) < self.batch_size:
            return
        
        # 从经验池中采样一个batch的数据
        batch = random.sample(self.replay_buffer, self.batch_size)
        state = torch.FloatTensor(np.array([e[0] for e in batch])).to(device)
        action = torch.FloatTensor(np.array([e[1] for e in batch])).to(device)
        reward = torch.FloatTensor(np.array([e[2] for e in batch])).to(device)
        next_state = torch.FloatTensor(np.array([e[3] for e in batch])).to(device)
        done = torch.FloatTensor(np.array([e[4] for e in batch])).to(device)
        
        # 计算target Q值
        target_Q = self.critic_target(next_state, self.actor_target(next_state))
        target_Q = reward + ((1 - done) * self.discount * target_Q).detach()
        
        # 计算当前Q值
        current_Q = self.critic(state, action)
        
        # 计算Critic损失
        critic_loss = F.mse_loss(current_Q, target_Q)
        
        # 更新Critic网络
        self.critic_optimizer.zero_grad()
        critic_loss.backward()
        self.critic_optimizer.step()
        
        # 计算Actor损失
        actor_loss = -self.critic(state, self.actor(state)).mean()
        
        # 更新Actor网络
        self.actor_optimizer.zero_grad()
        actor_loss.backward()
        self.actor_optimizer.step()
        
        # 更新Actor和Critic的目标网络
        for target_param, param in zip(self.critic_target.parameters(), self.critic.parameters()):
            target_param.data.copy_(self.tau * param.data + (1 - self.tau) * target_param.data)
            
        for target_param, param in zip(self.actor_target.parameters(), self.actor.parameters()):
            target_param.data.copy_(self.tau * param.data + (1 - self.tau) * target_param.data)

最后，我们可以使用DDPG算法来训练和测试模型：

# 定义一些环境参数
state_dim = 3
action_dim = 1
max_action = 1

# 创建DDPG对象
ddpg = DDPG(state_dim, action_dim, max_action)

# 训练模型
for i in range(10000):
    state = np.random.randn(state_dim)
    action = ddpg.select_action(state)
    next_state = np.random.randn(state_dim)
    reward = np.random.randn(1)
    done = np.random.randint(0, 2)
    
    ddpg.add_to_buffer(state, action, reward, next_state, done)
    ddpg.train()

# 测试模型
state = np.random.randn(state_dim)
action = ddpg.select_action(state)
print(action)

这就是基于PyTorch的DDPG算法代码。