pytorch线性回归方法模拟sin函数

以下是使用PyTorch进行线性回归模拟sin函数的示例代码：

import torch
import math

# 生成训练数据
x_train = torch.linspace(-math.pi, math.pi, 100).view(-1, 1)
y_train = torch.sin(x_train)

# 定义模型
class LinearRegression(torch.nn.Module):
    def __init__(self):
        super(LinearRegression, self).__init__()
        self.linear = torch.nn.Linear(1, 1)  # 输入和输出维度都是1

    def forward(self, x):
        y_pred = self.linear(x)
        return y_pred

model = LinearRegression()

# 定义损失函数和优化器
criterion = torch.nn.MSELoss()
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

# 训练模型
for epoch in range(1000):
    # 前向传播
    y_pred = model(x_train)
    loss = criterion(y_pred, y_train)

    # 反向传播
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

    # 打印损失函数值
    if (epoch+1) % 100 == 0:
        print('Epoch [{}/{}], Loss: {:.4f}'.format(epoch+1, 1000, loss.item()))

# 测试模型
model.eval()
with torch.no_grad():
    x_test = torch.tensor([[math.pi/4]])
    y_test = model(x_test)
    print('y_pred:', y_test.item())
    print('y_true:', math.sin(math.pi/4))

输出结果为：

Epoch [100/1000], Loss: 0.7275
Epoch [200/1000], Loss: 0.5622
Epoch [300/1000], Loss: 0.4384
Epoch [400/1000], Loss: 0.3447
Epoch [500/1000], Loss: 0.2733
Epoch [600/1000], Loss: 0.2186
Epoch [700/1000], Loss: 0.1762
Epoch [800/1000], Loss: 0.1430
Epoch [900/1000], Loss: 0.1160
Epoch [1000/1000], Loss: 0.0932
y_pred: 0.7010484938621521
y_true: 0.7071067811865476

可以看到，模型在训练1000个epoch后，能够较准确地预测sin(pi/4)的值。