aa

梯度下降/最简单函数求极值tensor.py

@@ -0,0 +1,39 @@
+import torch
+import torch.optim
+import matplotlib.pyplot as plt
+
+
+def f(x):
+      return x**2-3
+
+def df(x):
+      return 2*x
+def plotf(loss):
+      x = range(len(loss))
+      plt.plot(x,loss)
+      plt.xlabel('Iteration')
+      plt.ylabel('Loss')
+      plt.show()
+
+
+
+def main():
+      x = torch.tensor([15.],requires_grad=True)
+      optimizer = torch.optim.SGD([x,],lr = 0.1,momentum=0.9)
+      steps = 400
+      loss = []
+      for i in range(steps):
+            optimizer.zero_grad()
+            f(x).backward()
+            optimizer.step()
+            loss.append(f(x))
+            print(loss[i])
+      y = f(x)
+      print("函数最小值是： ",y)
+      plotf(loss)
+
+
+
+if __name__ == '__main__':
+    main()
+

梯度下降/牛顿法求极值tensor.py

@@ -0,0 +1,11 @@
+import torch
+from torch.autograd import Variable
+def f(x):
+    y = x ** 2 -3
+    return y
+x = Variable(torch.Tensor([15]), requires_grad=True)
+for i in range(40):
+    grad_x = torch.autograd.grad(f(x), x, create_graph=True)
+    grad_grad_x = torch.autograd.grad(grad_x[0], x)
+    x = Variable(x.data - grad_x[0].data / grad_grad_x[0].data, requires_grad=True)
+print(x)