new

Python高级编程/函数可变参数.py

@@ -0,0 +1,11 @@
+def change(age,*som):
+         print(age)
+         for i in som:
+           print(i)
+         return
+
+change(12,'name','year','mon','address')
+
+change('a1','b1')
+
+change('a2','b2',11)

优化算法/AA.py

@@ -0,0 +1,77 @@
+#coding:utf-8
+# https://mp.weixin.qq.com/s/ac-CgZj-avmPBraVvQUuBQ
+from __future__ import print_function
+import numpy as np
+import matplotlib.pyplot as plt
+class Function():
+    def __init__(self):
+        self.points_x = np.linspace(-20, 20, 1000)
+        self.points_y = self.f(self.points_x)
+
+    def f(self,x):
+        return (0.15*x)**2 + np.cos(x) + np.sin(3*x)/3 + np.cos(5*x)/5 + np.sin(7*x)/7
+
+    def df(self,x):
+        return (9/200)*x - np.sin(x) -np.sin(5*x) + np.cos(3*x) + np.cos(7*x)
+
+
+    def ddf(self,x):
+        return (9/200) - np.cos(x) -3*np.sin(x) - 5*np.cos(5*x) -7* np.sin(7*x)
+
+
+
+
+# #
+# #
+# # # AdaGrad
+# # for i in range(15):
+# #     # 绘制原来的函数
+# #     plt.plot(points_x, points_y, c="b", alpha=0.5, linestyle="-")
+# #     # 算法开始
+# #     lr = pow(1.5,-i)*32
+# #     delta = 1e-7
+# #     x = -20
+# #     r = 0
+# #     AdaGrad_x, AdaGrad_y = [], []
+# #     for it in range(1000):
+# #         AdaGrad_x.append(x), AdaGrad_y.append(f(x))
+# #         g = df(x)
+# #         r = r + g*g # 积累平方梯度
+# #         x = x - lr /(delta + np.sqrt(r)) * g
+# #
+# #     plt.xlim(-20, 20)
+# #     plt.ylim(-2, 10)
+# #     plt.plot(AdaGrad_x, AdaGrad_y, c="r", linestyle="-")
+# #     plt.scatter(AdaGrad_x[-1],AdaGrad_y[-1],90,marker = "x",color="g")
+# #     plt.title("AdaGrad,lr=%f"%(lr))
+# #     plt.savefig("AdaGrad,lr=%f"%(lr) + ".png")
+# #     plt.clf()
+# #
+# #
+# # # RMSProp
+# # for i in range(15):
+# #     # 绘制原来的函数
+# #     plt.plot(points_x, points_y, c="b", alpha=0.5, linestyle="-")
+# #     # 算法开始
+# #     lr = pow(1.5,-i)*32
+# #     delta = 1e-6
+# #     rou = 0.8
+# #     x = -20
+# #     r = 0
+# #     RMSProp_x, RMSProp_y = [], []
+# #     for it in range(1000):
+# #         RMSProp_x.append(x), RMSProp_y.append(f(x))
+# #         g = df(x)
+# #         r = rou * r + (1-rou)*g*g # 积累平方梯度
+# #         x = x - lr /(delta + np.sqrt(r)) * g
+# #
+# #     plt.xlim(-20, 20)
+# #     plt.ylim(-2, 10)
+# #     plt.plot(RMSProp_x, RMSProp_y, c="r", linestyle="-")
+# #     plt.scatter(RMSProp_x[-1],RMSProp_y[-1],90,marker = "x",color="g")
+# #     plt.title("RMSProp,lr=%f,rou=%f"%(lr,rou))
+# #     plt.savefig("RMSProp,lr=%f,rou=%f"%(lr,rou) + ".png")
+# #     plt.clf()
+# #
+
+# #

实例学习Numpy与Matplotlib/Numpy合并.py

@@ -12,6 +12,9 @@ print(b1.shape,b2.shape,b3.shape)
 b4 = np.concatenate([b2, b3])
 print(f"b4 = {b4} \n  ")
 
+b5 = np.vstack([b2, b3])
+print(f"b5 = {b5} \n  ")
+
 
 
 

梯度下降/最简单函数求极值.py

@@ -14,17 +14,15 @@ def plotf(loss):
       plt.ylabel('Loss')
       plt.show()
 
-
-
 def main():
       x = 15
       lr = 0.1
       steps = 40
       loss = []
       for i in range(steps):
-            loss.append(f(x))
-            print(loss[i])
             x = x-lr*df(x)
+            loss.append(f(x))
+      print(loss[i])
       # y = f(x)
       # print(y)
       plotf(loss)

梯度下降/最简单函数求极值tensor.py

@@ -27,7 +27,7 @@ def main():
             f(x).backward()
             optimizer.step()
             loss.append(f(x))
-            print(loss[i])
+            print(f(x))
       y = f(x)
       print("函数最小值是： ",y)
       plotf(loss)