1、实验目的和要求
实验目的:1.理解梯度下降的含义及作用。
- 使用python实现梯度下降。
- 使用python的3D绘图绘制梯度下降的结果。
实验要求:
书写提交实验报告
2、实验环境
Python3.9
Pycharm2022
Window10
3、实验内容:
使用gradient_descent()函数和python的Axes3D类绘制梯度下降法求函数f=x^2+y^2的最小值过程的散点图。
#coding:utf-8
import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d.axes3d import Axes3D
# 设置显示中文字体
from plotnine.themes.seaborn_rcmod import mpl
mpl.rcParams["font.sans-serif"] = ["SimHei"]
# 设置正常显示符号
mpl.rcParams["axes.unicode_minus"] = False
def _numerical_gradient_no_batch(f, x):
h = 1e-4 # 0.0001
grad = np.zeros_like(x)
for idx in range(x.size):
tmp_val = x[idx]
x[idx] = float(tmp_val) + h
fxh1 = f(x) # f(x+h)
x[idx] = tmp_val - h
fxh2 = f(x) # f(x-h)
grad[idx] = (fxh1 - fxh2) / (2 * h)
x[idx] = tmp_val # 还原值
return grad
#求梯度
def numerical_gradient(f, X):
if X.ndim == 1:
return _numerical_gradient_no_batch(f, X)
else:
grad = np.zeros_like(X)
for idx, x in enumerate(X):
grad[idx] = _numerical_gradient_no_batch(f, x)
return grad
def f(x,y):
return(1.0*x**2 + 1.0*y**2 )
#求最小值
def gradient_descent(f, init_x, lr=0.01, step_num=100):
x = init_x
x_history = []
for i in range(step_num):
x_history.append( x.copy() )
grad = numerical_gradient(f, x)
x -= lr * grad
return x, np.array(x_history)
def function(x):
return x[0]**2 + x[1]**2
init_x = np.array([3.0, -4.0])
lr = 0.1
step_num = 20
x0, x_history,= gradient_descent(function, init_x, lr=lr, step_num=step_num)
print(x0)
#定义坐标轴
fig = plt.figure()
ax = Axes3D(fig)
#ax = plt.axes(projection='3d')
#设置x、y、z轴
x= np.arange(-3, 3, 1) # 为了绘制函数的原图像
y= np.arange(-4, 4, 1)
x,y= np.meshgrid(x,y) #空间的点序列转换成网格点
#plt.rcParams['font.sans-serif']=['SimHei'] #用来正常显示中文标签
#设定显示范围
ax.set_xlabel('x0')
ax.set_ylabel('x1')
ax.set_zlabel('z')
plt.rcParams['font.sans-serif'] = ['Microsoft YaHei'] #汉显
#plt.title("显示图像")
#ax.set_title('函数图像')
ax.text2D(0.5, 0.95, "函数图像", transform=ax.transAxes)
ax.set_xlim(-3, 3)
ax.set_ylim(-4, 4)
ax.set_zlim(0, 25)
#生成三维数据
x1=x_history[:,0]
x2=x_history[:,1]
z0=f(x1,x2)
#绘制极值点散点图
ax.scatter(x1, x2, z0, cmap='spectral', c='r', label='airplane') # 绘制已经找到的极值点
#ax.legend() # 使坐标系为网格状
plt.show() # 显示
'''
def gradient_descent(f, init_x, lr=0.01, step_num=100):
x0 = init_x
x1 = init_x
y0 = init_y
y1 = init_y
x_history = []
y_history = []
z_history = []
for i in range(step_num):
x_history.append(x1.copy())
y_history.append(y1.copy())
z_history.append(f(x1,x2))
# 计算梯度
gradx = numerical_gradient(f, x0)
grady = numerical_gradient(f, y0)
x1 -= lr * gradx
y1 -= lr * grady
# 为新一轮迭代做准备
x0 = x1
y0 = y1
z0 = f(x0, y0)
return x, np.array(x_history),np.array(y_history)
'''