第十二周

1.

x = np.linspace(0,2,50)
y = np.sin(x-2) * np.sin(x-2) * np.exp(-x**2)
plt.figure(1)
plt.xlabel("x")
plt.ylabel("y")
plt.ylim(-0.1, 1)
plt.title(r"$f(x) = sin^2(x-2)e^{-x^2}$")
plt.annotate('local max', xy=(0.22, 0.9), xytext=(0.13, 0.5),
            arrowprops=dict(facecolor='black', shrink=0.05))
plt.plot(x,y)
plt.show()

2.

X = np.random.normal(loc=5,scale=5,size=(20,10))
b = np.random.normal(loc=0,scale=5,size=10)
z = np.random.normal(size=20)
y = np.dot(X,b) + z

b_e = np.ones(10)

def error(p,xxx,yyy):
    return np.dot(xxx,p)-yyy

b_e,cost=leastsq(error,b_e,args=(X,y)) 

plt.figure(2)

plt.xlabel("index")
plt.ylabel("value")
plt.plot(b, 'rx', label="True coefficients")
plt.plot(b_e, 'b', label="Estimated coefficients")
plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
plt.axis([0,9,-10,10])
plt.hlines(0,0,9 ,colors = "c", linestyles = "dashed")
plt.show()

3.

data = np.random.normal(scale=100, size=10000)
kernel = stats.gaussian_kde(data)
full_data = np.linspace(-500,500,10000)
est_data = kernel.evaluate(full_data)
plt.figure(3)
plt.hist(data, 30, density=True)
plt.plot(full_data, est_data)
plt.show()