bottom_analysis.py

import numpy as np
import matplotlib.pyplot as plt

data_init = np.loadtxt('./bottom1.txt')
data_middle = np.loadtxt('./bottom2.txt')
data_end = np.loadtxt('./bottom3.txt')

def plot(data_init, data_middle, data_end):
    plt.figure(figsize=(10, 4))
    plt.style.use('seaborn')
    plt.subplot(1, 3, 1)
    plt.vlines(x=0, ymin=0, ymax=4.5, colors='black', ls=':',)
    plt.plot(data_init[:,-2], data_init[:, 2], marker='o')
    plt.plot(data_init[:,-2], data_init[:,4], marker='*')
    plt.plot(data_init[:,-2], data_init[:,6], marker='*')
    plt.ylim((0, 4.5))
    plt.legend(['Training domain: C', 'Training domain: P', 'Training domain: R'])
    plt.title(r'(a) Start with bottom eigenvectors')
    plt.subplot(1, 3, 2)
    plt.ylim((0, 4.5))
    plt.vlines(x=200, ymin=0, ymax=4.5, colors='black', ls=':',)

    plt.plot(data_middle[:,-2], data_middle[:, 2], marker='o')
    plt.plot(data_middle[:,-2], data_middle[:,4], marker='*')
    plt.plot(data_middle[:,-2], data_middle[:,6], marker='*')
    plt.legend(['Training domain: C', 'Training domain: P', 'Training domain: R'])
    plt.title(r'(b) Switch to bottom eigenvectors')
    plt.subplot(1, 3, 3)
    plt.ylim((0, 4.5))
    plt.vlines(x=500, ymin=0, ymax=4.5, colors='black', ls=':',)
    plt.plot(data_end[:,-2], data_end[:, 2], marker='o')
    plt.plot(data_end[:,-2], data_end[:,4], marker='*')
    plt.plot(data_end[:,-2], data_end[:,6], marker='*')
    plt.legend(['Training domain: C', 'Training domain: P', 'Training domain: R'])
    plt.title(r'(c) No bottom eigenvectors')
    plt.tight_layout()
    plt.savefig('./bottom_vector.png')
    plt.savefig('./bottom_vector.pdf')
plot(data_init, data_middle, data_end)