#!/usr/bin/env python3
# -*- coding: utf-8 -*-


import numpy as np
from scipy.optimize import linprog, minimize_scalar
import matplotlib.pyplot as plt


demand = 10
path_number = 3


a = np.array([4, 25, 50])
b = np.array([1, 1, 1])


f = lambda x: np.sum(a * x + b * x ** 3 / 3)
k_max = 10000


plt.figure(figsize=(8, 5), dpi=500)


Iter = list()
Gap = list()
x = np.ones(path_number)
x = x / np.sum(x) * demand




for k in range(k_max):
    
    
    f_val = f(x)
    c = a + b * x ** 2
    
    y = np.zeros(path_number)
    y[np.argmin(c)] = demand
  
    d = y - x
    
    gap = -np.dot(c, y - x)
    print(f_val, gap)
    
    Iter.append(k)
    Gap.append(gap)
    
    alpha = 0.01
    x += alpha * d
plt.plot(Iter, Gap)
    
plt.legend(['decreasing step size', 'exact line search', 'inexact line search'])
plt.yscale('log')
