# -*- coding: utf-8 -*-

"""

Created on Fri Jun 28 18:28:28 2019

@author: Tomas

"""

import numpy as np

from scipy.optimize import linprog as lp

## load node to node matrix

NN = np.matrix([[0,0,0,0,0,0,1,0,1,0,0,0,0,0,0,0],

[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]])

# transform Node to Node matrix to Node Arc matrix, using the function made in basic utils EX0

NA, arcs = nodoNodoANodoArco(NN)

# Number of nodes of Plants, Stocks and Sale Points in order

pss = {"plants": 6, "stocks": 4, "sales": 6}

# Initialize Matrix of upper bounded constraints and equal constraints

Aub = np.zeros(NA.shape)

Aeq = np.zeros(NA.shape)

# Load values of Node Arc matriz into correspondix Matrix

Aub[:pss["plants"]+pss["stocks"], :] = NA[:pss["plants"]+pss["stocks"], :]

Aeq[pss["plants"]+pss["stocks"]:, :] = NA[pss["plants"]+pss["stocks"]:, :]

# load cost array of arcs dimensions

C = np.array([100,200,100,200,150,150,150,150,200,100,200,100,100,150,200,100,150,200,200,150,100,200,150,100])

# "linear equality constraint" of nodes dimensions

b = np.array([20,30,10,40,30,10,0,0,0,0,-30,-40,-10,-20,-20,-20])

# Initialize Split b to b for upper bound and b for equal constraints

bub = np.zeros(b.shape)

beq = np.zeros(b.shape)

bub[:pss["plants"]+pss["stocks"]] = b[:pss["plants"]+pss["stocks"]]

beq[pss["plants"]+pss["stocks"]:] = b[pss["plants"]+pss["stocks"]:]

## load bounds of arcs dimensions

#ubs = np.array([np.inf,np.inf,np.inf,np.inf,10,5,10,5,5,5,3,3,6,3,6,3,3,3,np.inf,np.inf,np.inf,np.inf,np.inf])

bounds = np.zeros((C.shape[0])).astype(tuple)

for i in range(0, C.shape[0]):

bounds[i] = (0,np.inf)

bounds = tuple(bounds)

# solve the linear prog for simplex and interior point

res = lp(C, A_ub = Aub, A_eq=Aeq, b_ub=bub ,b_eq=beq, bounds=bounds)

res_ip = lp(C, A_eq=NA, b_eq=b, bounds=bounds, method='interior-point')