tests de l'algo de separation

separationLib.py

@@ -1,81 +0,0 @@
-import networkx as nx
-from scipy.optimize import linprog
-import numpy as np
-
-def initGraph(Adj):
-    G = nx.Graph()
-    
-    for l in range(len(Adj)):
-        for c in range(l):
-            G.add_edge(l, c, weight = Adj[l][c])
-    return G
-
-def initObj(G):
-    c = []
-    extToEdge = {}
-    for e in G.edges():
-        extToEdge[ (e[0], e[1]) ] = len(c)
-        extToEdge[ (e[1], e[0]) ] = len(c)
-        c.append(G[ e[0] ][ e[1] ]['weight'])
-
-    return c, extToEdge
-
-def initCstr(G, extToEdge):
-    nbN, nbE = len(G.nodes()), len(G.edges())
-    b_eq = 2*np.ones((nbN, 1))
-    A_eq = np.zeros((nbN, nbE))
-
-    for nd in G.nodes():
-        for nei in G[nd]:
-            A_eq[nd][ extToEdge[(nd, nei)] ] = 1
-
-    return A_eq, b_eq
-
-def subG(G, x, extToEdge):
-    nbN = len(G.nodes())
-    subG = nx.Graph()
-    for (d, a) in G.edges():
-        if(x[ extToEdge[(d, a)] ] != 0):
-            subG.add_edge(d, a, weight = G[d][a]['weight'] * x[ extToEdge[(d, a)] ])
-    
-    return subG
-
-def addSubTour(A_ub, b_ub, G, cut, extToEdge):
-    nbN = len(G.nodes())
-    
-    newCstr = [0]*len(G.edges())
-     
-    for d in cut[0]:
-        for a in cut[1]:
-            newCstr[ extToEdge[(d, a)] ] = -1
-
-    return A_ub.append(newCstr), b_ub.append(-2)
-
-def separation(Adj):
-    G = initGraph(Adj)
-    
-    c, extToEdge = initObj(G)
-    
-    A_eq, b_eq = initCstr(G, extToEdge)
-   
-    A_ub = []
-    b_ub = []
-    res = linprog(c, A_eq = A_eq, b_eq = b_eq, bounds=((0, 1)))
-    x = res.x
-    print(x) 
-    n = 100
-    while(n < 100):
-        sG = subG(G, x, extToEdge)
-        cutV, cut = nx.stoer_wagner(sG)
-        print(cutV, cut)
-        if(cutV >= 2):
-            break
-        A_ub, b_ub = addSubTour(A_ub, b_ub, G, cut, extToEdge)
-        
-        res = linprog(c, A_eq = A_eq, b_eq = b_eq, A_ub = A_ub, b_ub = b_ub, bounds=((0, 1)))
-        x = res.x
-        
-        n += 1
-
-
-    return x, res.fun