from itertools import *
from scipy import optimize as opt
from copy import deepcopy
+import sys as sy
+import cv as cv
+import cv2 as cv2
error = 0.1
epsilon = 1E-10
vrate = 0.8
p = 0.7
coteCarre = 50
-distanceEmmissionMax = 30
+distanceEmmissionMax = 20
nbiter = 1000
POS = 1
POS_NUL = 2
POSINF1 = 3
init = []
+fichier_init="config_initiale_default.txt"
#lg= [(0,1,23),(1,0,15),(1,2,45)]
sink = n-1
+def distance(d1,d2):
+ return mt.sqrt(sum([(d1[t]-d2[t])**2 for t in d1]))
+
def genereGraph():
test = False
G.add_edge(io,ie,weight=dist)
G.add_edge(ie,io,weight=dist)
test = not(any([ not(nx.has_path(G,o,sink)) for o in G.nodes() if sink in G.nodes() and o != sink]))
- return G
+ return (G,l)
+
+
+def afficheGraph(G,l,tx,ty,sink):
+ r = 20
+ img = cv.CreateImage ((tx, ty), 32, 3)
+ cv.Rectangle(img, (0,0),(tx,ty), cv.Scalar(255,255,255), thickness=-1)
+ def px((x,y)):
+ return(int(tx*x/coteCarre),ty-int(ty*y/coteCarre))
+ for i in set(range(len(l)))-set([sink]):
+ (x,y) = l[i]
+ pix,piy = px((x,y))
+ demx = distanceEmmissionMax*tx/coteCarre
+ cv.Circle(img, (pix,piy),demx, cv.Scalar(125,125,125))
+
+ for i in set(range(len(l)))-set([sink]):
+ (x,y) = l[i]
+ pix,piy = px((x,y))
+ cv.Circle(img, (pix,piy),r, cv.Scalar(125,125,125),thickness=-1)
+
+ #sink
+ (x,y) = l[sink]
+ pix,piy = px((x,y))
+
+ cv.Rectangle(img, (pix-r/2,piy-r/2),(pix+r/2,piy+r/2), cv.Scalar(125,125,125), thickness=-1)
+
+ for i in range(len(l)):
+ for j in range(len(l)):
+
+ if np.linalg.norm(np.array(l[i])-np.array(l[j])) < distanceEmmissionMax :
+ (xi,yi) = l[i]
+ pixi,piyi = px((xi,yi))
+ (xj,yj) = l[j]
+ pixj,piyj = px((xj,yj))
+ cv.Line(img, (pixi,piyi), (pixj,piyj), cv.Scalar(125,125,125))
+
+
+ """
+ for i in range(len(l)):
+ (x,y) = l[i]
+ pix,piy = px((x,y))
+ print i
+ textColor = (0, 0, 255) # red
+ font = cv2.FONT_HERSHEY_SIMPLEX
+ imgp =
+ cv2.putText(img, str(i), (pix-r/4,piy-r/2),font, 3.0, textColor)#,thickn """
+ cv.SaveImage("SensorNetwork.png",img)
G = nx.DiGraph()
G.add_weighted_edges_from(lg)
#nx.draw(G)
#pb.show()
+(G,l) = genereGraph()
+N = G.nodes()
+#V = list(set(sample(N,int(len(N)*vrate)))-set([sink]))
+V = list(set(N)-set([sink]))
+source = V
+print "source",source
+afficheGraph(G,l,500,500,sink)
+#nx.draw(G)
+#pb.show()
+
+
+
#print G.edges(data=True)
#TODO afficher le graphe et etre sur qu'il est connexe
-N = G.nodes()
-
-
-#V = list(set(sample(N,int(len(N)*vrate)))-set([sink]))
-V = list(set(N)-set([sink]))
-source = V
-print "source",source
L = range(len(G.edges()))
beta = 1.3E-8
gamma = 55.54
delta = 0.2
+zeta = 0.1
amplifieur = 1
sigma2 = 3500
Bi = 5
-def distance(d1,d2):
- return mt.sqrt(sum([(d1[t]-d2[t])**2 for t in d1]))
def AfficheVariation (up,vp,lap,wp,thetap,etap,qp,Psp,Rhp,xp,valeurFonctionDualep):
vp = {}
for h in V:
if not ASYNC or random() < taux_succes:
- s = Rh[h]- mt.log(float(sigma2)/D)/(gamma*mt.pow(Ps[h],float(1)/3))
+ s = Rh[h]- mt.log(float(sigma2)/D)/(gamma*mt.pow(Ps[h],float(2)/3))
if abs(s) > mxg :
print "ds calcul v",abs(s),idxexp
mxg = abs(s)
Psp={}
#print "maj des des Psh"
def f_Ps(psh,h):
- #print "ds f_ps",psh, v[h]* mt.log(float(sigma2)/D)/(gamma*((psh**2)**(float(1)/3))) +la[h]*psh
- return v[h]* mt.log(float(sigma2)/D)/(gamma*mt.pow(float(2)/3)) +la[h]*psh
+ #print "ds f_ps",psh, v[h]* mt.log(float(sigma2)/D)/(gamma*((psh**2)**(float(2)/3))) +la[h]*psh
+ return v[h]* mt.log(float(sigma2)/D)/(gamma*mt.pow(float(2)/3)) +la[h]*psh
for h in V:
- if not ASYNC or random() < taux_succes:
- lah = 0.05 if la[h] == 0 else la[h]
- rep = (float(2*v[h]*mt.log(float(sigma2)/D))/mt.pow(3*gamma*lah,float(3)/5))
- Psp[h] = epsilon if rep <= 0 else rep
- else :
+ if not ASYNC or random() < taux_succes:
+ """
+ lah = 0.05 if la[h] == 0 else la[h]
+ rep = mt.pow(float(2*v[h]*mt.log(float(sigma2)/D))/(3*gamma*lah),float(3)/5)
+ Psp[h] = epsilon if rep <= 0 else rep
+ """
+ t= float(-3*la[h]+mt.sqrt(9*(la[h]**2)+64*zeta*v[h]*mt.log(float(sigma2)/D)/gamma))/(16*zeta)
+ #print t
+ rep = mt.pow(t,float(3)/5)
+ Psp[h]=rep
+ else :
Psp[h] = Ps[h]
-def __evalue_maj_theta__():
+def initialisation_():
+ global u, v, la, w, theta , q, Ps, Rh, eta, x,init
+ fd = open(fichier_init,"r")
+ l= fd.readline()
+ init_p = eval(l)
+ print init_p
+ theta = omega
+ (q,Ps,Rh,eta,x,u,v,la,w) = tuple([deepcopy(x) for x in init_p])
+ init = [deepcopy(q),deepcopy(Ps),deepcopy(Rh),deepcopy(eta),
+ deepcopy(x),deepcopy(u),deepcopy(v),deepcopy(la),deepcopy(w)]
+
+
+
+def __evalue_maj_theta__(nbexp,out=False):
global u, v, la, w, theta , q, Ps, Rh, eta, x, valeurFonctionDuale
nbexp = 10
res = {}
itermax = 100000
def __maj_theta(k):
+ mem = []
om = omega/(mt.pow(k,0.75))
return om
for idxexp in range(nbexp):
mxg = 0
- initialisation()
+ if not(out):
+ initialisation()
+ else :
+ initialisation_()
+
k = 1
arret = False
sm = 0
if k%100 ==0 :
print "k:",k,"erreur sur q", errorq, "et q:",q
print "maxg=", mxg
+ mem = [deepcopy(q),deepcopy(Ps),deepcopy(Rh),deepcopy(eta),
+ deepcopy(x),deepcopy(u),deepcopy(v),deepcopy(la),deepcopy(w)]
+ if k%4500 == 0 :
+ print "#########\n",mem,"\#########\n"
+ if k%4600 == 0 :
+ print "#########\n",mem,"\#########\n"
+
+
+
if smax - sm > 500:
print "variation trop grande"
print "init"
print "nbre d'iteration trop grand"
print "init"
print init
- exit
+ sy.exit(1)
print "###############"
print k
