def build_constraints(self):
""" @param n, r, p: number of exams, rooms and periods
Build the variables of the problem from the data
"""
n, r, p = self.dimensions['n'], self.dimensions['r'], self.dimensions['p']
for i in range(n):
# Add constraint: Each exam is planned in exactly one period
constraint = (gb.quicksum(self.vars['y'][i, l] for l in range(p)) == 1)
self.problem.addConstr(constraint, "c0")
# Add constraint: Each exam has enough seats
constraint = (gb.quicksum(self.vars['x'][i, k] * self.constants['c'][k] for k in range(r)) >= self.constants['s'][i])
self.problem.addConstr(constraint, "c1")
# Add constraint: Each exam is planned in exactly one period
constraint = (gb.quicksum(self.vars['y'][i, l] for l in range(p)) == 1)
self.problem.addConstr(constraint, "c0")
# Add constraint: Each room has at most one exam per period
for k in range(r):
for l in range(p):
constraint = (
gb.quicksum(self.vars['x'][i, k] * self.vars['y'][i, l] for i in range(n)) <=
self.constants['T'][k][l]
)
self.problem.addQConstr(constraint, "c2")
# Add constraint: There are no conflicts
for l in range(p):
constraint = (gb.quicksum([self.vars['y'][i, l] * self.vars['y'][j, l] * self.constants['Q'][i][j] for i, j in itertools.combinations(range(n), 2) if self.constants['Q'][i][j] == 1]) == 0)
self.problem.addQConstr(constraint, "c3")
return True
def mtztw(n,c,e,l):#############################################################
"""mtzts: model for the traveling salesman problem with time windows
Parameters:
- n: number of nodes
- c[i,j]: cost for traversing arc (i,j)
- e[i]: earliest date for visiting node i
- l[i]: latest date for visiting node i
Returns a model, ready to be solved.
"""
model = gurobipy.Model("tsptw - mtz")
x,u = {},{}
for i in range(1,n+1):
u[i] = model.addVar(lb=e[i], ub=l[i], vtype="C", name="u(%s)"%i)
for j in range(1,n+1):
if i != j:
x[i,j] = model.addVar(vtype="B", name="x(%s,%s)"%(i,j))
model.update()
for i in range(1,n+1):
model.addConstr(gurobipy.quicksum(x[i,j] for j in range(1,n+1) if j != i) == 1, "Out(%s)"%i)
model.addConstr(gurobipy.quicksum(x[j,i] for j in range(1,n+1) if j != i) == 1, "In(%s)"%i)
for i in range(1,n+1):
for j in range(2,n+1):
if i != j:
M = max(l[i] + c[i,j] - e[j], 0)
model.addConstr(u[i] - u[j] + M*x[i,j] <= M-c[i,j], "MTZ(%s,%s)"%(i,j))
model.setObjective(gurobipy.quicksum(c[i,j]*x[i,j] for (i,j) in x),gurobipy.GRB.MINIMIZE)
model.update()
model.__data = x,u
return model
def __loadModel(self):
# Create optimization model
self.__model = Model('PathBackup')
# Auxiliary variables for SOCP reformulation
U = {}
R = {}
# Create variables
for i,j in self.__links:
self.__BackupCapacity[i,j] = self.__model.addVar(lb=0, obj=1, name='Backup_Capacity[%s,%s]' % (i, j))
self.__model.update()
for p in self.__paths:
#LP Relaxation
#self.__bPath[self.__paths.index(p)] = self.__model.addVar(lb=0,obj=1,name='Backup_Path[%s]' % (self.__paths.index(p)))
self.__bPath[self.__paths.index(p)] = self.__model.addVar(vtype=GRB.BINARY,obj=1,name='Backup_Path[%s]' % (self.__paths.index(p)))
self.__model.update()
for i,j in self.__links:
U[i,j] = self.__model.addVar(obj=1,name='U[%s,%s]' % (i, j))
self.__model.update()
for i,j in self.__links:
for s,d in self.__links:
R[i,j,s,d] = self.__model.addVar(obj=1,name='R[%s,%s,%s,%s]' % (i,j,s,d))
self.__model.update()
self.__model.modelSense = GRB.MINIMIZE
self.__model.setObjective(quicksum(self.__BackupCapacity[i,j] for i,j in self.__links))
self.__model.update()
#------------------------------------------------------------------------#
# Constraints definition #
# #
# #
#------------------------------------------------------------------------#
# Link capacity constraints
for i,j in self.__links:
self.__model.addConstr(self.__BackupCapacity[i,j] >= quicksum(self.__mean[s,d]*quicksum(self.__bPath[self.__paths.index(p)] for p in self.__Pij[i,j,s,d]) for s,d in self.__links) + U[i,j]*self.__invstd,'[CONST]Link_Cap[%s][%s]' % (i, j))
self.__model.update()
# SCOP Reformulation Constraints
for i,j in self.__links:
self.__model.addConstr(quicksum(R[i,j,s,d]*R[i,j,s,d] for s,d in self.__links) <= U[i,j]*U[i,j],'[CONST]SCOP1[%s][%s]' % (i, j))
self.__model.update()
# SCOP Reformulation Constraints
for i,j in self.__links:
for s,d in self.__links:
self.__model.addConstr(self.__std[s,d]*quicksum(self.__bPath[self.__paths.index(p)] for p in self.__Pij[i,j,s,d]) == R[i,j,s,d],'[CONST]SCOP2[%s][%s][%s][%s]' % (i,j,s,d))
self.__model.update()
# Unique path
for s,d in self.__links:
self.__model.addConstr(quicksum(self.__bPath[self.__paths.index(p)] for p in self.__Psd[s,d]) == 1,'UniquePath[%s,%s]' % (s, d))
self.__model.update()
def buildIP(self, setNumTrucks = None):
model = gurobipy.Model('VRPModel')
model.setParam('OutputFlag',False)
# variables
x = {}
for i in range(len(self.confs)):
x[i] = model.addVar(vtype=gurobipy.GRB.BINARY, name=str(i))
model.update()
# constraints
for targetId in self.indices:
model.addConstr(gurobipy.quicksum(x[i] for i in self.indices[targetId]) >= 1,'target_%s' % (i))
if setNumTrucks:
model.addConstr(gurobipy.quicksum(x[i] for i in range(len(self.confs))) <= setNumTrucks,'at_most_%i_confs' % (setNumTrucks))
# objective
if setNumTrucks:
model.setObjective(gurobipy.quicksum( x[i]*(self.confs[i].val) for i in range(len(self.confs)) ) )
else:
model.setObjective(gurobipy.quicksum(x[i]*(self.confs[i].val + self.M) for i in range(len(self.confs))))
model.setAttr("modelSense", gurobipy.GRB.MINIMIZE)
model.setParam('TimeLimit', self.timeout)
model.update()
self.x = x
self.model = model
def addGreaterEqual(self, model, lhs_terms, xs, rhs, sname, aux_vars=None):
"""Adds the unwrapped version of
lhs_terms + xs *u >= rhs """
if aux_vars is None:
t1 = model.addVar()
t2 = model.addVar()
t3 = [model.addVar(lb=-grb.GRB.INFINITY) for ix in xrange(HORIZON_LENGTH)]
t4 = [model.addVar(lb=-grb.GRB.INFINITY) for ix in xrange(HORIZON_LENGTH)]
model.update()
else:
(t1, t2, t3, t4) = aux_vars
objs = [t1, t2]
objs += t3
objs += t4
# assume that x represents an expression, add extra variables to unpack it
# VG test type of x Can test if you want to
for x, t in zip(xs, t4):
objs.append(model.addConstr(x == t))
for ix, (row, t) in enumerate(zip(self.chol, t3)):
objs.append(model.addConstr(grb.quicksum(r * f for r, f in zip(row[: (ix + 1)], xs)) == t))
objs.append(model.addQConstr(grb.quicksum(t * t for t in t4) <= t1 * t1, name=sname + "Q1"))
objs.append(model.addQConstr(grb.quicksum(t * t for t in t3) <= t2 * t2, name=sname + "Q2"))
objs.append(
model.addConstr(
lhs_terms + grb.quicksum(mu * x for mu, x in zip(self.mean, xs)) - self.gamma1 * t1 - self.kappa * t2
>= rhs,
name=sname,
)
)
return objs
def build_constraints(self):
""" @param n, r, p: number of exams, rooms and periods
Build the variables of the problem from the data
"""
n, r, p = self.dimensions['n'], self.dimensions['r'], self.dimensions['p']
for i in range(n):
# Each exam i is scheduled on exactly one period
constraint = (gb.quicksum([self.vars['y'][i, l] for l in range(p)]) == 1)
self.problem.addConstr(constraint, "c0")
# enough seats for each student for exam i
constraint = (
gb.quicksum([self.vars['x'][i, k] * self.constants['c'][k] for k in range(r)]) >=
self.constants['s'][i]
)
self.problem.addConstr(constraint, "c1")
# No conflicts
for l in range(p):
constrant = (
gb.quicksum([self.vars['y'][j, l] * self.constants['Q'][i][j] for j in range(n) if j != i]) <=
(1 - self.vars['y'][i, l]) * n
)
self.problem.addConstr(constrant, "c2")
for k in range(r):
for l in range(p):
if self.constants['T'][k][l] == 0:
# We use room k if and only if the room is open
constraint = (self.vars['x'][i, k] + self.vars['y'][i, l] <= 1)
self.problem.addConstr(constraint, "c3")
return True
def ldp(ys):
"""
Use gurobi to compute the point in the convex hull of the points in [ys] which is closest to the origin.
@input ys: an array of size [dimension x number of points]
@returns ystar: a vector of size [dimension]
"""
grb_model.reset()
for v in grb_model.getVars():
grb_model.remove(v)
for c in grb_model.getConstrs():
grb_model.remove(c)
dim = ys.shape[0]
pts_per_obstacle = ys.shape[1]
ws = [grb_model.addVar(lb=0, ub=1) for j in range(pts_per_obstacle)]
vs = [grb_model.addVar(lb=-gurobipy.GRB.INFINITY, ub=gurobipy.GRB.INFINITY) for j in range(dim)]
grb_model.update()
for j in range(dim):
grb_model.addConstr(gurobipy.LinExpr(list(ys[j,:]), ws) == vs[j])
grb_model.addConstr(gurobipy.quicksum(ws) == 1)
grb_model.setObjective(gurobipy.quicksum([vs[j] * vs[j] for j in range(dim)]))
grb_model.optimize()
return np.array([v.x for v in vs])
def rampingCallback(self, where):
"""Callback to be used when using sparse ramping constraints"""
model = self.model
if where == grb.GRB.callback.MIPSOL:
for name, gen in self.trueGens.items():
if gen.fuel_type in ("Steam", "CT"):
continue
for hr in xrange(1, HORIZON_LENGTH):
f, g = self.prodVars[name, hr]
fm, gm = self.prodVars[name, hr - 1]
fval = numpy.array( model.cbGetSolution(f) )
fmval = numpy.array( model.cbGetSolution(fm) )
gval, gmval = model.cbGetSolution([g, gm])
#check to see if its violated
eco_max_m = gen.eco_max[hr - 1]
if model.cbGetSolution(self.stopVars[name, hr]) <= .9:
resid_star, value = self.affCG.suppFcn(fmval - fval)
value += gmval - gval
if value > gen.ramp_rate + 1e-5: #VG return to this
print "Ramping CallBack Used!"
self.affCG.model.cbLazy(grb.quicksum((fmi - fi) * ri for (fmi, fi, ri) in
zip(fm, f, resid_star) ) + gm - g <=
gen.ramp_rate + eco_max_m * self.stopVars[name, hr])
eco_max = gen.eco_max[hr]
if model.cbGetSolution(self.startVars[name, hr]) <= .9:
resid_star, value = self.affCG.suppFcn(fval - fmval)
value += gval - gmval
if value > gen.ramp_rate + 1e-5: #VG return to this
print "RampingCallBack Used!"
self.affCG.model.cbLazy(grb.quicksum((fi - fmi) * ri for (fi, fmi, ri) in
zip(f, fm, resid_star) ) + g - gm <=
gen.ramp_rate + eco_max * self.startVars[name, hr])
def tsp(D):
n = D.shape[0]
m = grb.Model("tsp")
x = {}
for i in xrange(n):
for j in xrange(i):
x[i,j] = m.addVar(vtype='B', lb=0, ub=1, obj=D[i,j], name='x_%d%d'%(i,j))
m.update()
for i in xrange(n):
m.addConstr(grb.quicksum(x[arc] for arc in pairs(i,n)) == 2, 'triv_%d'%i)
mincutval, it = 0, 0
while mincutval < 2:
m.optimize()
if m.status != grb.GRB.status.OPTIMAL:
print 'bad status:', m.status
return None
candidate = [arc for arc,var in x.iteritems() if var.x > 0]
g = UndirectedGraph()
g.add_vertices(xrange(n))
for edge in candidate:
g.add_edge(*edge)
S, T, mincutval = minimum_cut(g, 0)
m.addConstr(grb.quicksum(x[arc] for arc in cross(S,T)) >= 2, 'cut_%d'%it)
it += 1
print it
return candidate, m.objVal
def addGraphConstrs(self):
"""
add basic graph constraints
"""
num_nodes = self.num_nodes
# an edge/node is either selected or not
for i in xrange(1, num_nodes+1):
self.model.addConstr(gp.quicksum(self.nodeVars[i,t] for t in [0,1]) == 1.0)
for i in xrange(1, num_nodes+1):
for j in xrange(1, num_nodes+1):
self.model.addConstr(gp.quicksum(self.edgeVars[i,j,t] for t in [0,1]) == 1.0)
for i in xrange(1, num_nodes+1):
self.model.addConstr(gp.quicksum(self.dummyEdgeVars[i,t] for t in [0,1]) == 1.0)
# one or more edges from ROOT to any node
self.model.addConstr(gp.quicksum(self.dummyEdgeVars[i,1] for i in xrange(1, num_nodes+1)) >= 1.0)
# at most one edge between any pair of nodes (no self-loop)
for i in xrange(1, num_nodes+1):
for j in xrange(1, num_nodes+1):
self.model.addConstr(gp.LinExpr(self.edgeVars[i,j,1] + self.edgeVars[j,i,1]) <= 1.0)
# an edge may only be included if both its endpoints are included
for i in xrange(1, num_nodes+1):
self.model.addConstr(gp.LinExpr(self.nodeVars[i,1] - self.dummyEdgeVars[i,1]) >= 0.0)
for i in xrange(1, num_nodes+1):
for j in xrange(1, num_nodes+1):
self.model.addConstr(gp.LinExpr(self.nodeVars[i,1] - self.edgeVars[i,j,1]) >= 0.0)
self.model.addConstr(gp.LinExpr(self.nodeVars[j,1] - self.edgeVars[i,j,1]) >= 0.0)
return
def solveWeightReductionLP(self):
# define objective
self.WeightReductionLPModel.setObjective( gp.quicksum(self.LPVars[t]*self.g.allWeights[t] for t in self.g.allWeights) +
gp.quicksum(self.newLPVars[((u,v),)]*self.newWeightsVars[u,v] for (u,v) in self.edges) -
gp.quicksum(self.diffEdges[u,v] for (u,v) in self.edges) -
gp.quicksum(self.newWeightsVars[u,v]*self.newWeightsVars[u,v] for (u,v) in self.edges) -
gp.quicksum(self.newLPVars[((u,v),)]*self.newLPVars[((u,v),)] for (u,v) in self.edges) ,gp.GRB.MAXIMIZE)
self.WeightReductionLPModel.update()
# write the model
modelFile = 'model2.lp'
if (self.lpFile):
modelFile = self.lpFile;
self.WeightReductionLPModel.write(modelFile)
# self.WeightReductionLPModel.computeIIS();
# self.WeightReductionLPModel.write('model2.ilp')
# return
# solve the model
self.WeightReductionLPModel.optimize()
if self.WeightReductionLPModel.status == gp.GRB.status.OPTIMAL:
for (u,v) in self.edges:
if self.LPVars[((u,v),)].x > 0:
print('(%s,%s)' % (u,v))
print('=====')
newWeights = {}
for (u,v) in self.edges:
text = '(%s,%s)' % (u,v)
text = text + ', w = %s' % (self.newWeightsVars[u,v].x)
newWeights[(u,v)] = self.newWeightsVars[u,v].x
if self.newLPVars[((u,v),)].x > 0:
text = text + ' *'
#print(text)
self.newWeights = newWeights
def sampleConstraint(self, ybar, gbar, b, tag, numPts, isLessEqual, ixD=None):
"""Samples numPts versions of the constraint
ybar ^T M_k d + gbar <= d_{ix} + b """
#assume the sample mean is exact for the sampling
#simply use a gaussian approximation around the sample mean with shrunken covar
zetas = numpy.random.randn( self.d, numPts )
zetas /= map(numpy.linalg.norm, zetas.T)
zetas *= self.kappa
us = self.mu + self.invChol * zetas
Mus = numpy.dot(self.Mk.T, self.u)
#VG Check the dimensions of Mus and us and stuff
pdb.set_trace()
if ixD is None:
for ix in xrange(numPts):
self.model.addConstr(
grb.quicksum( ybar_i * mu_i for ybar_i, mu_i in zip(ybar, Mus[:, ix] ) )
+ gbar <= b, name = tag + "Sample%d" % ix )
return
else:
for ix in xrange(numPts):
self.model.addConstr(
grb.quicksum( ybar_i * mu_i for ybar_i, mu_i in zip(ybar, Mus[:, ix] ) )
+ gbar <= b + us[ixD, ix], name = tag +"Sample%d" % ix )
return
def buildLP(self):
model = gurobipy.Model('trimModel')
model.setParam('OutputFlag',False)
x = {}
for i in range(self.nConfs):
x[i] = model.addVar(vtype=gurobipy.GRB.CONTINUOUS, name="x_%s" % (i), ub=1.0)
s = {}
for j in range(self.nTargets):
s[j] = model.addVar(vtype=gurobipy.GRB.CONTINUOUS, name="s_%s" % (i))
model.update()
model.addConstr(gurobipy.quicksum(x[i] for i in range(self.nConfs) ) >= self.trimParam, '%s_confs' % (self.trimParam))
for chosenConfIndex in self.chosenConfsIndices:
model.addConstr(x[chosenConfIndex] >= 1.0, '%s_conf' % (chosenConfIndex))
for j in range(self.nTargets):
relevantConfs = [i for i in range(self.nConfs) if j in self.confsTargets[i]]
model.addConstr(gurobipy.quicksum(x[i] for i in relevantConfs) <= s[j], 'col_%s' % (j))
model.update()
model.setObjective(gurobipy.quicksum(s[j] * s[j] for j in range(self.nTargets)))
model.setAttr("modelSense", gurobipy.GRB.MINIMIZE)
model.update()
model.write('tmp.lp')
self.x = x
self.model = model
def logProbMaximizeIntraMinimizeInter(self, alpha=0.5):
"""
Maximize ∑ α.xij.log(pij) + (1-α).(1-xij).log(1-pij)
j>i
Parameters
----------
alpha : float, optional
0 < α < 1 in above equation
"""
intra = grb.quicksum([np.log(self.graph[n][m][PROBABILITY])*self.x[n,m]
for (n,m) in self.x
if self.graph.has_edge(n,m)
and self.graph[n][m][PROBABILITY] > 0])
inter = grb.quicksum([np.log(1-self.graph[n][m][PROBABILITY])*(1-self.x[n,m])
for (n,m) in self.x
if self.graph.has_edge(n,m)
and self.graph[n][m][PROBABILITY] < 1])
self.model.setObjective((1-alpha)*intra+alpha*inter, grb.GRB.MAXIMIZE)
self.model.optimize()
return self.getAnnotations()
def _build_constraints(self):
m = self.model
zones = self.data.zones
lines = self.data.lines
linelimit = self.variables.linelimit
gens_for_zones = self.data.gens_for_country
windreserve = self.variables.windcap
gcap = self.variables.gcap
export = self.variables.export
demand = self.data.zonaldemand
capshed = self.variables.capshed
# Power Balance constraint in each zone at each time
self.constraints.powerbalance = {}
for z in zones:
self.constraints.powerbalance[z] = m.addConstr(
gb.quicksum(gcap[g] for g in gens_for_zones[z])
+ windreserve[z]
,gb.GRB.EQUAL,
(demand[z] - capshed[z]) + export[z])
# Export constraint from each zone
self.constraints.exporting = {}
for z in zones:
self.constraints.exporting[z] = m.addConstr(
export[z], gb.GRB.EQUAL,
gb.quicksum(linelimit[l] for l in lines if l[0] == z) - gb.quicksum(linelimit[l] for l in lines if l[1] == z))
def build_model(plants, warehouses, capacity, demand, fixed_costs, trans_costs):
# decision variables
m = Model("facility")
is_open = []
for p in plants:
is_open.append(m.addVar(vtype=GRB.BINARY,
name="is_open[{}]".format(p)))
trans_qty = []
for w in warehouses:
trans_qty.append([])
for p in plants:
trans_qty[w].append(m.addVar(vtype=GRB.CONTINUOUS,
name="trans_qty[{}.{}]".format(p, w),
lb=0.0))
m.update()
# objective function
m.setObjective(quicksum(fixed_costs[p] * is_open[p]
for p in plants) +
quicksum(trans_costs[w][p] * trans_qty[w][p]
for w in warehouses
for p in plants),
GRB.MINIMIZE)
# constraints
for p in plants:
m.addConstr(quicksum(trans_qty[w][p] for w in warehouses) <= capacity[p] * is_open[p],
"Capacity({})".format(p))
for w in warehouses:
m.addConstr(quicksum(trans_qty[w][p] for p in plants) == demand[w],
"Demand({})".format(w))
m.update()
return m
def addPiecewiseCostsAffine(model, gen_dict, prod_vars, USet):
"""Creates variables and the defining constraints for the piecewise cost functions
for the variable cost """
ys_all, cost_vars = {}, {}
true_gens = filter(lambda g: g.res_type == "GEN" and g.fuel_type <> "FixedImport", gen_dict.values())
for gen in true_gens:
name = gen.name
# There is a nicety that all the true generators pertain H1*H36
assert gen.offerBlocks.keys() == ["H1*H36"]
offerBlocks = gen.offerBlocks["H1*H36"]
# For numerical stability, makes sense to clip these at eco_max.
eco_max = max(gen.eco_max.values())
for numKnots, b in enumerate(offerBlocks):
if b.size >= eco_max:
break
sizes = [b.size for b in offerBlocks[:numKnots]] + [eco_max]
prices = [b.price for b in offerBlocks[:numKnots]] + [offerBlocks[numKnots].price]
size_diffs = [s - sm for s, sm in zip(sizes, [0] + sizes)]
numKnots += 1
# simpler formulation relies on the fact that size_diffs are all > 0
for iHr in xrange(HORIZON_LENGTH):
ys_all[name, iHr] = [model.addVar(ub=1.0) for ix in xrange(numKnots)]
cost_vars[name, iHr] = model.addVar(obj=1.0)
# single call to save time at memory
model.update()
# requisition the variables for the affineness
aux_vars = USet.addVecVars(model, len(true_gens) * HORIZON_LENGTH)
for ixGen, gen in enumerate(true_gens):
name = gen.name
offerBlocks = gen.offerBlocks["H1*H36"]
eco_max = max(gen.eco_max.values())
for numKnots, b in enumerate(offerBlocks):
if b.size >= eco_max:
break
sizes = [b.size for b in offerBlocks[:numKnots]] + [eco_max]
prices = [b.price for b in offerBlocks[:numKnots]] + [offerBlocks[numKnots].price]
size_diffs = [s - sm for s, sm in zip(sizes, [0] + sizes)]
numKnots += 1
for iHr in xrange(HORIZON_LENGTH):
model.addConstr(
cost_vars[name, iHr]
== grb.quicksum(y * p * s for y, p, s in zip(ys_all[name, iHr], prices, size_diffs))
)
f, g = prod_vars[name, iHr]
t1, t2 = USet.createNormVars(model, f, aux_vars[HORIZON_LENGTH * ixGen + iHr])
USet.addLessEqualFast(
model,
g,
f,
grb.quicksum(y * s for y, s in zip(ys_all[name, iHr], size_diffs)),
"PieceWiseCost" + name + str(iHr),
t1,
t2,
)
return cost_vars
def ilp(costMatrix):
#Invalid_Connections : -1
if costMatrix.shape==(0,0):
return []
dist_mat=numpy.copy(costMatrix)
dist_mat[costMatrix==-1]=10e10
size_x = dist_mat.shape[0]
size_y = dist_mat.shape[1]
size_min = int(numpy.amin([size_x,size_y]))
from gurobipy import Model, quicksum, GRB
m=Model("mip1")
COS,VAR={},{}
for i in range(size_x):
x_cos, x_var = [],[]
for j in range(size_y):
COS[i,j]=dist_mat[i,j]
VAR[i,j]=m.addVar(vtype='B',name="["+str(i)+","+str(j)+"]")
m.update()
# Set objective
m.setObjective( quicksum(\
COS[x,y]*VAR[x,y]
for x in range(size_x) \
for y in range(size_y) \
),GRB.MINIMIZE)
# Constrains HORIZONTAL
for i in range(size_x):
m.addConstr( quicksum\
(VAR[i,y] for y in range(size_y)) <= 1)
# Constrains VERTICAL
for i in range(size_y):
m.addConstr( quicksum\
(VAR[x,i] for x in range(size_x)) <= 1)
m.addConstr(quicksum(\
VAR[x,y] for x in range(size_x) for y in range(size_y)) == int(size_min))
m.setParam("OutputFlag",False)
m.optimize()
res=numpy.zeros(dist_mat.shape,dtype=bool)
for i in range(size_x):
for j in range(size_y):
res[i,j]=VAR[i,j].x
binMatrix = numpy.zeros( costMatrix.shape,dtype=bool )
binMatrix[res==1]=1
binMatrix[costMatrix==-1]=0
return binMatrix
def build_constraints(self):
for l in self.dimensions['p']:
# Avoid conflicts between groups
self.problem.addConstr(gb.quicksum([self.vars['x'][i, l] for i in range(self.dimensions['c']) if l in self.constants['available'][i]]) <= 1)
for i in range(self.dimensions['c']):
# each group has to have at least one time slot
self.problem.addConstr(gb.quicksum([self.vars['x'][i, l] for l in range(self.dimensions['p'])]) >= 1)
return True
def convexify(self, penalty_coeff):
for constraint in self.constraints:
constraint.clean()
self.convexified_constr = []
for c in self.constraints:
self.convexified_constr.extend(constraint.convexify(self.model, penalty_coeff))
penalty_obj = grb.quicksum(self.convexified_constr)
# self.obj_sqp = self.obj + penalty_coeff * self.nonlinear_cnts.convexify(self.model, penalty_coeff)
self.obj_sqp = grb.quicksum(self.obj_quad) + penalty_obj
def build_objectif(self):
""" @param n, r, p: number of exams, rooms and periods
Build the constants of the problem from the data
"""
n, r, p = self.dimensions['n'], self.dimensions['r'], self.dimensions['p']
obj1 = gb.quicksum([self.vars['x'][i, k] * self.constants['s'][i] for i, k in itertools.product(range(n), range(r))])
obj2 = [[gb.quicksum([self.vars['y'][i, l] * self.constants['h'][l] - self.vars['y'][j, l] * self.constants['h'][l] for l in range(p)]) for i in range(n)] for j in range(n)]
obj = obj1 - gb.quicksum([gb.quicksum([self.constants['Q'][i][j] * obj2[i][j] * obj2[i][j] for i, j in itertools.combinations(range(n), 2) if self.constants['Q'][i][j] == 1])])
self.problem.setObjective(obj, gb.GRB.MINIMIZE)
return True
def build_objectif(self):
""" @param n, r, p: number of exams, rooms and periods
Build the constants of the problem from the data
"""
n, r = self.dimensions['n'], self.dimensions['r']
crit = (
self.c * gb.quicksum([self.vars['x'][i, k] * self.constants['s'][i] for i in range(n) for k in range(r)]) -
gb.quicksum([self.vars['z'][i, j] for i in range(n) for j in range(i + 1, n)])
)
self.problem.setObjective(crit, gb.GRB.MINIMIZE)
return True
def _build_constraints(self):
m = self.model
times = self.data.times
zones = self.data.zones
lines = self.data.lines
linelimit = self.variables.linelimit
Load = self.data.zonalconsumption
loadshed = self.variables.loadshed
gprod = self.variables.gprod
gens_for_country = self.data.gens_for_country
export = self.variables.export
windprod = self.variables.windprod
solarprod = self.variables.solarprod
generators = self.data.generators
gprod = self.variables.gprod
hydrocoeff = self.data.hydrocoeff
# Power Balance constraint in each zone at each time
self.constraints.powerbalance = {}
for z in zones:
for t in times:
self.constraints.powerbalance[z, t] = m.addConstr(
gb.quicksum(gprod[g, t] for g in gens_for_country[z])
+ windprod[z, t] + solarprod[z, t]
,gb.GRB.EQUAL,
(Load[z, t] - loadshed[z, t]) + export[z, t])
# Export constraint at each node at each time
self.constraints.exporting = {}
for t in times:
for z in zones:
self.constraints.exporting[z, t] = m.addConstr(
export[z, t], gb.GRB.EQUAL,
gb.quicksum(linelimit[l, t] for l in lines if l[0] == z) - gb.quicksum(linelimit[l, t] for l in lines if l[1] == z))
# Hydro constraint
self.constraints.hydro = {}
for g in generators.index:
if generators.primaryfuel[g] == "Hydro":
if not (generators.country[g] == "MKD" or generators.country[g] == "MNE" or generators.country[g] == "HRV"):
self.constraints.hydro[g] = m.addConstr(
gb.quicksum(gprod[g, t] for t in times), gb.GRB.LESS_EQUAL,
hydrocoeff * generators.capacity[g] * len(times))
def _build_constraints(self):
m = self.model
zones = self.data.zones
gens_for_zones = self.data.gens_for_country
gcap = self.variables.gcap
demand = self.variables.demand
generators = self.data.generators
dispatch = self.variables.dispatch
export = self.variables.export
lines = self.data.lines
linelimit = self.variables.linelimit
# Power Balance constraint in each zone
self.constraints.powerbalance = {}
for z in zones:
self.constraints.powerbalance[z] = m.addConstr(
gb.quicksum(gcap[g] for g in gens_for_zones[z])
,gb.GRB.EQUAL,
demand[z] + export[z])
# Dispatched status for each generator
self.constraints.dispatch = {}
for g in generators.index:
self.constraints.dispatch[g] = m.addConstr(
generators.capacity[g] * dispatch[g],
gb.GRB.EQUAL, gcap[g])
# Export constraint from each zone
self.constraints.exporting = {}
for z in zones:
self.constraints.exporting[z] = m.addConstr(
export[z], gb.GRB.EQUAL,
gb.quicksum(linelimit[l] for l in lines if l[0] == z) - gb.quicksum(linelimit[l] for l in lines if l[1] == z))
# Hydro constraint (cannot be a PLR)
self.constraints.hydro = {}
for g in generators.index:
if generators['primaryfuel'][g] == 'Hydro':
self.constraints.hydro[g] = m.addConstr(
generators.capacity[g] * dispatch[g],
gb.GRB.EQUAL, 0)
def _build_objective(self):
m = self.model
gens = self.MP.data.generators
geninfo = self.MP.data.geninfo
VOLL = self.MP.data.VOLL
dumploadprice = self.MP.data.dumploadprice
m.setObjective(
gb.quicksum((geninfo.price[g] + geninfo.uppremium[g])*self.variables.gprod_rt_up[g] for g in gens) +
gb.quicksum((- geninfo.price[g] + geninfo.downpremium[g])*self.variables.gprod_rt_down[g] for g in gens) -
VOLL*(self.variables.loadserved-self.variables.loadserved_DA) +
dumploadprice * self.variables.dumpload
)
def add_voice_position_constraints(self):
"""Ensure S > A > T > B for all t."""
for t in self.times:
for v in range(self.max_voice - 1):
constraint = \
quicksum([p * self[p, v, t] for p in self.pitches]) <= \
quicksum([p * self[p, v + 1, t] for p in self.pitches]) - 1
self.model.addConstr(constraint,
"satb_u" + str(v) + "_v" + str(v + 1) + "_t" + str(t))
def find_feasible_start(n_colors, h, statespace, conflicts, verbose=False):
model = Model("TimeFeasibility")
p = len(h)
y = {}
# y[i,k] = if color i gets slot l
for i in range(n_colors):
for l in range(p):
y[i,l] = model.addVar(vtype=GRB.BINARY, name="y_%s_%s" % (i,l))
model.update()
# Building constraints...
# c1: all get one
for i in range(n_colors):
model.addConstr( quicksum([ y[i, l] for l in range(p) ]) == 1, "c1")
# c2: each slot needs to be used tops once
for l in range(p):
model.addConstr( quicksum([ y[i, l] for i in range(n_colors) ]) <= 1, "c2")
### c3: statespace constraints
for i in range(n_colors):
#print l, h[l], i, [s for s in statespace]
model.addConstr( quicksum([ y[i, l] for l in range(p) if h[l] not in statespace[i] ]) == 0, "c3")
# objective: minimize conflicts
#obj = quicksum([ y[i,l] * y[j,l] for l in range(p) for i in range(n_colors) for j in range(i+1, n_colors) ])
obj = quicksum([ sum(y[i,l] for i in range(n_colors)) for l in range(p) ])
#obj = 0
model.setObjective(obj, GRB.MINIMIZE)
if not verbose:
model.params.OutputFlag = 0
model.optimize()
# return best room schedule
color_schedule = []
if model.status == GRB.INFEASIBLE:
return color_schedule
for i in range(n_colors):
for l in range(p):
v = model.getVarByName("y_%s_%s" % (i,l))
if v.x == 1:
color_schedule.append(h[l])
break
return color_schedule
def solve_lp(graph):
points = range(graph.n)
lines = graph.lines
gamma_lp = set_up_model("connected_components_lp_2d")
connected_components = graph.connected_components
len_ccs = len(connected_components)
edges = list(graph.edges.iter_subset(points))
x = {}
for (p, q) in edges:
x[p, q] = gamma_lp.addVar(name='edge|%s - %s|' % (p, q))
t = gamma_lp.addVar(obj=1.0)
gamma_lp.modelSense = grb.GRB.MINIMIZE
gamma_lp.update()
# correct number of edges
gamma_lp.addConstr(quicksum(x[i, j] for (i, j) in edges) == (len_ccs - 1))
# crossing number range:
#gamma_lp.addConstr(0 <= t <= math.sqrt(graph.n))
# crossing constraints
for line in lines:
s = quicksum(x[p, q] for (p, q) in edges if has_crossing(line,
graph.get_line_segment(p, q)))
if s != 0.0:
gamma_lp.addConstr(
s <= t)
# connectivity constraint
for cc1 in connected_components:
gamma_lp.addConstr(quicksum(x[p, q] for (p, q) in edges if p < q and
p in cc1 and q not in cc1) +
quicksum(x[q, p] for (q, p) in edges if p > q and
p in cc1 and q not in cc1)
>= 1.)
gamma_lp.optimize()
if gamma_lp.status == grb.GRB.status.OPTIMAL:
add_best_edge(graph, x)
return
else:
format_string = 'Vars:\n'
for var in gamma_lp.getVars():
format_string += "%s\n" % var
print "number of lines = %s" % len(graph.lines)
print '%s\nlp model=%s' % (format_string, gamma_lp)
import spanningtree.plotting
spanningtree.plotting.plot(graph)
raise StandardError('Model infeasible')
def solve_lp_and_round(graph, points):
'''
creates a Gurobi model for Sariel LP formulation and round it
deterministically
return the selection of edges that are in the fractional solution
'''
lines = graph.lines
gamma_lp = set_up_model("sariels_lp_2d")
n = len(points)
edges = list(graph.edges.iter_subset(points))
x = {}
for (p, q) in edges:
x[p, q] = gamma_lp.addVar(obj=graph.euclidean_distance(p, q),
name='edge|%s - %s|' % (p, q))
t = gamma_lp.addVar(obj=1.0)
gamma_lp.modelSense = grb.GRB.MINIMIZE
gamma_lp.update()
# crossing number range:
gamma_lp.addConstr(0 <= t <= math.sqrt(n))
# crossing constraints
for line in lines:
s = quicksum(x[p, q] for (p, q) in edges if has_crossing(line,
graph.get_line_segment(p, q)))
if s != 0.0:
gamma_lp.addConstr(
s <= t)
# connectivity constraint
for p in points:
gamma_lp.addConstr(
quicksum(x[p, q] for q in points if p < q) +
quicksum(x[q, p] for q in points if p > q)
>= 1)
gamma_lp.optimize()
if gamma_lp.status == grb.GRB.status.OPTIMAL:
round_solution = []
for (p, q) in edges:
val = x[p, q].X
sample = random.random()
if sample <= val:
graph.solution.update(p, q, True)
round_solution.append((p, q))
return round_solution
def verifier(model, spec):
"""Find a policy that maximizes the probability of specifiation realization"""
if type(model) == MC:
model_type = 'MC'
elif type(model) == MDP:
model_type = 'MDP'
else:
raise NameError("Given model is not supported for verification")
if len(spec) == 2:
# reach-avoid specification
if len(set.intersection(set(spec[0]),set(spec[1]))) != 0:
print("The specification creates conflict --- reach set updated by removing avoid elements")
spec[1] = list(set.difference(set(spec[1]),set(spec[0])))
for b in spec[0]:
model.make_absorbing(b)
# define linear program model
grb_model = grb.Model(name=model_type+" LP")
grb_model.setParam('OutputFlag', False )
# define variables
n_vars = len(model.states)
xs_vars = {i : grb_model.addVar(vtype=grb.GRB.CONTINUOUS,
lb=0,
ub=1,
name="x_{0}".format(i))
for i in range(n_vars)}
# define constraints
constraints = dict()
n_consts = 0
constraints.update({j : grb_model.addConstr(
lhs=xs_vars[spec[1][j]],
sense=grb.GRB.EQUAL,
rhs=1,
name="constraint_{0}".format(j))
for j in range(len(spec[1]))})
n_consts += len(spec[1])
for s in model.states:
if s not in spec[0]:
if type(model) == MC:
constraints.update({n_consts : grb_model.addConstr(
lhs=xs_vars[s],
sense=grb.GRB.GREATER_EQUAL,
rhs=grb.quicksum([model.transitions[s,j]*xs_vars[j]
for j in model.states]),
name="constraint_{0}".format(n_consts))})
n_consts += 1
elif type(model) == MDP:
for a in model.actions[model.enabled_actions[s]]:
constraints.update({n_consts : grb_model.addConstr(
lhs=xs_vars[s],
sense=grb.GRB.GREATER_EQUAL,
rhs=grb.quicksum([model.transitions[s,a,j]*xs_vars[j]
for j in model.states if model.transitions[s,a,j]!=0]),
name="constraint_{0}".format(n_consts))})
n_consts += 1
# define objective
objective = grb.quicksum([xs_vars[i] for i in range(n_vars)])
# add objective
grb_model.ModelSense = grb.GRB.MINIMIZE
grb_model.setObjective(objective)
# solve
grb_model.write(model_type+".lp")
grb_model.optimize()
# store data
if grb_model.status == grb.GRB.status.OPTIMAL:
grb_model.write(model_type+".sol")
opt_df = pd.DataFrame.from_dict(xs_vars, orient="index",
columns = ["variable_object"])
opt_df.reset_index(inplace=True)
opt_df["solution_value"] = opt_df["variable_object"].apply(lambda item: item.X)
else:
raise NameError("Given specification is not handled")
vars_val = [v.x for v in grb_model.getVars()]
# find the optimal policy
opt_policy = None
if type(model) == MDP:
opt_policy = []
for s in model.states:
n_opt_act = 0
opt_act = []
for a in model.actions[model.enabled_actions[s]]:
if isclose(vars_val[s],
np.sum([model.transitions[s,a,ss]*vars_val[ss]
for ss in model.states if model.transitions[s,a,ss]!=0]),
abs_tol=1e-24):
opt_act.append(a)
n_opt_act += 1
opt_policy.append(opt_act)
return (vars_val, opt_policy)
def create_model_4(depot, hubs, clients, cost_matrix_1, cost_matrix_2):
"""
Model 4 = A simple CVRP with no clients
:param robots: an object with location and number
:param depot: location of the depot station, name of the depot
:param hubs: each hub has a location
:param clients: clients have names,
:param cost_matrix_1: a dict that uses the keys of locations to give you the distance (Depot - Hubs)
:param cost_matrix_2: a dict that uses the keys of locations to give you the distance (Hubs-clients, Hubs-clients)
(ex. cost_matrix_1[h.name,d.name] = distance between hub i and depot j)
:return: Gurobi model with the related constraints
"""
### Creating model:
model = gp.Model("2E-VRP")
#####################################################
# 1. sets & parameters (parameters should be already defined in __init__)
#####################################################
# a. define maximum number of robots:
# TODO: move this to create an object list of robots
n_c = len(clients) # number of customers
R_max = len(clients)
R = list(range(R_max)) # list of robots
# Dep = list(range(len(depot))) # list of depots - this one is wrong
C = [c.name for c in clients] # list of clients
D_c = [c.demand for c in clients] # list of demands
L = list(set([l[0] for l in cost_matrix_2.keys()])) # L = m+n
# TODO: transform L to be a set of str.
H = list(range(len(hubs) + 1)) # list of hubs (hubs 0,1,2)
Loc_c = [c.loc for c in clients] # list of locations
Loc_c_dict = {c.name: c.loc
for c in clients
} # dict of locations - will not be used most probably
R_cap = 50 # maximum robot capacity
R_dist = 200
V_cap = 300
# defining the number of trucks by calculating the total demand
total_demand = math.fsum(c.demand for c in clients)
t = math.ceil((total_demand / V_cap))
#V = list(range(t)) # list of trucks
# fixed costs
c_hub = 50
c_truck = 50
c_robot = 10
# variable cost
c_truck_distance = 5
p = 4 # p will be replaced by the sum of
#####################################################
# 2. decision variables
#####################################################
# hub open/not-open
o = model.addVars(H, vtype=GRB.BINARY, name="o")
# DH = [0]+H # in our case it is only one depot, the model is extendable to include multiple (if we..)
DH = list(range(len(hubs) + 2))
# TODO: remember in the distance matrix hubs are 0,1,2 and depots are 3
y = model.addVars(
DH, DH, vtype=GRB.BINARY,
name="y") # TODO: you need the subtour elimination constraint
# TODO: set the variable to 0 when dh1 = dh2
# TODO: Check if we need the decision variable for the truck
# 1 <= u[.] <= p+1
u = model.addVars(DH, vtype=GRB.INTEGER, ub=p + 1, lb=1,
name="u") # u[V,DH]
#####################################################
# 3. constraints
#####################################################
# (8) - all the trucks must return to the depot station
#model.addConstr((gp.quicksum(y[h, 0] for h in DH[1:]) == 1), name="trucks1")
# (9) - all trucks must depart from the depart station
#model.addConstr((gp.quicksum(y[0, h] for h in DH[1:]) == 1), name="trucks2")
# (10) - number of trucks leaving = number of trucks returning
model.addConstrs((gp.quicksum(y[h1, h2] for h1 in DH[1:] if h1 != h2)
== gp.quicksum(y[h2, h1] for h1 in DH[1:] if h1 != h2)
for h2 in DH[1:]),
name="trucks3")
# (11) - sum of all trucks going to the same hub is 1 if hub is open
model.addConstrs((gp.quicksum(y[h1, h2]
for h1 in DH[1:] if h1 != h2) == o[h2 - 1]
for h2 in DH[1:]),
name="if_truck_then_hub")
# (12) - setting the starting point from main Depot
model.addConstr(
(u[0] == 1),
name="start_from_depot") # when changed to 1 the model is infeasible
# (13) - order of satellites should be between 2 and p+1
model.addConstrs((u[h] >= 2 for h in DH[1:]), name="lowest_hub_order")
model.addConstrs((u[h] <= p + 1 for h in DH[1:]),
name="lowest_hub_order_2")
# (14) - respecting the tour order
model.addConstrs(((u[h1] - u[h2] + 1) <= (p) * (1 - y[h1, h2])
for h1 in range(2, p + 1)
for h2 in range(2, p + 1) if h1 != h2),
name="respect_order"
) # DH should be replaced by the sum of the open hubs
# (16) - if the hub is closed then u should be zero
model.addConstrs((u[h] <= (p + 1) * o[h - 1] for h in DH[1:]),
name="no_hub_no_u")
# (17) - adding constraint 17 from the article for testing
model.addConstr((gp.quicksum(o[h] for h in H) == p), name="constraint_17")
# we shouldn't need constraint 17 if the variable o is activated by a constraint x which is already written
model.addConstrs((y[dh2, dh1] + y[dh1, dh2] <= 1 for dh1 in DH
for dh2 in DH if dh1 != dh2),
name="no_subtours_3")
model.addConstrs((gp.quicksum(y[dh1, dh2] for dh1 in DH if dh1 != dh2) <= 1
for dh2 in DH),
name="no_subtours_1")
model.addConstrs((gp.quicksum(y[dh2, dh1] for dh1 in DH if dh1 != dh2) <= 1
for dh2 in DH),
name="no_subtours_2")
#####################################################
# 4. Objective function
#####################################################
# 1. truck distance cost
cost_truck_var = gp.quicksum(
cost_matrix_1[dh1 - 1, dh2 - 1] * c_truck_distance * y[dh1, dh2]
for dh1 in DH[1:] for dh2 in DH[1:] if dh1 != dh2)
# 2. truck fixed cost
cost_truck_fixed = c_truck * t # this will change in v2
model.setObjective(cost_truck_var + cost_truck_fixed)
#####################################################
# 5. Saving LP model (remember to change versions)
#####################################################
model.write("../output/lp_model/2EVRP_truck_u.lp")
#### Optimize!
model.optimize()
status = model.Status
if status in [GRB.INF_OR_UNBD, GRB.INFEASIBLE, GRB.UNBOUNDED]:
print("Model is either infeasible or unbounded.")
sys.exit(0)
elif status != GRB.OPTIMAL:
print("Optimization terminated with status {}".format(status))
sys.exit(0)
#####################################################
# 6. analyzing solutions
#####################################################
print('Optimal solution found with total cost: %g' % model.objVal)
for v in model.getVars():
if v.x >= 0.5:
print('%s %g' % (v.varName, v.x))
# 1. analyzing hubs
print("Analyzing hubs")
# 2. analyzing robots
print("Analyzing Robots")
# 3. analyzing tours (echelon 1)
print("Analyzing tours (depot - hubs)")
# 4. analyzing tours (echelon 2)
print("Analyzing tours (hubs - clients)")
def _build_objective_StochElecDA(self):
var = self.variables
#generators = self.edata.generators
gfpp = self.edata.gfpp
nongfpp = self.edata.nongfpp
gendata = self.edata.generatorinfo
HR = self.edata.generatorinfo.HR
gaspriceda = self.edata.GasPriceDA
gaspriceRT = self.edata.GasPriceRT
SCdata = self.edata.SCdata
swingcontr = self.edata.swingcontracts
time = self.edata.time
Map_Eg2Gn = self.edata.Map_Eg2Gn
scenarios = self.edata.scenarios
m = self.model
scenarioprob = {}
scenarioprob = {
s: self.edata.scen_wgp[s][2]
for s in self.edata.scen_wgp.keys()
}
scengprt = {
s: self.edata.scen_wgp[s][0]
for s in self.edata.scen_wgp.keys()
}
#scenarioprob = self.data.scenprob['Probability'].to_dict()
P_up = defaults.RESERVES_UP_PREMIUM
P_dn = defaults.RESERVES_DN_PREMIUM
# !NB Re-dispatch cost = Day-ahead energy cost (No premium)
m.setObjective(
# Day-ahead energy cost
# Non Gas Generators
gb.quicksum(gendata.lincost[gen] * var.Pgen[gen, t] for gen in nongfpp
for t in time) +
# Gas Generators = Nodal Gas Price * HR * Power Output
gb.quicksum(gaspriceda[t][Map_Eg2Gn[gen]] * HR[gen] * var.Pgen[gen, t]
for gen in gfpp for t in time) +
# Gas Generators with Contracts
gb.quicksum(SCdata.lambdaC[sc, gen] * HR[gen] * var.PgenSC[gen, t]
for gen in gfpp for sc in swingcontr for t in time) +
# Real-time redispatch cost
# Probability
gb.quicksum(scenarioprob[s] * (
# Non Gas Generators
gb.quicksum(gendata.lincost[gen] *
(P_up * var.RUp[gen, s, t] - P_dn * var.RDn[gen, s, t])
for gen in nongfpp for t in time) +
# Gas Generators
gb.quicksum(gaspriceRT[t][Map_Eg2Gn[gen]][scengprt[s]] * HR[gen] *
(P_up * var.RUp[gen, s, t] - P_dn * var.RDn[gen, s, t])
for gen in gfpp for t in time) +
# Gas Generators with Contracts
gb.quicksum(
SCdata.lambdaC[sc, gen] * HR[gen] *
(P_up * var.RUpSC[gen, s, t] - P_dn * var.RDnSC[gen, s, t])
for gen in gfpp for sc in swingcontr for t in time) +
# Load Shedding Penalty
gb.quicksum(defaults.VOLL * var.Lshed[s, t] for t in time))
for s in scenarios),
gb.GRB.MINIMIZE)
def create_model():
"""
Funcion que crea modelo de optimizacion de multiples posiciones sin descomposicion
"""
start_date = args.start_date
filepath = args.filepath
time_limit = args.time_limit
mip_focus = args.mip_focus
mip_gap = args.mip_gap
m = Model("SSTPA V3")
m.setParam('TimeLimit', time_limit)
m.setParam('MIPFocus', mip_focus)
m.setParam('MIPGap', mip_gap)
# Parse params dict to variables
params = parse_params(filepath, start_date)
N = params['N']
F = params['F']
S = params['S']
I = params['I']
R = params['R']
L = params['L']
M = 10 ** 10
EL = params['EL']
EV = params['EV']
PI = params['PI']
#################
# * VARIABLES * #
#################
# x_nf: x[partido, fecha]
# 1 si el partido n se programa finalmente
# en la fecha f
# 0 en otro caso.
x = m.addVars(N, F, vtype=GRB.BINARY, name="x")
# y_is: y[equipo][patron_localias]
# 1 si al equipo i se le asigna el patron
# de localias s
# 0 en otro caso
y = {i: m.addVars(S[i], vtype=GRB.BINARY, name="y") for i in I}
# p_jilf: P[equipo, equipo, fecha, fecha]
# discreta, cant de puntos del equipo j al finalizar la fecha f
# con la info de los resultados hasta la fecha l inclusive
# en el MEJOR conjunto de resultados futuros para el equipo i
p_m = m.addVars(I, I, F, F, vtype=GRB.INTEGER, name="p_m")
# p_jilf: P[equipo, equipo, fecha, fecha]
# discreta, cant de puntos del equipo j al finalizar la fecha f
# con la info de los resultados hasta la fecha l inclusive
# en el PEOR conjunto de resultados futuros para el equipo i
p_p = m.addVars(I, I, F, F, vtype=GRB.INTEGER, name="p_p")
# v_nilf : v[partido, equipo, fecha, fecha]
# binaria, 1 si el equipo local gana el partido n
# de la fecha f teniendo informacion finalizada la fecha l
# en el MEJOR conjunto de resultados futuros para el equipo i
v_m = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="v_m")
# v_nilf : v[partido, equipo, fecha, fecha]
# binaria, 1 si el equipo local gana el partido n
# de la fecha f teniendo informacion finalizada la fecha l
# en el MEJOR conjunto de resultados futuros para el equipo i
v_p = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="v_p")
# a_nilf: a[partido,equipo,fecha,fecha]
# 1 si el equipo visitante gana el partido n de la fecha f
# teniendo información finalizada la fecha l
# en el MEJOR conjunto de resultados para el equipo i
a_m = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="a_m")
# a_nilf: a[partido,equipo,fecha,fecha]
# 1 si el equipo visitante gana el partido n de la fecha f
# teniendo información finalizada la fecha l
# en el PEOR conjunto de resultados para el equipo i
a_p = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="a_p")
# e_nilf: e[partido,equipo,fecha,fecha]
# binaria, toma el valor 1 si se empata el
# partido n de la fecha f, con la info
# de los resultados hasta la fecha l inclusive
# en el MEJOR conjunto de resultados futuros para el euqipo i
e_m = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="e_m")
# e_nilf: e[partido,equipo,fecha,fecha]
# binaria, toma el valor 1 si se empata el
# partido n de la fecha f, con la info
# de los resultados hasta la fecha l inclusive
# en el PEOR- conjunto de resultados futuros para el euqipo i
e_p = m.addVars(N, I, F, F, vtype=GRB.BINARY, name="e_p")
# alfa_jil : alfa[equipo,equipo,fecha]
# binaria, toma el valor 1 si el equipo j termina con menos puntos
# que el equipo i en el
# MEJOR conjunto de
# resultados futuros para el equipo i considerando que
# se está en la fecha l
alfa_m = m.addVars(I, I, F, vtype=GRB.BINARY, name="alfa_m")
# alfa_jil : alfa[equipo,equipo,fecha]
# binaria, toma el valor 1 si el equipo j tiene termina
# con menos puntos que el equipo i, en el PEOR conjunto de
# resultados futuros para el equipo i considerando que
# se está en la fecha l
alfa_p = m.addVars(I, I, F, vtype=GRB.BINARY, name="alfa_p")
# beta_il: beta[equipo,fecha]
# discreta, indica la mejor posicion
# que puede alcanzar el equipo i al final del
# torneo, mirando desde la fecha l en el MEJOR
# conjunto de resultados futuros para el equipo i
beta_m = m.addVars(I, F, vtype=GRB.INTEGER, name="beta_m")
# beta_il: beta[equipo, fecha]
# discreta, indica la mejor posicion
# que puede alcanzar el equipo i al final del
# torneo, mirando desde la fecha l en el PEOR
# conjunto de resultados futuros para el equipo i
beta_p = m.addVars(I, F, vtype=GRB.INTEGER, name="beta_p")
#####################
# * RESTRICCIONES * #
#####################
# R2
for n in N:
m.addConstr((quicksum(x[n, f] for f in F) == 1), name=f"R2-{n}")
# R3
for i in I:
for f in F:
_exp = LinExpr(quicksum(x[n, f] for n in N if EL[i][n] + EV[i][n] == 1))
m.addConstr(_exp == 1, name=f"R3-{i}-{f}")
# R4
m.addConstrs((quicksum(y[i][s] for s in S[i]) == 1 for i in I), name="R4")
# R6
for f in F:
for i in I:
_exp1 = LinExpr(quicksum(x[n, f] for n in N if EL[i][n] == 1))
_exp2 = LinExpr(quicksum(y[i][s] for s in S[i] if L[s][f] == 1))
m.addConstr(_exp1 == _exp2, name=f"R6-{f}-{i}")
# R7
for f in F:
for i in I:
_exp1 = LinExpr(quicksum(x[n, f] for n in N if EV[i][n] == 1))
_exp2 = LinExpr(quicksum(y[i][s] for s in S[i] if L[s][f] == 0))
m.addConstr(_exp1 == _exp2, name=f"R7-{f}-{i}")
# R8
for n in N:
for i in I:
for f in F:
for l in F:
if f > l:
_exp = LinExpr(v_m[n, i, l, f] + e_m[n, i, l, f] + a_m[n, i, l, f])
m.addConstr(x[n, f] == _exp, name=f"R8-{n}-{i}-{f}-{l}")
# R8
for n in N:
for i in I:
for f in F:
for l in F:
if f > l:
_exp = LinExpr(v_p[n, i, l, f] + e_p[n, i, l, f] + a_p[n, i, l, f])
m.addConstr(x[n, f] == _exp, name=f"R9-{n}-{i}-{f}-{l}")
# R10
for j in I:
for i in I:
for f in F:
for l in F:
_exp1 = LinExpr(quicksum(quicksum(R[j][n] * x[n, theta]
for n in N if EL[j][n] + EV[j][n] == 1)
for theta in F if theta <= l))
_exp2 = LinExpr(quicksum(quicksum(3 * v_m[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EL[j][n] == 1))
_exp3 = LinExpr(quicksum(quicksum(3 * a_m[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EV[j][n] == 1))
_exp4 = LinExpr(quicksum(quicksum(e_m[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EL[j][n] + EV[j][n] == 1))
m.addConstr(p_m[j, i, l, f] == PI[j] + _exp1 + _exp2 + _exp3 + _exp4,
name=f"R10-{j}-{i}-{f}-{l}")
# R10
for j in I:
for i in I:
for f in F:
for l in F:
_exp1 = LinExpr(quicksum(quicksum(R[j][n] * x[n, theta]
for n in N if EL[j][n] + EV[j][n] == 1)
for theta in F if theta <= l))
_exp2 = LinExpr(quicksum(quicksum(3 * v_p[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EL[j][n] == 1))
_exp3 = LinExpr(quicksum(quicksum(3 * a_p[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EV[j][n] == 1))
_exp4 = LinExpr(quicksum(quicksum(e_p[n, i, l, theta]
for theta in F if theta > l and theta <= f)
for n in N if EL[j][n] + EV[j][n] == 1))
m.addConstr(p_p[j, i, l, f] == PI[j] + _exp1 + _exp2 + _exp3 + _exp4,
name=f"R11-{j}-{i}-{f}-{l}")
# R12
for l in F:
for i in I:
for j in I:
if j != i:
m.addConstr(M - M * alfa_m[j, i, l] >= 1 + p_m[j, i, l, F[-1]] - p_m[i, i, l, F[-1]],
name=f"R12-{l}-{i}-{j}")
# R13
for l in F:
for i in I:
for j in I:
if j != i:
m.addConstr(M * alfa_p[j, i, l] >= 1 + p_p[j, i, l, F[-1]] - p_p[i, i, l, F[-1]],
name=f"R13-{l}-{i}-{j}")
# R14
for i in I:
for l in F:
_exp = LinExpr(quicksum(alfa_m[j, i, l] for j in I if i != j))
m.addConstr(beta_m[i, l] == len(I) - _exp, name=f"R14-{i}-{l}")
# R15
for i in I:
for l in F:
_exp = LinExpr(quicksum(alfa_p[j, i, l] for j in I if i != j))
m.addConstr(beta_p[i, l] == len(I) - _exp, name=f"R15-{i}-{l}")
########################
# * FUNCION OBJETIVO * #
########################
_obj = quicksum(quicksum(beta_p[i, l] - beta_m[i, l] for i in I) for l in F)
m.setObjective(_obj, GRB.MAXIMIZE)
return m
N_nurses_it = model.addVars(time,
clinics,
vtype=GRB.INTEGER,
name="N_nurses_it")
H_nurses_it = model.addVars(time,
clinics,
vtype=GRB.INTEGER,
name="H_nurses_it")
F_nurses_it = model.addVars(time,
clinics,
vtype=GRB.INTEGER,
name="F_nurses_it")
############################# OBJECTIVE FUNCTION ###########################
transport_part = gp.quicksum(Kgmt[T - 1][M - 1][G - 1] * Ngmt[T, M, G]
for G in gmsd for M in manufacturers
for T in time)
transport_part += gp.quicksum(Ksgt[T - 1][G - 1][S - 1] * Nsgt[T, G, S]
for S in svs for G in gmsd for T in time)
transport_part += gp.quicksum(Krst[T - 1][S - 1][R - 1] * Nrst[T, S, R]
for R in rvs for S in svs for T in time)
transport_part += gp.quicksum(Kdrt[T - 1][R - 1][D - 1] * Ndrt[T, R, D]
for D in dvs for R in rvs for T in time)
transport_part += gp.quicksum(Kidt[T - 1][D - 1][I - 1] * Nidt[T, D, I]
for I in clinics for D in dvs for T in time)
inventory_part = gp.quicksum(hgt[T - 1][G - 1] * Igt[T, G] for G in gmsd
for T in time)
inventory_part += gp.quicksum(hst[T - 1][S - 1] * Ist[T, S] for S in svs
for T in time)
inventory_part += gp.quicksum(hrt[T - 1][R - 1] * Irt[T, R] for R in rvs
lbs = np.ones(num_node, dtype=np.uint8)
vals = lbs
resource = num_node * resource_rate
model = gp.Model('ADG')
obj = model.addVar(vtype=GRB.CONTINUOUS, name="obj")
x = model.addVars(x_keys, ub=1, vtype=GRB.INTEGER, name='x')
r = model.addVars(range(num_node), vtype=GRB.CONTINUOUS, name='r')
b = model.addVars(b_keys, vtype=GRB.CONTINUOUS, name='b')
model.addConstr(r.sum() <= resource, name='total resource sum')
# model.addConstrs((b[(i, j, k)] <= w_mtx[j, k] * r[j] for (i, j, k) in b_keys), name='borrowing limitations')
model.addConstrs((gp.quicksum((np.ones_like(knbr_dict[i]) - x.select(i, knbr_dict[i])) * vals[knbr_dict[i]]) <= obj
for i in range(num_node)), name='minimize maximum')
for ra in range(num_node):
model.addConstrs((r[k] - gp.quicksum(b.select(ra, k, '*')) + gp.quicksum(b.select(ra, '*', k))
>= lbs[k] * x[(ra, k)] for k in knbr_dict[ra]),
name='reallocation results for node {}'.format(i))
model.addConstrs((gp.quicksum(b.select(ra, k, '*')) <= r[k] for k in range(num_node)),
name='total out-borrowings')
model.setObjective(obj, GRB.MINIMIZE)
pre_end = time.time()
time_diff(start_all, pre_end, "preprocessing")
model.optimize()
def blocki_node_triange_count(net0, epsilon, k, reps):
net = renumber_node_ids(net0)
num_nodes = nx.number_of_nodes(net)
num_edges = nx.number_of_edges(net)
# Linear Programming Model
lpm = grb.Model()
lpm.Params.LogFile = "gurobi.log"
lpm.Params.LogToConsole = 0
lpm.Params.Threads = 32
print("Prepare to add vars", num_nodes, num_edges)
w = lpm.addVars(2 * num_edges, name="w")
x = lpm.addVars(num_nodes, name="x")
print("Finish to add vars")
print("x", x[5241])
w_map = {}
count = 0
for (u, v) in net.edges():
w_map[(u, v)] = w[count]
count += 1
w_map[(v, u)] = w[count]
count += 1
print("Prepare to add x constraints")
lpm.addConstrs(x[i] >= 0 for i in range(num_nodes))
print("Added x constraints")
for (u, v) in net.edges():
lpm.addConstr(w_map[(u, v)] >= 0)
lpm.addConstr(w_map[(v, u)] >= 0)
lpm.addConstr(w_map[(u, v)] >= 1 - x[u] - x[v])
lpm.addConstr(w_map[(v, u)] >= 1 - x[u] - x[v])
lpm.addConstr(w_map[(u, v)] <= 1)
lpm.addConstr(w_map[(v, u)] <= 1)
for i in range(num_nodes):
lpm.addConstr(grb.quicksum(w_map[(i, j)]
for j in nx.neighbors(net, i)) <= k)
print("Add all constraints")
lpm.setObjective(grb.quicksum(x[i] for i in range(num_nodes)),
grb.GRB.MINIMIZE)
lpm.optimize()
# return lpm.ObjVal
x_star = {}
for v in lpm.getVars():
if "x" in v.varName:
index = int(v.varName[2:-1])
x_star[index] = v.x
#if "w" in v.varName:
# if v.x < 1.0:
# print(v.x)
trim_net = nx.Graph()
for (u, v) in net.edges():
if x_star[u] < 1/4 and x_star[v] < 1/4:
trim_net.add_edge(u, v)
count = common.triangle_count(trim_net)[0]
print("Trim count:", count)
result = {}
for eps in epsilon:
delta = eps
if delta == 1.0:
delta = 0.999 #to avoid division by 0
result[eps] = count + np.random.laplace(0,
S_beta(beta=-eps / 2 * math.log(delta),
d_hat=2 * num_edges / num_nodes,
k=k,
c=4) / eps,
reps)
return result
def __init__(self,
n_vertices,
edges,
constraints,
k,
gamma,
verbosity=0,
symmetry_breaking=True,
overlap=False,
single_cut=False,
timeout=None):
self.check_graph(n_vertices, edges)
self.n_vertices = n_vertices
self.k = k
self.verbosity = verbosity
self.timeout = timeout
model = Model('graph_clustering')
mvars = []
for i in range(k):
cvars = []
for j in range(n_vertices):
v = model.addVar(lb=0.0, ub=1.0, vtype=GRB.BINARY)
cvars.append(v)
mvars.append(cvars)
model.update()
ineq_sense = GRB.GREATER_EQUAL if overlap else GRB.EQUAL
# constraint: each vertex in exactly/at least one cluster
for v in range(n_vertices):
model.addConstr(quicksum([mvars[i][v] for i in range(k)]),
ineq_sense, 1)
# symmetry-breaking constraints
if symmetry_breaking:
model.addConstr(mvars[0][0], GRB.EQUAL, 1)
for i in range(2, k):
model.addConstr(
quicksum([mvars[i - 1][j] for j in range(n_vertices)]) <=
quicksum([mvars[i][j] for j in range(n_vertices)]))
obj_expr = LinExpr()
wsum = sum(w for (_, _, w) in constraints)
gamma = gamma / wsum
# indicators for violation of cl constraints
for (u, v, w) in constraints:
for i in range(k):
y = model.addVar(lb=0.0, ub=1.0, vtype=GRB.BINARY)
model.update()
model.addConstr(y >= mvars[i][u] + mvars[i][v] - 1)
obj_expr.add(y, -w * gamma)
# size of smallest cluster
s = model.addVar(lb=0.0, ub=n_vertices, vtype=GRB.INTEGER)
model.update()
for i in range(k):
model.addConstr(
s <= quicksum([mvars[i][v] for v in range(n_vertices)]))
s_coef = 1 / n_vertices if overlap else k / n_vertices
obj_expr.add(s_coef * s)
model.setObjective(obj_expr, GRB.MAXIMIZE)
model.params.OutputFlag = self.verbosity
model.Params.PreCrush = 1
model.Params.LazyConstraints = 1
model._cutfinder = Cut_Finder(n_vertices, edges)
model._vars = mvars
model._k = k
model._relobj = None
model._impcounter = 0
model._single_cut = single_cut
# runtime information
model._root_cuttime = 0
model._tree_cuttime = 0
self.model = model
def BarycenterLinf(images):
""" Compute the Kantorovich Wasserstein Barycenter of order 1 with Linf as ground distance """
K = len(images)
n = len(images[0])
s = int(np.sqrt(n))
def ID(x, y):
return x * s + y
# Build model
m = Model()
m.setParam(GRB.Param.Method, 2)
m.setParam(GRB.Param.Crossover, 0)
m.setParam(GRB.Param.NumericFocus, 1)
m.setParam(GRB.Param.OutputFlag, 0)
m.setAttr(GRB.Attr.ModelSense, 1)
# Create variables
Z = {}
for i in range(n):
Z[i] = m.addVar(obj=0)
X = {}
for k in range(K):
for i in range(s):
for j in range(s):
X[k, ID(i, j)] = {}
X[k, n] = {}
A = []
for k in range(K):
for i in range(s):
for j in range(s - 1):
X[k, ID(i, j)][k, ID(i, j + 1)] = m.addVar(obj=1)
X[k, ID(i, j + 1)][k, ID(i, j)] = m.addVar(obj=1)
if k == 0:
A.append((ID(i, j), ID(i, j + 1)))
A.append((ID(i, j + 1), ID(i, j)))
for i in range(s - 1):
for j in range(s):
X[k, ID(i, j)][k, ID(i + 1, j)] = m.addVar(obj=1)
X[k, ID(i + 1, j)][k, ID(i, j)] = m.addVar(obj=1)
if k == 0:
A.append((ID(i, j), ID(i + 1, j)))
A.append((ID(i + 1, j), ID(i, j)))
for i in range(s - 1):
for j in range(s - 1):
X[k, ID(i, j)][k, ID(i + 1, j + 1)] = m.addVar(obj=1)
X[k, ID(i + 1, j + 1)][k, ID(i, j)] = m.addVar(obj=1)
X[k, ID(i, j + 1)][k, ID(i + 1, j)] = m.addVar(obj=1)
X[k, ID(i + 1, j)][k, ID(i, j + 1)] = m.addVar(obj=1)
if k == 0:
A.append((ID(i, j), ID(i + 1, j + 1)))
A.append((ID(i + 1, j + 1), ID(i, j)))
A.append((ID(i, j + 1), ID(i + 1, j)))
A.append((ID(i + 1, j), ID(i, j + 1)))
A = tuplelist(A)
m.update()
# Flow variables
for i in range(n):
Fs = A.select(i, '*')
Bs = A.select('*', i)
for k in range(K):
m.addConstr(
quicksum(X[k, i][k, j] for _, j in Fs) -
quicksum(X[k, j][k, i] for j, _ in Bs) == images[k][i] - Z[i])
m.addConstr(quicksum(Z[i] for i in range(n)) == 1.0)
# Solve the model
m.optimize()
return m.getAttr(GRB.Attr.ObjVal), [Z[i].X for i in range(n)]
def create_model(hubs, clients, cost_matrix_1, cost_matrix_2):
"""
Model to assign robots to clients respecting:
1. Robot maximum capacity
2. Robot maximum distance
3. Client Time Windows
:param hubs:
:param clients:
:param cost_matrix_1:
:param cost_matrix_2:
:return: robot assignment and demand for each hub
"""
### Start Time:
start_time = time.time()
### Creating model:
model = gp.Model("RAP")
#####################################################
# 1. sets & parameters (parameters should be already defined in __init__)
#####################################################
# a. define maximum number of robots:
n_c = len(clients) # number of customers
R_max = len(
clients
) - 25 #TODO: The number of solutions (for Instance 1 - robots needed: 5)
R = list(range(R_max)) # list of robots
C = [c.name for c in clients] # list of clients
D_c = [c.demand for c in clients] # list of demands
# time windows and duration:
st = {c.name: c.st for c in clients}
t1 = {c.name: c.tw1 for c in clients}
t2 = {c.name: c.tw2 for c in clients}
Loc_c = [c.loc for c in clients] # list of locations
Loc_c_dict = {c.name: c.loc
for c in clients
} # dict of locations - will not be used most probably
L = list(set([l[0] for l in cost_matrix_2.keys()])) # L = m+n
H = list(range(len(hubs))) # list of hubs (hubs 0,1,2)
R_cap = 100 # maximum robot capacity
R_dist = 400
# fixed costs
c_hub = 50
c_robot = 10
# set upper bound to the last operation last time window
op_time = max(t2.values()) + 15
#####################################################
# 2. decision variables
#####################################################
# hub open/not-open
o = model.addVars(H, vtype=GRB.BINARY, name="o")
# robot-client-hub assignment
x = model.addVars(C, R, H, vtype=GRB.BINARY, name="x")
# robot - hub assignment
rob = model.addVars(R, H, vtype=GRB.BINARY, name="rob")
# edges assignment to robots
y = model.addVars(L, L, R, vtype=GRB.BINARY, name="y")
# Start time of service
# TODO: Get the maximum latest time window
t = model.addVars(L, ub=op_time,
name="t") # for instance 1, the maximum time is 263
# Artificial variables to correct time window upper and lower limits
xa = model.addVars(C, name="xa")
xb = model.addVars(C, name="xb")
#####################################################
# 3. constraints
#####################################################
# A robot must be assigned to a client
model.addConstrs(
(gp.quicksum(x[c, r, h] for r in R for h in H) == 1 for c in C),
name="client_must_be_served")
# (4) - if robot serves a client, the corresponding hub must be open
model.addConstrs(
(gp.quicksum(x[c, r, h] for r in R) <= o[h] for c in C for h in H),
name="no_hub_no_robot")
model.addConstrs((rob[r, h] <= o[h] for r in R for h in H),
name="no_hub_no_robot_2")
# [2] - At most one robot can be assigned to a job
model.addConstrs((x.sum(c, '*', h) <= 1 for c in C for h in H),
name="one_robot")
# Robot serves from a single hub:
# (D) - each client is served by a single robot from a single hub
# model.addConstrs((gp.quicksum(x[c, r, h] for h in H for r in R) == 1 for c in C), name="Just_one_hub")
# (C) - Robot maximum capacity constraint
model.addConstrs((gp.quicksum(D_c[int(c)] * x[c, r, h] for c in C
for h in H) <= R_cap * rob.sum(r, '*')
for r in R),
name="robot_cap")
# (A) - distance covered by every robot should be less than maximum allowed distance
model.addConstrs((gp.quicksum(cost_matrix_2[i, j] * y[i, j, r] for i in L
for j in L if i != j) <= R_dist for r in R),
name="robot_dist")
# Robot tour constraints
model.addConstrs(
(gp.quicksum(y[int(j), int(i), r]
for j in L if int(j) != int(i)) == gp.quicksum(x[i, r, h]
for h in H)
for r in R for i in C),
name="Tour1")
model.addConstrs(
(gp.quicksum(y[int(i), int(j), r]
for j in L if int(j) != int(i)) == gp.quicksum(x[i, r, h]
for h in H)
for r in R for i in C),
name="Tour2")
# Additional Tour Constraints:
model.addConstrs((y[int(j), int(i), r] + y[int(i), int(j), r] <= 1
for r in R for i in L for j in L if int(j) != int(i)),
name="Tour3")
# Ensuring Tours in "y" are respecting the same hub the robot is serving from
model.addConstrs((gp.quicksum(y[int(c), n_c + h, r]
for c in C) == rob[r, h] for r in R
for h in H),
name="sameHub1_y")
model.addConstrs((gp.quicksum(y[n_c + h, int(c), r]
for c in C) == rob[r, h] for r in R
for h in H),
name="sameHub2_y")
# Ensuring robot assignment in "x" are respecting the same hub the robot is serving from
model.addConstrs((gp.quicksum(x[c, r, h]
for c in C) <= (n_c + 1) * (rob[r, h])
for r in R for h in H),
name="sameHub_x")
#model.addConstrs((gp.quicksum(x[c, r, h] for c in C) <= (n_c+1)*(rob[r, h]) for r in R for h in H),
# name="sameHub2_x")
# Temporal constraints for client locations (here we assume distance and time have 2:1 ratio)
#M = {(i, j): op_time + st[i] + (cost_matrix_2[int(i), int(j)]/2) for i in C for j in C}
#model.addConstrs((t[int(j)] >= t[int(i)] + st[i] + (cost_matrix_2[int(i), int(j)]/2)
# - M[i, j] * (1 - gp.quicksum(y[int(i), int(j), r] for r in R))
# for i in C for j in C), name="tempoClients")
# Temporal constraints for hub locations
#M = {(i, j): op_time + (cost_matrix_2[int(i), int(j)]/2) for i in H for j in C}
#model.addConstrs((t[int(j)] >= t[int(i)] + (cost_matrix_2[int(i), int(j)]/2)
# - M[i, j] * (1 - gp.quicksum(y[int(i), int(j), r] for r in R)) for i in H for j in C),
# name="tempoHub")
# Time window constraints
#model.addConstrs((t[int(c)] + xa[c] >= t1[c] for c in C), name="timeWin1")
#model.addConstrs((t[int(c)] - xb[c] <= t2[c] for c in C), name="timeWin2")
# Robot serves from one hub only
model.addConstrs((gp.quicksum(rob[r, h] for h in H) <= 1 for r in R),
name="one_hub_per_robot")
# Testing
# model.addConstr((gp.quicksum(o[h] for h in H) == 3), name="all_hubs_open")
########################################################
# Subtour Elimination
########################################################
# Miller-Tucker-Zemlin formulation (Bektas version):
# How to define p?? The maximum number of clients a robot can visit, or the maximum number of hub a truck can visit
# 1 <= u[.] <= p+1
# p1 = len(L)
# u1 = model.addVars(L, H, vtype=GRB.INTEGER, ub=p1 + 1, lb=1, name="u")
# (4) - respecting the tour order
# model.addConstrs(((u1[i,h] - u1[j,h] + p1 * (gp.quicksum(y[i, j, r] for r in R))) <= p1 - 1
# for i in range(0, p1) for j in range(0, p1) for h in H if i != j),
# name="respect_order_clients") # DH should be replaced by the sum of the open hubs
p1 = n_c
u1 = model.addVars(L, vtype=GRB.INTEGER, ub=p1 + 1, lb=1, name="u")
# (4) - respecting the tour order
model.addConstrs(
((u1[i] - u1[j] + p1 * (gp.quicksum(y[i, j, r] for r in R))) <= p1 - 1
for i in range(0, p1) for j in range(0, p1) if i != j),
name="respect_order_clients"
) # DH should be replaced by the sum of the open hubs
#########################################################
# 2nd part - Truck assignment
#########################################################
# defining the number of trucks by calculating the total demand
V_cap = 300 # Truck Max Capacity
total_demand = math.fsum(c.demand for c in clients)
t = math.ceil((total_demand / V_cap))
V = list(range(
t +
1)) # list of trucks - adding an additional truck for safety reasons
c_truck = 50 # Truck Fixed Cost
c_truck_distance = 5 # Truck Variable Cost
DH = list(range(len(hubs) + 1)) # List of Depot and Hubs
z = model.addVars(V, DH, DH, vtype=GRB.BINARY, name="z")
# Truck hub assignment
w = model.addVars(V, vtype=GRB.BINARY, name="w")
# (8) - all the trucks must return to the depot station
model.addConstrs(
(gp.quicksum(z[v, len(hubs), h] for h in DH[:-1]) == w[v] for v in V),
name="trucks1")
# (9) - all trucks must depart from the depart station
model.addConstrs(
(gp.quicksum(z[v, h, len(hubs)] for h in DH[:-1]) == w[v] for v in V),
name="trucks2")
# (11) - sum of all trucks going to the same hub is 1 if hub is open
model.addConstrs((gp.quicksum(z[v, h1, h2] for v in V
for h1 in DH if h1 != h2) == o[h2]
for h2 in DH[:-1]),
name="if_truck_then_hub")
# (11new) - Modified to have more than one truck visiting the hub.
#model.addConstrs((gp.quicksum(z[v, h1, h2] for v in V for h1 in DH if h1 != h2)
# >= o[h2] for h2 in DH[:-1]), name="if_truck_then_hub")
# (10) - mirroring Constraint 11
model.addConstrs((gp.quicksum(z[v, h1, h2] for v in V
for h1 in DH if h1 != h2) == gp.quicksum(
z[v, h2, h1] for v in V
for h1 in DH if h1 != h2) for h2 in DH),
name="trucks3")
# Truck tour constraints
#model.addConstrs((gp.quicksum(z[v, j, h] for j in DH for v in V if j != h)
# >= o[h] for h in H),
# name="Tour1_truck")
#model.addConstrs((gp.quicksum(z[v, h, j] for j in DH for v in V if j != h)
# >= o[h] for h in H),
# name="Tour2_truck")
# Eliminating Subtours
#model.addConstrs((gp.quicksum(z[v, dh1, dh2] for v in V for dh1 in DH[:-1] if dh1 != dh2)
# <= 1 for dh2 in DH), name="no_subtours_1")
#model.addConstrs((gp.quicksum(z[v, dh2, dh1] for v in V for dh1 in DH[:-1] if dh1 != dh2)
# <= 1 for dh2 in DH), name="no_subtours_2")
#model.addConstrs((z[v, dh2, dh1] + z[v, dh1, dh2] <= 1
# for v in V for dh1 in DH for dh2 in DH if dh1 != dh2), name="no_subtours_3")
# Demand per Hub
D_h = model.addVars(H, ub=total_demand, name="D_h")
model.addConstrs((D_h[h] == gp.quicksum(D_c[int(c)] * x[c, r, h] for c in C
for r in R) for h in H),
name="total_demand_per_hub")
# Maximum Truck Capacity
model.addConstrs((gp.quicksum(D_h[h] * z[v, i, h] for i in DH
for h in H) <= V_cap * w[v] for v in V),
name="truck_capacity")
# Miller-Tucker-Zemlin formulation (Bektas 2006 version):
# How to define p?? The maximum number of clients a robot can visit, or the maximum number of hub a truck can visit
# 1 <= u[.] <= p+1
p = len(DH) #n_c
u2 = model.addVars(DH, vtype=GRB.INTEGER, ub=p + 1, lb=1, name="u2")
# (u1) - setting the starting point from main Depot
model.addConstr((u2[len(hubs)] == 1), name="start_from_depot")
# (u2-3) - order of satellites should be between 2 and p+1
model.addConstrs((u2[h] >= 2 for h in DH[:-1]), name="lowest_hub_order")
model.addConstrs((u2[h] <= p + 1 for h in DH[:-1]),
name="highest_hub_order_2")
# (4) - respecting the tour order
model.addConstrs(
((u2[h1] - u2[h2] + p * (gp.quicksum(z[v, h1, h2]
for v in V))) <= p - 1
for h1 in range(0, p - 1) for h2 in range(0, p - 1) if h1 != h2),
name="respect_order"
) # DH should be replaced by the sum of the open hubs
# (16) - if the hub is closed then u should be zero
#model.addConstrs((u[h] <= (p + 1) * o[h - 1] for h in DH[1:]), name="no_hub_no_u")
#####################################################
# 4. Objective function
#####################################################
# 1. robot distance cost
cost_robot_var = gp.quicksum(cost_matrix_2[c1, c2] * y[c1, c2, r]
for c1 in L for c2 in L if c1 != c2
for r in R)
# 2. robot fixed cost
cost_robot_fixed = gp.quicksum(c_robot * rob[r, h] for r in R for h in H)
# 3. hub fixed cost
cost_hub_fixed = gp.quicksum(c_hub * o[h] for h in H)
# 4. truck distance cost
cost_truck_var = gp.quicksum(cost_matrix_1[dh1, dh2] * c_truck_distance *
z[v, dh1, dh2] for v in V for dh1 in DH
for dh2 in DH if dh1 != dh2)
# 5. truck fixed cost
cost_truck_fixed = c_truck * gp.quicksum(
w[v] for v in V) # this will change in v2
model.setObjective(
cost_robot_var + cost_robot_fixed + cost_hub_fixed + cost_truck_fixed +
cost_truck_var, GRB.MINIMIZE)
#####################################################
# 5. Saving LP model (remember to change versions)
#####################################################
model.write("../output/lp_model/RAP_TRP_model6_v2.lp")
#### Optimize!
model.optimize()
#####################################################
# 6. analyzing solutions
#####################################################
print('Optimal solution found with total cost: %g' % model.objVal)
for v in model.getVars():
if v.x >= 0.5:
print('%s %g' % (v.varName, v.x))
### Printing execution Time:
print("Executed in %s Minutes" % ((time.time() - start_time) / 60))
# Robots:
print("Analyzing Solutions...")
print("1. Robots:")
robots = {r: [] for r in R}
active_robots = []
for r in R:
robot = ""
hub = ""
clients_served = []
clients_served_real = []
for h in H:
for c in C:
if x[c, r, h].X > 0.5:
robot = str(r + 1)
hub = str(h + 1)
clients_served.append(int(c) + 1)
clients_served_real.append(int(c))
if robot:
print("Robot {}, will be serving from hub {}.".format(robot, hub))
print("The robot will serve the following clients: {}".format(
clients_served))
active_robots.append(r)
if hub:
clients_served_real.append(n_c + int(hub) - 1)
robots[r] = clients_served_real
# Robot tours:
print("2. Robot tours:")
tours = {r: [] for r in R}
for r in R:
links = list()
for i in L:
for j in L:
if y[i, j, r].X > 0.5:
links.append((i, j))
tours[r] = links
print("Links visited by each robot: ")
print(tours)
#####################################################
# 7. printing tours
#####################################################
import matplotlib.pyplot as plt
import networkx as nx
G = nx.DiGraph()
list_nodes = list(range(1, len(L))) # list(range(1, len(L) + 1))
G.add_nodes_from(list_nodes)
nodes_clients = {int(c.name): c.loc for c in clients}
nodes_hubs = {n_c + h: hubs[h] for h in H}
node_pos = {**nodes_clients, **nodes_hubs}
# Create a list of nodes in shortest path
for r in active_robots:
# Create a list of edges in shortest path
red_edges = [(i, j) for i in L for j in L if y[i, j, r].x > 0.5]
for i in L:
for j in L:
if y[i, j, r].x > 0.5:
G.add_edge(i, j)
# If the node is in the shortest path, set it to red, else set it to white color
node_col = [
'white' if not node in robots[r] else 'red' for node in G.nodes()
]
# If the edge is in the shortest path set it to red, else set it to white color
edge_col = [
'black' if not edge in red_edges else 'red' for edge in G.edges()
]
# Draw the nodes
nx.draw_networkx(G, node_pos, node_color=node_col, node_size=450)
# Draw the node labels
# nx.draw_networkx_labels(G1, node_pos,node_color= node_col)
# Draw the edges
nx.draw_networkx_edges(G, node_pos, edge_color=edge_col)
# Draw the edge labels
#cost_matrix_2_to_int = cost_matrix_2
#for sub in cost_matrix_2_to_int:
# cost_matrix_2_to_int[sub] = int(cost_matrix_2_to_int[sub])
#nx.draw_networkx_edge_labels(G, node_pos, edge_color=edge_col, edge_labels=cost_matrix_2_to_int)
# Remove the axis
plt.axis('off')
# TODO: Add description of the plot
# Remove edges
G.remove_edges_from(list(G.edges()))
# Show the plot
# plt.show()
# Save the plot
#plt.savefig("../output/plots/scenario1_satellite_double/Model_5_tour-robot_Ca2-3-15_{}.png".format(r + 1))
#plt.savefig("../output/plots/scenario2_robot_capacity/Model_5_tour-robot_Ca2-3-15_{}.png".format(r + 1))
#plt.savefig("../output/plots/scenario3_satellite_cost_robot_distance/Model_5_tour-robot_Ca2-3-15_{}.png".format(r + 1))
#plt.savefig("../output/plots/normal_scenario/Model_5_tour-robot_Ca2-3-15_{}.png".format(r + 1))
plt.savefig(
"../output/plots/Ca1-3,30/Model_6_tour-robot_Ca1-3,30_robot_{}.png"
.format(r + 1))
plt.clf()
a = m.addVars(I, F, vtype=GRB.BINARY, name="a")
# d_if: d[equipo, fecha]
# 1 si el partido del equipo i en la fecha f
# es atractivo por poder descender.
# 0 en otro caso.
d = m.addVars(I, F, vtype=GRB.BINARY, name="d")
print(f"** VARIABLES TIME: {time.time() - start_model}")
#####################
#* RESTRICCIONES *#
#####################
# R2
m.addConstrs((quicksum(x[n, f] for f in F) == 1 for n in N), name="R2")
# R3
m.addConstrs((quicksum(x[n, f] for n in N if EL[i][n] + EV[i][n] == 1) == 1
for i in I for f in F),
name="R3")
# R4
for i in I:
m.addConstr((quicksum(y[i][s] for s in S[i]) == 1), name="R4")
# R6
for i in I:
m.addConstrs((quicksum(x[n, f] for n in N if EL[i][n] == 1) == quicksum(
y[i][s] for s in S[i] if L[s][f] == 1) for f in F),
name="R6")
def constr_4(m, assign, z, C_pop, matches, T_n, mean_T_pop, Treat, Covar):
m.addConstrs(z[i] >= -quicksum(
((assign.T[t] * C_pop[i]) / (matches * T_n)).sum()
for t in Treat) - mean_T_pop[i] for i in Covar)
def optimize(hp, Rt, bp, slopes, theta):
"""
:param R: Returns at time t (Gross)
:param hp: Post-decision variable (pre-return) at time t-1
:param bp: Breakpoints at time t
:param slopes: Slopes at time t
:return: Post-decision variable at time t + DeltaV
"""
grad = [] # gradient deltaV_(t-1)
h = Rt * hp # element wise multiplication
x, y = getcoord(bp, slopes)
# print(x)
# print(y)
N = len(bp) - 1 # no of holdings, excluding cash
m = Model()
m.setParam('OutputFlag', False)
"""Add Variables"""
# add x_i, y_i as variables (all x first, then all y)
# 2N variables
xv = array([m.addVar() for _ in range(N)]) # Buys at time t
yv = array([m.addVar() for _ in range(N)]) # Sales at time t
# also add hpv for convenience (we'll have equality const relating to xv,yv)
hpv = array([m.addVar(lb=0) for _ in range(N + 1)]) # h_plus at time t
m.update()
"""Add Objective Function"""
outputCashFlow = (1 + theta) * quicksum(xv) - (1 - theta) * quicksum(yv)
m.setPWLObj(hpv[0], x[0], y[0])
for i in range(1, N + 1):
m.setPWLObj(hpv[i], x[i], y[i])
"""Add Constraints"""
eqCstrs = [
m.addConstr(hpv[i] - h[i] == xv[i - 1] - yv[i - 1])
for i in range(1, N + 1)
]
eqCstrs.append(m.addConstr(hpv[0] - h[0] == -1 * outputCashFlow))
holdingCstrs = [
m.addConstr(-xv[i - 1] + yv[i - 1] <= h[i]) for i in range(1, N + 1)
]
budgetCstr = m.addConstr(outputCashFlow <= h[0])
m.ModelSense = -1 # maximize
m.update()
try:
m.optimize()
except GurobiError as e:
pass
print("failed")
# print(m.Status)
# optimal variables
# print([(v.varName, v.X) for v in m.getVars()])
# optimal h vector
hopt = []
for i in range(N + 1):
hopt.append(hpv[i].X)
# get optimal function value
Vold = m.ObjVal
# solve N+1 times, incrementing hp component wise
for i in range(N + 1):
hp1 = np.copy(hp)
hp1[i] = hp[i] + 1
h = Rt * hp1 # element wise multiplication
x, y = getcoord(bp, slopes)
N = len(bp) - 1 # no of holdings, excluding cash
m = Model()
m.setParam('OutputFlag', False)
"""Add Variables"""
# add x_i, y_i as variables (all x first, then all y)
# 2N variables
xv = array([m.addVar() for _ in range(N)]) # Buys at time t
yv = array([m.addVar() for _ in range(N)]) # Sales at time t
# also add hpv for convenience (we'll have equality const relating to xv,yv)
hpv = array([m.addVar(lb=0) for _ in range(N + 1)]) # h_plus at time t
m.update()
"""Add Objective Function"""
outputCashFlow = (1 + theta) * quicksum(xv) - (1 -
theta) * quicksum(yv)
m.setPWLObj(hpv[0], x[0], y[0])
for i in range(1, N + 1):
m.setPWLObj(hpv[i], x[i], y[i])
"""Add Constraints"""
eqCstrs = [
m.addConstr(hpv[i] - h[i] == xv[i - 1] - yv[i - 1])
for i in range(1, N + 1)
]
eqCstrs.append(m.addConstr(hpv[0] - h[0] == -1 * outputCashFlow))
holdingCstrs = [
m.addConstr(-xv[i - 1] + yv[i - 1] <= h[i])
for i in range(1, N + 1)
]
budgetCstr = m.addConstr(outputCashFlow <= h[0])
m.ModelSense = -1 # maximize
m.update()
try:
m.optimize()
except GurobiError as e:
pass
print("failed")
# get optimal function value
Vnew = m.ObjVal
grad.append(Vnew - Vold)
return (np.asarray(hopt), np.asarray(grad))
def k_medoids(distances, number_clusters, timelimit=100, mipgap=0.0001):
"""
Parameters
----------
distances : 2d array
Distances between each pair of node points. `distances` is a
symmetrical matrix (dissimmilarity matrix).
number_clusters : integer
Given number of clusters.
timelimit : integer
Maximum time limit for the optimization.
"""
# Distances is a symmetrical matrix, extract its length
length = distances.shape[0]
# Create model
model = gp.Model("k-Medoids-Problem")
# Create variables
x = {} # Binary variables that are 1 if node i is assigned to cluster j
y = {} # Binary variables that are 1 if node j is chosen as a cluster
for j in xrange(length):
y[j] = model.addVar(vtype="B", name="y_"+str(j))
for i in xrange(length):
x[i,j] = model.addVar(vtype="B", name="x_"+str(i)+"_"+str(j))
# Update to introduce the variables to the model
model.update()
# Set objective - equation 2.1, page 509, [1]
obj = gp.quicksum(distances[i,j] * x[i,j]
for i in xrange(length)
for j in xrange(length))
model.setObjective(obj, gp.GRB.MINIMIZE)
# s.t.
# Assign all nodes to clusters - equation 2.2, page 509, [1]
# => x_i cannot be put in more than one group at the same time
for i in xrange(length):
model.addConstr(sum(x[i,j] for j in xrange(length)) == 1)
# Maximum number of clusters - equation 2.3, page 509, [1]
model.addConstr(sum(y[j] for j in xrange(length)) == number_clusters)
# Prevent assigning without opening a cluster - equation 2.4, page 509, [1]
for i in xrange(length):
for j in xrange(length):
model.addConstr(x[i,j] <= y[j])
for j in xrange(length):
model.addConstr(x[j,j] >= y[j])
# Sum of main diagonal has to be equal to the number of clusters:
model.addConstr(sum(x[j,j] for j in xrange(length)) == number_clusters)
# Set solver parameters
model.Params.TimeLimit = timelimit
model.Params.MIPGap = mipgap
# Solve the model
model.optimize()
# Get results
r_x = np.array([[x[i,j].X for j in range(length)]
for i in range(length)])
r_y = np.array([y[j].X for j in xrange(length)])
r_obj = model.ObjVal
return (r_y, r_x.T, r_obj)
def mtsp_solver(cost, salesmen=1, min_cities=None, max_cities=None, **kwargs):
"""
Multiple traveling salesmen MILP solver using the Gurobi MILP optimiser.
:rtype : Returns tuple with the routes for each salesman, the objective value, and a model object.
:param cost: Cost matrix for travelling from point to point.
:param salesmen: Number of salesmen taking part in the solution.
:param min_cities: Optional parameter of minimum cities to be visited by each salesman.
:param max_cities: Optional parameter of maximum cities to be visited by each salesman.
"""
n = cost.shape[0]
if min_cities is None:
K = 2
else:
K = min_cities
if max_cities is None:
L = n
else:
L = max_cities
m = gurobipy.Model()
# Create variables
e_vars = {}
for i in range(n):
for j in range(n):
e_vars[i, j] = m.addVar(obj=cost[i, j],
vtype=gurobipy.GRB.BINARY,
name='e' + str(i) + '_' + str(j))
m.update()
for i in range(n):
e_vars[i, i].ub = 0
u_vars = {}
for i in range(n):
u_vars[i] = m.addVar(lb=0,
ub=L,
vtype=gurobipy.GRB.INTEGER,
name='u' + str(i))
m.update()
# Add degree-2 constraint, and forbid loops
m.addConstr(
gurobipy.quicksum(e_vars[0, i] for i in range(1, n)) == salesmen)
m.addConstr(
gurobipy.quicksum(e_vars[i, 0] for i in range(1, n)) == salesmen)
for i in range(1, n):
m.addConstr(
gurobipy.quicksum(e_vars[i, j] for j in range(n) if i != j) == 1)
for i in range(1, n):
m.addConstr(
gurobipy.quicksum(e_vars[j, i] for j in range(n) if i != j) == 1)
for i in range(1, n):
m.addConstr(u_vars[i] + (L - 2) * e_vars[0, i] - e_vars[i, 0] <= L - 1)
for i in range(1, n):
m.addConstr(u_vars[i] + e_vars[0, i] + (2 - K) * e_vars[i, 0] >= 2)
for i in range(1, n):
m.addConstr(e_vars[0, i] + e_vars[i, 0] <= 1)
for i in range(1, n):
for j in range(1, n):
if i != j:
m.addConstr(u_vars[i] - u_vars[j] + L * e_vars[i, j] +
(L - 2) * e_vars[j, i] <= L - 1)
m.update()
m._vars = e_vars
m._uvars = u_vars
m.params.OutputFlag = int(kwargs.get('output_flag', 0))
m.params.TimeLimit = float(kwargs.get('time_limit', 60.0))
m.optimize()
solution = m.getAttr('X', e_vars)
selected = [(i, j) for i in range(n) for j in range(n)
if solution[i, j] > 0.5]
routes = []
for i in range(salesmen):
routes.append([])
next_city = 0
finished = False
while not finished:
for j in range(len(selected)):
if selected[j][0] == next_city:
routes[i].append(next_city)
next_city = selected[j][1]
selected.pop(j)
break
if next_city == 0:
finished = True
return routes, m.objVal, m
def optimize_path(self,
active_set=None,
inactive_set_val=None,
initial_soln=None,
verbose=False):
'''
:param initial_soln: Initial guess for solution to optimization problem
:param active_set: Every variable in the active set will be optimized over
Every Variable not in the active set will be set to the constant as
provided in inital_soln
:return: Waypoints corresponding to path
'''
m = gp.Model('Planner')
m.Params.timeLimit = self.time_limit
m.Params.Presolve = self.presolve
x, u, q, cvx_constraints, dyn_constraints \
= (list(), list(), list(), list(), list())
if active_set is not None:
assert active_set.dtype == bool, "active set must be np array of bools"
active_set = active_set.copy(
) #we change this later so make a copy
else:
active_set = np.ones((self.T, len(self.env.get_cvx_ineqs()[0])),
dtype=bool)
if inactive_set_val is not None:
assert inactive_set_val.dtype == bool, "inactive_set_value must be bools"
inactive_set_val.copy() #we change this later so make a copy
else:
inactive_set_val = active_set.copy()
for t in range(self.T):
x.append(
m.addMVar(4,
lb=-GRB.INFINITY,
vtype=GRB.CONTINUOUS,
name=f"x_{t}"))
u.append(
m.addMVar(2,
lb=-GRB.INFINITY,
vtype=GRB.CONTINUOUS,
name=f"u_{t}"))
q.append(x[-1][:2])
# THis wont work in general it has to be specified as part of
#one cnvex region in the environment
# q[-1].setAttr("lb", 0.0)
# q[-1].setAttr("ub", 100.0)
# m.addConstr(q[-1] <= 100*np.ones((2,)))
xbinvars, x_row_constraints = (list(), list())
const_constraints = list()
assert np.any(active_set[t]) or np.any(inactive_set_val[t]),\
"[ERROR] Inconsistent problem definition. Each waypoint must have " \
"at least one convex region constraint active."
for region_idx_ in range(self.num_regions):
bin_var, row_constraints = self._add_cvx_region_constraint(
q[-1], region_idx_, t, m)
xbinvars.append(bin_var)
x_row_constraints.append(row_constraints)
if not active_set[t, region_idx_]:
const_constraints.append(
m.addConstr(bin_var == inactive_set_val[t,
region_idx_],
name=f"x_{t}_cvx_{region_idx_}_FON"))
anycvx = m.addConstr(gp.quicksum(xbinvars) >= 1,
name=f"x_{t}_cvx_any")
cvx_constraints.append(
(xbinvars, x_row_constraints, const_constraints, anycvx))
init_constraint = m.addMConstrs(A=np.eye(4),
x=x[0],
sense=GRB.EQUAL,
b=np.array(self.env.start),
name='x_init')
end_constraint = m.addMConstrs(A=np.eye(4),
x=x[-1],
sense=GRB.EQUAL,
b=np.array(self.env.end),
name='x_f')
for t in range(self.T - 1):
dyn_constraints.append(
m.addConstr(
x[t + 1] - x[t] == \
0.5 * self.h_k * ( \
(self.dyn.A @ x[t] + self.dyn.A @ x[t + 1]) + \
(self.dyn.B @ u[t] + self.dyn.B @ u[t + 1])),
name=f'dyn_{t}'))
m.setObjective(sum([ut @ np.eye(2) @ ut for ut in u]), GRB.MINIMIZE)
if initial_soln is not None:
(x_init, u_init) = initial_soln
self._initialize_variables(x, u, x_init, u_init)
m.optimize()
runtime = m.getAttr('Runtime')
if m.getAttr('Status') == GRB.INFEASIBLE:
print('[WARNING] Model provided is infeasible.')
return x_init, u_init, float(
'Inf'), active_set, inactive_set_val, runtime
elif m.getAttr('Status') == GRB.INF_OR_UNBD:
assert False, "[ERROR] Should never happen."
elif m.getAttr('Status') == GRB.TIME_LIMIT:
print('[INFO] MIP time-limit termination.')
print(cvx_constraints)
self._update_active_set(active_set, inactive_set_val,
[c[0] for c in cvx_constraints])
self.last_x = x
self.last_u = u
self.last_m = m
xnp = np.stack([xt.getAttr('X') for xt in x])
unp = np.stack([ut.getAttr('X') for ut in u])
objective = m.getObjective().getValue()
return xnp, unp, objective, active_set, inactive_set_val, runtime
def assignmentFunction(allowableChanges,
methodFlag): #don't need init solution
#from math import floor, sqrt
#for non-integer division, convert to floats!
#import pandas as pd
import numpy as np
import gurobipy as gp
import random as rnd
sol = Solution([0, 0], 1, 0)
#Because Python doesn't have select-case
if methodFlag == 0: #SMA"Warmstart" -- may drop
print "zero"
elif methodFlag == 1: #SMAColdstart
#sets
KD_Off = range(74, 160)
B_Off = range(0, 74)
KD_Ass = range(0, 71)
B_Ass = range(73, 139)
D_Ass = range(139, 162)
#Officers = list(set().union(KD_Off,B_Off))
#Assignments = list(set().union(KD_Ass, B_Ass, D_Ass))
#Match KD Assignments
KD_D_Ass = D_Ass[0:15]
p_kd = load_obj('p_kd')
yg = load_obj('yg')
#need opref and apref
n = len(KD_Off)
#rankO = np.zeros(n+1).tolist() #Officer to assignment currO[assignment]
#rankO = [-1]*(n)
#provides ordinal ranking of how far "down" each officers list is current match
#rankA = [-1]*(n)
KD_OtoA = {}
opref = {}
apref = {}
for o in KD_Off:
opref[o] = p_kd[o]
KD_OtoA[o] = -1
for assignment in KD_Ass:
apref[assignment] = {}
preference = 0
for year in sorted(yg.keys()):
preference += 1
for officer in yg[year]:
apref[assignment][officer] = preference
# for assignment in B_Ass:
# apref[assignment]=[0]*(n)
# for officer in KD_Off:
# apref[assignment][officer] = 2
# for officer in B_Off:
# apref[assignment][officer] = 1
#
# for assignment in KD_D_Ass:
# apref[assignment] = [0]*(n)
unmatched = KD_Off #listing of unmatched officers
noKD = [] #bookkeeping for KD Officers not getting a KD Assignment
nextKD_D = KD_D_Ass.pop()
while unmatched: #implict boolean false for empty list
#find first unmatched officer)
officer = rnd.choice(unmatched) # = rankA[1:].index(0)+1
#rankA[officer] = rankA[officer] + 1
#don't loop, choose next best.
#if len(opref[officer]) > 1:
# possibilities = opref[officer].pop()
#else:
# possibilities = opref[officer]
while 1:
try:
possibilities = opref[officer].pop()
except IndexError:
KD_OtoA[officer] = nextKD_D
nextKD_D = KD_D_Ass.pop()
break
if len(possibilities) > 0:
possibility = rnd.choice(possibilities)
possibilities.remove(possibility)
if not possibilities:
opref[officer].append(possibilities)
if possibility in KD_Ass:
break
print possibility
print "-----\nO:" + str(officer)
print "p:-----"
print possibilities
#if len(possibilities) >= 1:
# possibility = rnd.choice(possibilities)
# possibilities.remove(possibility)
#else:
# possibility = possibilities[0]
if len(possibilities) > 0:
#if sublist not empty, replace on stack
opref[officer].append(possibilities)
#change to rank++ below
try:
possibilitypref = apref[possibility]
except KeyError:
print "KEYERROR!"
break
try:
#incumbent = KD_OtoA.index(possibility)
incumbent = KD_OtoA.keys()[KD_OtoA.values().index(possibility)]
except ValueError:
incumbent = -1
#no incumbent
if incumbent == -1:
unmatched.remove(officer)
rankO[possibility] = possibilitypref[officer]
KD_OtoA[officer] = possibility #rankA[officer] = possibility
#if assignment prefers officer to incumbent
elif possibilitypref[officer] > possibilitypref[incumbent]:
#match officer to possibility
#break match of possibility if needed
KD_OtoA[incumbent] = -1
unmatched.append(incumbent)
KD_OtoA[officer] = possibility
rankO[possibility] = possibilitypref[officer]
unmatched.remove(officer)
#elif possibilitypref[officer] == possibilitypref[incumbent]:
print KD_OtoA
sol.finalSolution = KD_OtoA
sol.resultStatusFlag = 0
sol.changes = 0
#save_obj(OtoA, 'smaA')
elif methodFlag == 2: #LPWarmstart
kd_m = gp.Model()
y = load_obj('y')
lpA = load_obj('lpA')
#sets
KD_Off = range(74, 160)
B_Off = range(0, 74)
KD_Ass = range(0, 71)
B_Ass = range(73, 139)
D_Ass = range(139, 162)
Officers = list(set().union(KD_Off, B_Off))
Assignments = list(set().union(KD_Ass, B_Ass, D_Ass))
#handlemods
#allowableChanges[0]: Assignment Restrictions
restrictions = rnd.sample(Officers, allowableChanges[0])
A_r = {}
for restriction in restrictions:
A_r[restriction] = rnd.sample(
Assignments, int(rnd.uniform(.05, .1) * len(Assignments)))
#allowableChanges[1]: Directed Assignment
A_d = {}
availO = set(Officers)
availO -= set(restrictions)
directeds = rnd.sample(list(availO), allowableChanges[1])
availA = set(Assignments)
for directed in directeds:
availA -= set(A_d.values())
A_d[directed] = rnd.sample(list(availA), 1)
A_d[directed] = A_d[directed][0]
#allowableChanges[2]: Rejected Match
availO -= set(directeds)
rejects = []
while len(rejects) < allowableChanges[2]:
reject = rnd.sample(list(availO), 1)[0]
if int(lpA[reject]) != 999:
rejects.append(reject)
availO.remove(reject)
#allowableChanges[3]: Unforecast Assignment
#Need to remove all references, these are just the dummy assignments
#mildly embarassing... but reference in write up
#Phase I, slot KD Offs/objective f_1
x = {}
for o in KD_Off:
for a in KD_Ass:
x[(o, a)] = kd_m.addVar(vtype=gp.GRB.BINARY,
obj=y[o],
name="O{0:03d}".format(o) +
"A{0:03d}".format(a))
if lpA[o] == a:
x[(o, a)].Start = 1
else:
x[(o, a)].Start = 0
kd_m.ModelSense = gp.GRB.MINIMIZE #MINIMIZE
for a in KD_Ass:
kd_m.addConstr(gp.quicksum(x[(o, a)] for o in KD_Off),
gp.GRB.EQUAL, 1)
for o in KD_Off:
if o in list(restrictions):
if not list(set(A_r[o]) & set(KD_Ass)):
kd_m.addConstr(
gp.quicksum(x[(o, a)]
for a in list(set(A_r[o]) & set(KD_Ass))),
gp.GRB.EQUAL, 1)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 0)
elif o in list(directeds):
if A_d[o] in list(KD_Ass):
kd_m.addConstr(x[o, A_d[o]], gp.GRB.EQUAL, 1)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 0)
elif o in list(rejects):
if lpA[o] in list(KD_Ass):
kd_m.addConstr(x[o, int(lpA[o])], gp.GRB.EQUAL, 0)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
kd_m.update()
kd_m.setParam('OutputFlag', False)
#kd_m.write('lp.mps')
kd_m.optimize()
#print kd_m.Status
try:
y_star = kd_m.objVal
except:
return Solution(-1 * np.ones(160), -1, -1)
#Continue with phase2, slot everyone- obj f_2
C = load_obj('C')
x = {} #reallocating memory
bd_m = gp.Model()
for o in Officers:
for a in Assignments:
x[(o, a)] = bd_m.addVar(vtype=gp.GRB.BINARY,
obj=C[(o, a)],
name="O{0:03d}".format(o) +
"A{0:03d}".format(a))
bd_m.ModelSense = gp.GRB.MINIMIZE #-1
for a in Assignments:
bd_m.addConstr(gp.quicksum(x[(o, a)] for o in Officers),
gp.GRB.EQUAL, 1)
for o in Officers:
if o in list(restrictions):
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in A_r[o]),
gp.GRB.EQUAL, 1)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
elif o in list(directeds):
bd_m.addConstr(x[o, A_d[o]], gp.GRB.EQUAL, 1)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
elif o in list(rejects):
bd_m.addConstr(x[o, lpA[o]], gp.GRB.EQUAL, 0)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
else:
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
bd_m.addConstr(
gp.quicksum(
gp.quicksum(y[o] * x[(o, a)] for a in KD_Ass)
for o in Officers), gp.GRB.LESS_EQUAL, 1.0015 * y_star)
bd_m.setParam('OutputFlag', False)
bd_m.update()
bd_m.optimize()
#print bd_m.Status
#unassigned =[]
sol.finalSolution = np.zeros(160)
try:
for v in bd_m.getVars():
if v.x > 0:
if int(v.varName[-3:]) >= 140:
#unassigned.append(int(v.varName[1:4]))
sol.finalSolution[int(v.varName[1:4])] = 999
else:
sol.finalSolution[int(v.varName[1:4])] = int(
v.varName[-3:])
sol.finalSolution = sol.finalSolution.astype(int)
sol.resultStatusFlag = 0 #Good Execution
#save_obj(sol.finalSolution, 'lpA') #saved locally first time and referenced later
#lpA = load_obj('lpA')
sol.changes = sum(i != j for i, j in zip(sol.finalSolution, lpA))
except:
sol = Solution(-1 * np.ones(160), -1, -1)
elif methodFlag == 3: #LPColdstart
kd_m = gp.Model()
y = load_obj('y')
lpA = load_obj('lpA')
#sets
KD_Off = range(74, 160)
B_Off = range(0, 74)
KD_Ass = range(0, 71)
B_Ass = range(73, 139)
D_Ass = range(139, 162)
Officers = list(set().union(KD_Off, B_Off))
Assignments = list(set().union(KD_Ass, B_Ass, D_Ass))
#handlemods
#allowableChanges[0]: Assignment Restrictions
restrictions = rnd.sample(Officers, allowableChanges[0])
A_r = {}
for restriction in restrictions:
A_r[restriction] = rnd.sample(
Assignments, int(rnd.uniform(.05, .1) * len(Assignments)))
#allowableChanges[1]: Directed Assignment
A_d = {}
availO = set(Officers)
availO -= set(restrictions)
directeds = rnd.sample(list(availO), allowableChanges[1])
availA = set(Assignments)
for directed in directeds:
availA -= set(A_d.values())
A_d[directed] = rnd.sample(list(availA), 1)
A_d[directed] = A_d[directed][0]
#allowableChanges[2]: Rejected Match
availO -= set(directeds)
rejects = []
while len(rejects) < allowableChanges[2]:
reject = rnd.sample(list(availO), 1)[0]
if int(lpA[reject]) != 999:
rejects.append(reject)
availO.remove(reject)
#allowableChanges[3]: Unforecast Assignment
#Need to remove all references, these are just the dummy assignments
#mildly embarassing... but reference in write up
#Phase I, slot KD Offs/objective f_1
x = {}
for o in KD_Off:
for a in KD_Ass:
x[(o, a)] = kd_m.addVar(vtype=gp.GRB.BINARY,
obj=y[o],
name="O{0:03d}".format(o) +
"A{0:03d}".format(a))
kd_m.ModelSense = gp.GRB.MINIMIZE #MINIMIZE
for a in KD_Ass:
kd_m.addConstr(gp.quicksum(x[(o, a)] for o in KD_Off),
gp.GRB.EQUAL, 1)
for o in KD_Off:
if o in list(restrictions):
if not list(set(A_r[o]) & set(KD_Ass)):
kd_m.addConstr(
gp.quicksum(x[(o, a)]
for a in list(set(A_r[o]) & set(KD_Ass))),
gp.GRB.EQUAL, 1)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 0)
elif o in list(directeds):
if A_d[o] in list(KD_Ass):
kd_m.addConstr(x[o, A_d[o]], gp.GRB.EQUAL, 1)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 0)
elif o in list(rejects):
if lpA[o] in list(KD_Ass):
kd_m.addConstr(x[o, int(lpA[o])], gp.GRB.EQUAL, 0)
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
else:
kd_m.addConstr(gp.quicksum(x[(o, a)] for a in KD_Ass),
gp.GRB.EQUAL, 1)
kd_m.update()
kd_m.setParam('OutputFlag', False)
#kd_m.write('lp.mps')
kd_m.optimize()
#print kd_m.Status
try:
y_star = kd_m.objVal
except:
return Solution(-1 * np.ones(160), -1, -1)
#Continue with phase2, slot everyone- obj f_2
C = load_obj('C')
x = {} #reallocating memory
bd_m = gp.Model()
for o in Officers:
for a in Assignments:
x[(o, a)] = bd_m.addVar(vtype=gp.GRB.BINARY,
obj=C[(o, a)],
name="O{0:03d}".format(o) +
"A{0:03d}".format(a))
bd_m.ModelSense = gp.GRB.MINIMIZE #-1
for a in Assignments:
bd_m.addConstr(gp.quicksum(x[(o, a)] for o in Officers),
gp.GRB.EQUAL, 1)
for o in Officers:
if o in list(restrictions):
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in A_r[o]),
gp.GRB.EQUAL, 1)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
elif o in list(directeds):
bd_m.addConstr(x[o, A_d[o]], gp.GRB.EQUAL, 1)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
elif o in list(rejects):
bd_m.addConstr(x[o, lpA[o]], gp.GRB.EQUAL, 0)
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
else:
bd_m.addConstr(gp.quicksum(x[(o, a)] for a in Assignments),
gp.GRB.EQUAL, 1)
bd_m.addConstr(
gp.quicksum(
gp.quicksum(y[o] * x[(o, a)] for a in KD_Ass)
for o in Officers), gp.GRB.LESS_EQUAL, 1.0015 * y_star)
bd_m.setParam('OutputFlag', False)
bd_m.update()
bd_m.optimize()
#print bd_m.Status
#unassigned =[]
sol.finalSolution = np.zeros(160)
try:
for v in bd_m.getVars():
if v.x > 0:
if int(v.varName[-3:]) >= 140:
#unassigned.append(int(v.varName[1:4]))
sol.finalSolution[int(v.varName[1:4])] = 999
else:
sol.finalSolution[int(v.varName[1:4])] = int(
v.varName[-3:])
sol.finalSolution = sol.finalSolution.astype(int)
sol.resultStatusFlag = 0 #Good Execution
#save_obj(sol.finalSolution, 'lpA') #saved locally first time and referenced later
#lpA = load_obj('lpA')
sol.changes = sum(i != j for i, j in zip(sol.finalSolution, lpA))
except:
sol = Solution(-1 * np.ones(160), -1, -1)
else: #Error
print "Parameter Error: Invalid methodFlag"
sol = Solution(-1 * np.ones(160), -1, -1)
return sol
def movrp_solver(cost,
start=None,
finish=None,
num_agents=None,
min_cities=None,
max_cities=None,
**kwargs):
"""
Open vehicle routing problem solver for a single vehicle using the Gurobi MILP optimiser.
:param cost: Cost matrix for traveling from point to point.
:param start: Optional starting point for the tour. If none is provided the first point of the array is chosen
:param finish: Optional ending point of the tour. If none is provided the last point of the array is chosen
:return: Returns the route the cost and the model.
"""
# Number of points
n = cost.shape[0]
# Check for default values
if start is None:
start = 0
if finish is None:
finish = n - 1
if num_agents is None:
num_agents = 1
if min_cities is None:
K = 0
else:
K = min_cities
if max_cities is None:
L = n
else:
L = max_cities
m = gurobipy.Model()
# Create model variables
e_vars = {}
for i in range(n):
for j in range(n):
e_vars[i, j] = m.addVar(obj=cost[i, j],
vtype=gurobipy.GRB.BINARY,
name='e' + str(i) + '_' + str(j))
m.update()
for i in range(n):
e_vars[i, i].ub = 0
m.update()
u_vars = {}
for i in range(n):
u_vars[i] = m.addVar(vtype=gurobipy.GRB.INTEGER, name='u' + str(i))
m.update()
# None exits the finish point
m.addConstr(gurobipy.quicksum(e_vars[finish, j] for j in range(n)) == 0)
m.update()
m.addConstr(
gurobipy.quicksum(e_vars[j, finish] for j in range(n)) == num_agents)
m.update()
# From all other points someone exits
for i in range(n):
if i != finish and i != start:
m.addConstr(gurobipy.quicksum(e_vars[i, j] for j in range(n)) == 1)
m.update()
# None enters the starting point
m.addConstr(gurobipy.quicksum(e_vars[j, start] for j in range(n)) == 0)
m.update()
m.addConstr(
gurobipy.quicksum(e_vars[start, j] for j in range(n)) == num_agents)
m.update()
# To all other points someone enters
for i in range(n):
if i != start and i != finish:
m.addConstr(gurobipy.quicksum(e_vars[j, i] for j in range(n)) == 1)
m.update()
# Sub-tour elimination constraint
for i in range(n):
for j in range(n):
if i != j:
m.addConstr(u_vars[i] - u_vars[j] + n * e_vars[i, j] <= n - 1)
m.update()
# for i in range(n):
# if i != start and i != finish:
# m.addConstr(
# u_vars[i] + (L - 2) * gurobipy.quicksum(e_vars[k, i] for k in (start,finish)) -
# gurobipy.quicksum(e_vars[i, k] for k in (start,finish)) <= (L - 1)
# )
# m.update()
#
# for i in range(n):
# if i != start and i != finish:
# m.addConstr(
# u_vars[i] + gurobipy.quicksum(e_vars[k, i] for k in (start,finish)) +
# (2 - K) * gurobipy.quicksum(e_vars[i, k] for k in (start,finish)) >= 2
# )
# m.update()
#
# for k in (start,finish):
# for i in range(n):
# if i != start and i != finish:
# m.addConstr(e_vars[k, i] + e_vars[i, k] <= 1)
# m.update()
#
# for i in range(n):
# if i != start and i != finish:
# for j in range(n):
# if i != j and j != start and j != finish:
# m.addConstr(u_vars[i] - u_vars[j] + L * e_vars[i, j] + (L - 2) * e_vars[j, i] <= L - 1)
# m.update()
m._vars = e_vars
m._uVars = u_vars
m.params.OutputFlag = int(kwargs.get('output_flag', 0))
m.params.TimeLimit = float(kwargs.get('time_limit', 60.0))
m.params.MIPGap = float(kwargs.get('mip_gap', 0.0))
m.optimize()
solution = m.getAttr('X', e_vars)
selected = [(i, j) for i in range(n) for j in range(n)
if solution[i, j] > 0.5]
routes = []
for i in range(num_agents):
routes.append([])
next_city = selected[i][0]
finished = False
routes[i].append(next_city)
next_city = selected[i][1]
while not finished:
for j in range(len(selected)):
if selected[j][0] == next_city:
routes[i].append(next_city)
next_city = selected[j][1]
break
if next_city == finish:
routes[i].append(next_city)
finished = True
return routes, m.objVal, m
def solve_model(n, U, w, b_data):
assert(n > 0 and U > 0)
print((n, U))
print(b_data)
with gp.Model("MaxCov") as model:
delta = model.addVars(n, vtype=GRB.BINARY, name="delta")
# z does not require depot
z = model.addVars(U, n, lb=1, ub=n, vtype=GRB.INTEGER, name="z")
# X_{u}[i,j]
# n + 1 because of depot
X = model.addVars(U, n + 1, n + 1, vtype=GRB.BINARY, name="X")
b = model.addVars(U, vtype=GRB.CONTINUOUS, name="b")
model.addConstrs((b[i] == b_data[i] for i in range(U)), name="b_const")
model.addConstrs((X.sum(u, '*', j) == X.sum(u, j, "*")
for u in range(U)
for j in range(n + 1)), name="X_req0")
#depot constr
model.addConstrs((X.sum(u, n, '*') == 1
for u in range(U)),
name="depot_constr")
# hamilton constrs
model.addConstrs((z[u, j] - z[u, i] >= X[u, i, j] + n * (X[u, i, j] - 1)
for u in range(U)
for i in range(n) # depot is excluded
for j in range(n)), name="hamilton_constr")
# cost constraints
model.addConstrs(gp.quicksum(w[i][j] * X[u, i, j]
for i in range(n)
for j in range(n)) +
#depot start
gp.quicksum(w[n + u][k] * X[u, n, k] for k in range(n)) +
#depot arrive
gp.quicksum(w[i][ n + u] * X[u, i, n] for i in range(n)) <=
b[u] for u in range(U))
# delta constraint
model.addConstrs(delta[i] <= X.sum("*", i, '*')
for i in range(n))
# add delta maximisation objective
model.ModelSense = GRB.MAXIMIZE
model.setObjectiveN(delta.sum("*"), index=0, priority=2, name= "MaxDelta")
model.update()
model.write("test.lp")
model.optimize()
status = model.Status
if status in (GRB.INF_OR_UNBD, GRB.INFEASIBLE, GRB.UNBOUNDED):
print('Model cannot be solved because it is infeasible or unbounded')
sys.exit(0)
if status != GRB.OPTIMAL:
print('Optimization was stopped with status ' + str(status))
sys.exit(0)
return _parse_results(model)
def submit_assigment_problem(env):
# Number of workers required for each shift
shifts, shiftRequirements = gp.multidict({
"Mon1": 3,
"Tue2": 2,
"Wed3": 4,
"Thu4": 4,
"Fri5": 5,
"Sat6": 5,
"Sun7": 3,
"Mon8": 2,
"Tue9": 2,
"Wed10": 3,
"Thu11": 4,
"Fri12": 5,
"Sat13": 7,
"Sun14": 5,
})
# Amount each worker is paid to work one shift
workers, pay = gp.multidict({
"Amy": 10,
"Bob": 12,
"Cathy": 10,
"Dan": 8,
"Ed": 8,
"Fred": 9,
"Gu": 11,
})
# Worker availability
availability = gp.tuplelist([
('Amy', 'Tue2'), ('Amy', 'Wed3'), ('Amy', 'Thu4'), ('Amy', 'Sun7'),
('Amy', 'Tue9'), ('Amy', 'Wed10'), ('Amy', 'Thu11'), ('Amy', 'Fri12'),
('Amy', 'Sat13'), ('Amy', 'Sun14'), ('Bob', 'Mon1'), ('Bob', 'Tue2'),
('Bob', 'Fri5'), ('Bob', 'Sat6'), ('Bob', 'Mon8'), ('Bob', 'Thu11'),
('Bob', 'Sat13'), ('Cathy', 'Wed3'), ('Cathy', 'Thu4'),
('Cathy', 'Fri5'), ('Cathy', 'Sun7'), ('Cathy', 'Mon8'),
('Cathy', 'Tue9'), ('Cathy', 'Wed10'), ('Cathy', 'Thu11'),
('Cathy', 'Fri12'), ('Cathy', 'Sat13'), ('Cathy', 'Sun14'),
('Dan', 'Tue2'), ('Dan', 'Thu4'), ('Dan', 'Fri5'), ('Dan', 'Sat6'),
('Dan', 'Mon8'), ('Dan', 'Tue9'), ('Dan', 'Wed10'), ('Dan', 'Thu11'),
('Dan', 'Fri12'), ('Dan', 'Sat13'), ('Dan', 'Sun14'), ('Ed', 'Mon1'),
('Ed', 'Tue2'), ('Ed', 'Wed3'), ('Ed', 'Thu4'), ('Ed', 'Fri5'),
('Ed', 'Sat6'), ('Ed', 'Mon8'), ('Ed', 'Tue9'), ('Ed', 'Thu11'),
('Ed', 'Sat13'), ('Ed', 'Sun14'), ('Fred', 'Mon1'), ('Fred', 'Tue2'),
('Fred', 'Wed3'), ('Fred', 'Sat6'), ('Fred', 'Mon8'), ('Fred', 'Tue9'),
('Fred', 'Fri12'), ('Fred', 'Sat13'), ('Fred', 'Sun14'),
('Gu', 'Mon1'), ('Gu', 'Tue2'), ('Gu', 'Wed3'), ('Gu', 'Fri5'),
('Gu', 'Sat6'), ('Gu', 'Sun7'), ('Gu', 'Mon8'), ('Gu', 'Tue9'),
('Gu', 'Wed10'), ('Gu', 'Thu11'), ('Gu', 'Fri12'), ('Gu', 'Sat13'),
('Gu', 'Sun14')
])
# Start environment, get model in this environment
with gp.Model("assignment", env=env) as m:
# Assignment variables: x[w,s] == 1 if worker w is assigned to shift s.
# Since an assignment model always produces integer solutions, we use
# continuous variables and solve as an LP.
x = m.addVars(availability, ub=1, name="x")
# Set tags encoding the assignments for later retrieval of the schedule.
# Each tag is a JSON string of the format
# {
# "Worker": "<Name of the worker>",
# "Shift": "String representation of the shift"
# }
#
for k, v in x.items():
name, timeslot = k
d = {"Worker": name, "Shift": shiftname[timeslot]}
v.VTag = json.dumps(d)
# The objective is to minimize the total pay costs
m.setObjective(gp.quicksum(pay[w] * x[w, s] for w, s in availability),
GRB.MINIMIZE)
# Constraints: assign exactly shiftRequirements[s] workers to each shift
reqCts = m.addConstrs(
(x.sum('*', s) == shiftRequirements[s] for s in shifts), "_")
# Submit this model for batch optimization to the cluster manager
# and return its batch ID for later querying the solution
batchID = m.optimizeBatch()
return batchID
Const_coeff = np.zeros((long_l, long_l + 1))
for i in range(0, long_l):
for j in range(0, (long_l - i)):
if (j != 0):
Const_coeff[i, long_l - j] = n - j * p + 5 - 5 * i
else:
Const_coeff[i, long_l - j] = mu - w
rest_coeff = np.ones((long_l, 2))
for i in range(0, long_l):
rest_coeff[i, 1] = i * p
t = 100000
y = m.addVars(binvar, name="y", obj=t, vtype=GRB.BINARY)
m.setObjective(t, GRB.MINIMIZE)
# m.addConstrs(gp.quicksum(distr[i]*mu_coeff[i] for i in range(len(mu_coeff)))==mu)
# m.addConstrs(gp.quicksum(distr[i]*so_coeff[i] for i in range(len(so_coeff)))==sm)
m.addConstrs(
gp.quicksum(Const_coeff[i][j] * distr[j]
for j in range(len(distr))) + pary * y[i] <= mu
for i in range(len(y)))
m.addConstr(gp.quicksum(y[i] for i in range(len(y))) == 1)
m.addConstrs(gp.quicksum(t - 5 * i * y[i] >= 0 for i in range(len(y))))
m.optimize()
for v in m.getVars():
print("%s %s %8.2f %s %8.2f %s %8.2f %s %8.2f" %
(v.Varname, "=", v.X, ", reduced cost = ", abs(
v.RC), ", from coeff = ", v.SAObjLow, "to coeff = ", v.SAObjUp))
print(" ")
def populate_dual_subproblem(data):
"""
Function that populates the Benders Dual Subproblem, as suggested by the
paper "Minimal Infeasible Subsystems and Bender's cuts" by Fischetti,
Salvagnin and Zanette.
:param data: Problem data structure
:param upper_cost: Link setup decisions fixed in the master
:param flow_cost: This is the cost of the continuous variables of the
master problem, as explained in the paper
:return: Numpy array of Gurobi model objects
"""
# Gurobi model objects
subproblems = np.empty(
shape=(data.periods, data.commodities), dtype=object)
# Construct model for period/commodity 0.
# Then, copy this and change the coefficients
subproblem = Model('subproblem_(0,0)')
# Ranges we are going to need
arcs, periods, commodities, nodes = xrange(data.arcs.size), xrange(
data.periods), xrange(data.commodities), xrange(data.nodes)
# Other data
demand, var_cost = data.demand, data.variable_cost
# Origins and destinations of commodities
origins, destinations = data.origins, data.destinations
# We use arrays to store variable indexes and variable objects. Why use
# both? Gurobi wont let us get the values of individual variables
# within a callback.. We just get the values of a large array of
# variables, in the order they were initially defined. To separate them
# in variable categories, we will have to use index arrays
flow_vars = np.empty_like(arcs, dtype=object)
# Populate all variables in one loop, keep track of their indexes
# Data for period = 0, com = 0
for arc in arcs:
flow_vars[arc] = subproblem.addVar(
obj=demand[0, 0]*var_cost[arc], lb=0., ub=1., name='flow_a{}'.format(arc))
subproblem.update()
# Add constraints
for node in nodes:
out_arcs = get_2d_index(data.arcs, data.nodes)[0] == node + 1
in_arcs = get_2d_index(data.arcs, data.nodes)[1] == node + 1
lhs = quicksum(flow_vars[out_arcs]) - quicksum(flow_vars[in_arcs])
subproblem.addConstr(lhs == 0., name='flow_bal{}'.format(node))
subproblem.update()
# Store variables
subproblem._all_variables = flow_vars.tolist()
# Set parameters
subproblem.setParam('OutputFlag', 0)
subproblem.modelSense = GRB.MINIMIZE
subproblem.params.threads = 2
subproblem.params.LogFile = ""
subproblem.update()
subproblems[0, 0] = subproblem
for period, com in product(periods, commodities):
if (period, com) != (0, 0):
model = subproblem.copy()
model.ModelName = 'subproblem_({},{})'.format(period, com)
flow_cost = data.demand[period, com] * var_cost
model.setObjective(LinExpr(flow_cost.tolist(), model.getVars()))
model.setAttr('rhs', model.getConstrs(), [0.0] * data.nodes)
model._all_variables = model.getVars()
model.update()
subproblems[period, com] = model
return subproblems
def model2(FileName):
def Read_booking(FileName):
Booking_df = pd.read_csv(FileName)
L = [] # 積み地の集合
D = [] # 揚げ地の集合
T = [] # 港の集合
lport_name = list(Booking_df["PORT_L"].unique())
dport_name = list(Booking_df["PORT_D"].unique())
lport_name = lport_name[:-1]
dport_name = dport_name[:-1]
for t in range(len(lport_name)):
L.append(t)
for t in range(len(dport_name)):
D.append(t + max(L) + 1)
T = L + D
Check_port = []
key1 = Booking_df.columns.get_loc('CPORT')
for j in range(len(Booking_df)):
count = 0
if str(Booking_df.iloc[j, key1]) != 'nan':
for p in lport_name:
if p == str(Booking_df.iloc[j, key1]):
Check_port.append(count)
else:
count = count + 1
# 港番号のエンコード
for k in range(len(lport_name)):
Booking_df = Booking_df.replace(lport_name[k], L[k])
for k in range(len(dport_name)):
Booking_df = Booking_df.replace(dport_name[k], D[k])
Booking = Booking_df.iloc[:, 0:Booking_df.columns.get_loc('PORT_L')]
key2 = Booking.columns.get_loc('LPORT')
key3 = Booking.columns.get_loc('DPORT')
key4 = Booking.columns.get_loc('Units')
key5 = Booking.columns.get_loc('RT')
key6 = Booking.columns.get_loc('Weight')
count1 = 0
for col1 in L:
for col2 in D:
units = rt = weight = count2 = 0
for k in range(len(Booking_df)):
if Booking_df.iloc[k, key2] == col1 and Booking_df.iloc[k, key3] == col2:
count2 += 1
units += Booking_df.iloc[k, key4]
rt += Booking_df.iloc[k, key4] * Booking_df.iloc[k, key5]
weight += Booking_df.iloc[k,
key4] * Booking_df.iloc[k, key6]
if count2 > 0:
Booking.iloc[count1, key2] = col1
Booking.iloc[count1, key3] = col2
Booking.iloc[count1, key4] = units
Booking.iloc[count1, key5] = rt
Booking.iloc[count1, key6] = int(weight / units)
count1 += 1
print(Booking)
for t in range(count1, len(Booking_df)):
Booking = Booking.drop(Booking.index[count1])
print(Booking)
Booking["Index"] = 0
for k in range(len(Booking)):
Booking.iloc[k, Booking.columns.get_loc('Index')] = k
A = list(Booking["RT"])
J = list(Booking["Index"])
U = list(Booking["Units"])
G = list(Booking["Weight"])
Port = Booking.iloc[:, key2:key3+1]
# 注文をサイズ毎に分類
J_small = []
J_large = []
for j in J:
if A[j] <= 100:
J_small.append(j)
if 100 < A[j]:
J_large.append(j)
return T, L, D, J, U, A, G, J_small, J_large, Port, Check_port, Booking
# 前処理
# ==============================================================================================
# ファイルロード
BookingFile = FileName
HoldFile = "data/hold.csv"
MainLampFile = "data/mainlamp.csv"
BackMainLampFile = "data/back_mainlamp.csv"
AfrMainLampFile = "data/afr_mainlamp.csv"
StressFile = "data/stress_mainlamp.csv"
Gang2File = "data/gangnum_2.csv"
Gang3File = "data/gangnum_3.csv"
# 注文情報の読み込み
# T, L, D, J, U, A, G, J_small,J_medium, J_large, Port, Check_port, Booking,diveded_dic = read_booking.Read_booking(BookingFile)
T, L, D, J, U, A, G, J_small, J_large, Port, Check_port, Booking = Read_booking(BookingFile)
print(J)
# 船体情報の読み込み1
I, B, I_pair, I_next, I_same, I_lamp, I_deck, RT_benefit, delta_s, min_s, max_s, delta_h, max_h, Hold_encode, Hold= read_hold.Read_hold(HoldFile)
# 船体情報の読み込み2
Ml_Load, Ml_Back, Ml_Afr, Stress, GANG2, GANG3 = read_other.Read_other(MainLampFile, BackMainLampFile, AfrMainLampFile, StressFile, Gang2File, Gang3File, Hold_encode)
J_t_load = [] # J_t_load:港tで積む注文の集合
J_t_keep = [] # J_t_keep:港tを通過する注文の集合
J_t_dis = [] # J_t_dis:港tで降ろす注文の集合
J_lk = [] # J_lk:J_t_load + J_t_keep
J_ld = [] # J_ld:J_t_load + J_t_dis
for t in T:
J_load = []
J_keep = []
J_dis = []
lk = []
ld = []
tmp_load = list(Port.iloc[:, 0])
tmp_dis = list(Port.iloc[:, 1])
N = len(J)
k = 0
for i in L:
if k < i:
k = i
count = 0
for t_l in tmp_load:
if t == t_l:
J_load.append(count)
lk.append(count)
ld.append(count)
count = count + 1
count = 0
for t_d in tmp_dis:
if t == t_d:
J_dis.append(count)
ld.append(count)
count = count + 1
for t_k in range(N):
if t > tmp_load[t_k] and t < tmp_dis[t_k]:
J_keep.append(J[t_k])
lk.append(t_k)
J_t_load.append(J_load)
J_t_keep.append(J_keep)
J_t_dis.append(J_dis)
J_lk.append(lk)
J_ld.append(ld)
# モデリング1(定数・変数の設定)
# ==============================================================================================
# Gurobiパラメータ設定
GAP_SP = gp.Model()
GAP_SP.setParam("TimeLimit", 300)
GAP_SP.setParam("MIPFocus", 1)
GAP_SP.setParam("LPMethod", 1)
GAP_SP.setParam("MIPGap",0.01)
GAP_SP.printStats()
# ハイパーパラメータ設定
# 各目的関数の重み
w1 = 1
w2 = 1
w3 = 1
w4 = 1
w5 = 0
# 最小分割RT
v_min = 10
# 目的関数1のペナルティ
penal1_z = 10
# 目的関数2のペナルティ
penal2_load = 1
penal2_dis = 10
# 目的関数3のペナルティ
penal3_m = 10
# 目的関数4のチェックポイント
check_point = Check_port
# 目的関数5のペナルティ
penal5_k = 1000
# 最適化変数
# V_ij:注文jをホールドiにkRT割り当てる
V_ij = {}
for i in I:
for j in J:
V_ij[i, j] = GAP_SP.addVar(
lb=0, ub=A[j], vtype=gp.GRB.CONTINUOUS, name=f"V_ij({i},{j})")
# 目的関数1
# X_ij:注文jをホールドiに割り当てるなら1、そうでなければ0
X_ij = GAP_SP.addVars(I, J, vtype=gp.GRB.BINARY)
# Y_keep_it:港tにおいてホールドiを通過する注文があるなら1、そうでなければ0
Y_keep_it = GAP_SP.addVars(I, T, vtype=gp.GRB.BINARY)
# Y_it1t2:ホールドiにおいてt1で積んでt2で降ろす注文があるなら1、そうでなければ0
Y_it1t2 = GAP_SP.addVars(I, T, T, vtype=gp.GRB.BINARY)
# Z_it1t2:ホールドiにおいてt2を通過する注文があるとき異なる乗せ港t1の数分のペナルティ
Z_it1t2 = GAP_SP.addVars(I, L, D, vtype=gp.GRB.BINARY)
OBJ1 = gp.quicksum(
w1 * penal1_z * Z_it1t2[i, t1, t2] for i in I for t1 in L for t2 in D)
# 目的関数2
# Y_load_i1i2t:港tにおいてホールドペア(i1,i2)で積む注文があるなら1
Y_load_i1i2t = GAP_SP.addVars(I, I, L, vtype=gp.GRB.BINARY)
# Y_keep_i1i2t:港tにおいてホールドペア(i1,i2)を通過する注文があるなら1
Y_keep_i1i2t = GAP_SP.addVars(I, I, T, vtype=gp.GRB.BINARY)
# Y_dis_i1i2t:港tにおいてホールドペア(i1,i2)で揚げる注文があるなら1
Y_dis_i1i2t = GAP_SP.addVars(I, I, D, vtype=gp.GRB.BINARY)
# ホールドペア(i1,i2)においてtで注文を積む際に既にtで積んだ注文があるときのペナルティ
Z1_i1i2t = GAP_SP.addVars(I, I, L, vtype=gp.GRB.BINARY)
# ホールドペア(i1,i2)においてtで注文を揚げる際にtを通過する注文があるときのペナルティ
Z2_i1i2t = GAP_SP.addVars(I, I, D, vtype=gp.GRB.BINARY)
OBJ2_1 = gp.quicksum(
penal2_load * Z1_i1i2t[i1, i2, t] for i1 in I for i2 in I for t in L)
OBJ2_2 = gp.quicksum(
penal2_dis * Z2_i1i2t[i1, i2, t] for i1 in I for i2 in I for t in D)
OBJ2 = w2 * (OBJ2_1 + OBJ2_2)
# 目的関数3
# M_it:港tにおいてホールドiが作業効率充填率を超えたら1
M_it = GAP_SP.addVars(I, T, vtype=gp.GRB.BINARY)
# M_ijt:港tにおいてホールドiに自動車を積むまでに作業効率充填率を上回ったホールドに自動車を通すペナルティ
M_ijt = GAP_SP.addVars(I, J, T, lb=0, vtype=gp.GRB.CONTINUOUS)
OBJ3 = gp.quicksum(w3 * penal3_m * M_ijt[i, j, t]
for i in I for j in J for t in T)
# 目的関数4
# N_jt:チェックポイントにおけるホールドiの残容量
N_it = GAP_SP.addVars(I, check_point, lb=0, vtype=gp.GRB.CONTINUOUS)
OBJ4 = gp.quicksum(w4 * N_it[i, t] * RT_benefit[i]
for i in I for t in check_point)
# 目的関数5
# K1_it:lampで繋がっている次ホールドが充填率75%を上回ったら1、そうでなければ0
K1_it = GAP_SP.addVars(I_lamp, L, vtype=gp.GRB.BINARY)
# K2_it:ホールドiが1RT以上のスペースがあったら1、そうでなければ0
K2_it = GAP_SP.addVars(I, L, lb=0, vtype=gp.GRB.BINARY)
# K3_it:目的関数5のペナルティ
K3_it = GAP_SP.addVars(I_lamp, L, lb=0, vtype=gp.GRB.CONTINUOUS)
OBJ5 = gp.quicksum(w5 * penal5_k * K3_it[i, t] for i in I_lamp for t in L)
# 目的関数の設計
OBJ = OBJ1 + OBJ2 + OBJ3 - OBJ4 + OBJ5
GAP_SP.setObjective(OBJ, gp.GRB.MINIMIZE)
# モデリング2(制約)
# ==============================================================================================
# 基本制約
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 割当てた注文がコンパートメント毎にリソースを超えない
GAP_SP.addConstrs(gp.quicksum(V_ij[i, j] for j in J) <= B[i] for i in I)
# 全注文内の自動車の台数を全て割り当てる
GAP_SP.addConstrs(gp.quicksum(V_ij[i, j] for i in I) == A[j] for j in J)
# VとXの制約
GAP_SP.addConstrs(V_ij[i, j] / A[j] <= X_ij[i, j] for i in I for j in J)
# 目的関数の制約
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 目的関数1の制約
for t in T:
GAP_SP.addConstrs(X_ij[i, j] <= Y_keep_it[i, t]
for i in I for j in J_t_keep[t])
for t in D:
t2 = t
for t1 in L:
J_sub = [] # J_sub:t1で積んでt2で揚げる注文の集合
for j in J_t_dis[t2]:
if j in J_t_load[t1]:
J_sub.append(j)
GAP_SP.addConstrs(X_ij[i, j] <= Y_it1t2[i, t1, t2]
for i in I for j in J_sub)
GAP_SP.addConstrs(Z_it1t2[i, t1, t2] <=
Y_it1t2[i, t1, t2] for i in I)
GAP_SP.addConstrs(Z_it1t2[i, t1, t2] <=
Y_keep_it[i, t2] for i in I)
GAP_SP.addConstrs(
Z_it1t2[i, t1, t2] >= Y_it1t2[i, t1, t2] + Y_keep_it[i, t2] - 1 for i in I)
# 目的関数2の制約
for t in T:
for i1, i2 in I_pair:
GAP_SP.addConstrs(X_ij[i1, j] + X_ij[i2, j] <=
2 * Y_keep_i1i2t[i1, i2, t] for j in J_t_keep[t])
if t in L:
GAP_SP.addConstrs(
X_ij[i1, j] + X_ij[i2, j] <= 2 * Y_load_i1i2t[i1, i2, t] for j in J_t_load[t])
GAP_SP.addConstr(Z1_i1i2t[i1, i2, t]
<= Y_load_i1i2t[i1, i2, t])
GAP_SP.addConstr(Z1_i1i2t[i1, i2, t]
<= Y_keep_i1i2t[i1, i2, t])
GAP_SP.addConstr(
Z1_i1i2t[i1, i2, t] >= Y_load_i1i2t[i1, i2, t] + Y_keep_i1i2t[i1, i2, t] - 1)
if t in D:
GAP_SP.addConstrs(
X_ij[i1, j] + X_ij[i2, j] <= 2 * Y_dis_i1i2t[i1, i2, t] for j in J_t_dis[t])
GAP_SP.addConstr(Z2_i1i2t[i1, i2, t] <= Y_dis_i1i2t[i1, i2, t])
GAP_SP.addConstr(Z2_i1i2t[i1, i2, t]
<= Y_keep_i1i2t[i1, i2, t])
GAP_SP.addConstr(
Z2_i1i2t[i1, i2, t] >= Y_dis_i1i2t[i1, i2, t] + Y_keep_i1i2t[i1, i2, t] - 1)
# 目的関数3の制約
for t in T:
GAP_SP.addConstrs(M_it[i, t] >= - Stress[i] + (gp.quicksum(V_ij[i, j]
for j in J_t_keep[t]) / B[i]) for i in I)
for i1 in I:
I_primetmp = []
for k in Ml_Load.iloc[i1, :]:
I_primetmp.append(k)
I_prime = [x for x in I_primetmp if str(x) != 'nan']
I_prime.pop(0)
GAP_SP.addConstrs(M_ijt[i1, j, t] >= gp.quicksum(
M_it[i2, t] for i2 in I_prime) for j in J_ld[t])
# 目的関数4の制約
for t in check_point:
GAP_SP.addConstrs(N_it[i, t] <= B[i] - gp.quicksum(V_ij[i, j]
for j in J_lk[t]) for i in I)
# 目的関数5の制約
GAP_SP.addConstrs(K1_it[i, t] >= - 0.75 + gp.quicksum(V_ij[i, j]
for j in J_lk[t]) / B[i] for i in I_lamp for t in L)
GAP_SP.addConstrs(K2_it[i, t] >= 1 - (gp.quicksum(V_ij[i, j]
for j in J_lk[t]) + 1) / B[i] for i in I for t in L)
for i in range(len(Ml_Back)):
i1 = Ml_Back.iloc[i, 0]
I_backtmp = []
for k in Ml_Back.iloc[i, :]:
I_backtmp.append(k)
I_back_i1 = [x for x in I_backtmp if str(x) != 'nan']
I_back_i1.pop(0)
GAP_SP.addConstrs(K3_it[i1, t] >= len(
I) * (K1_it[i1, t] - 1) + gp.quicksum(K2_it[i2, t] for i2 in I_back_i1) for t in L)
# 特殊制約1(注文の分割制約)
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# J_smallを分割しない
GAP_SP.addConstrs(gp.quicksum(X_ij[i, j] for i in I) == 1 for j in J_small)
# 10RT以下に分割しない
GAP_SP.addConstrs(V_ij[i, j] + v_min * (1 - X_ij[i, j])
>= v_min for i in I for j in J)
# 特殊制約2(移動経路制約)
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 港を通過する荷物が移動の邪魔をしない
for t in T:
for i1 in I:
I_primetmp = []
Frtmp = []
for k in Ml_Load.iloc[i1, :]:
I_primetmp.append(k)
for k in Ml_Afr.iloc[i1, :]:
Frtmp.append(k)
I_prime = [int(x) for x in I_primetmp if str(x) != 'nan']
Fr = [x for x in Frtmp if str(x) != 'nan']
I_prime.pop(0)
Fr.pop(0)
N_prime = len(I_prime)
for k in range(N_prime):
i2 = int(I_prime[k])
GAP_SP.addConstrs(gp.quicksum(
V_ij[i2, j1] for j1 in J_t_keep[t]) / B[i2] <= 1 + Fr[k] - X_ij[i1, j2] for j2 in J_ld[t])
# 特殊制約3(船体重心の制約)
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# 船の上下前後の配置バランスが閾値を超えない
for t in T:
# 荷物を全て降ろしたとき
GAP_SP.addConstr(gp.quicksum(
delta_h[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_t_keep[t] for i in I) <= max_h)
GAP_SP.addConstr(gp.quicksum(
delta_s[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_t_keep[t] for i in I) <= max_s)
GAP_SP.addConstr(gp.quicksum(
delta_s[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_t_keep[t] for i in I) >= min_s)
# 荷物を全て載せたとき
GAP_SP.addConstr(gp.quicksum(
delta_h[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_lk[t] for i in I) <= max_h)
GAP_SP.addConstr(gp.quicksum(
delta_s[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_lk[t] for i in I) <= max_s)
GAP_SP.addConstr(gp.quicksum(
delta_s[i] * U[j] * G[j] * V_ij[i, j] / A[j] for j in J_lk[t] for i in I) >= min_s)
# 特殊制約4(ギャングの制約)
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"""
count = 0
for t in L:
gang = gang_num[count]
if gang == 2:
gang_low = GANG2[0]
gang_high = GANG2[1]
#ギャングの担当領域毎に均等に注文を分ける
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) <= 100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) >= -100)
if gang == 3:
gang_low = GANG3[0]
gang_mid = GANG3[1]
gang_high = GANG3[2]
#ギャングの担当領域毎に均等に注文を分ける
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_mid for j in J_t_load[t]) <= 100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_mid for j in J_t_load[t]) >= -100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) <= 100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_low for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) >= -100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_mid for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) <= 100)
GAP_SP.addConstr(gp.quicksum(V_ij[i1,j] for i1 in gang_mid for j in J_t_load[t]) - gp.quicksum(V_ij[i2,j] for i2 in gang_high for j in J_t_load[t]) >= -100)
"""
# 最適化計算
# ==============================================================================================
print("\n========================= Solve Assignment Problem =========================")
GAP_SP.optimize()
# 解の保存
# ==============================================================================================
# 目的関数の値
val_opt = GAP_SP.ObjVal
# OBJ1のペナルティ
penal1 = 0
for i in I:
for t1 in L:
for t2 in D:
if Z_it1t2[i, t1, t2].X > 0:
penal1 = penal1 + Z_it1t2[i, t1, t2].X
# OBJ2のペナルティ
penal2_1 = penal2_2 = 0
for i1 in I:
for i2 in I:
for t in L:
if Z1_i1i2t[i1, i2, t].X > 0:
penal2_1 = penal2_1 + 1
# print(f"ホールド{i1},{i2}で積み地ペナルティ")
for t in D:
if Z2_i1i2t[i1, i2, t].X > 0:
penal2_2 = penal2_2 + 1
# print(f"ホールド{i1},{i2}で揚げ地ペナルティ")
# OBJ3のペナルティ
penal3 = 0
for i in I:
for j in J:
for t in T:
if M_ijt[i, j, t].X > 0:
penal3 = penal3 + M_ijt[i, j, t].X
# OBJ4のペナルティ
benefit4 = 0
for t in check_point:
for i in I:
if N_it[i, t].X > 0:
benefit4 = benefit4 + N_it[i, t].X
# OBJ5のペナルティ
penal5 = 0
for t in L:
for i in I_lamp:
if K3_it[i, t].X > 0:
penal5 = penal5 + K3_it[i, t].X
# 解の書き込み
answer = []
assign = []
for i in I:
for j in J:
if V_ij[i, j].X > 0:
# assign_data[ホールド番号,積載RT,積み地,揚げ地]
answer.append([i, j])
assign.append([0, V_ij[i, j].X, "L", "D"])
# 残リソース
I_left_data = Hold.iloc[:, 0:2]
for k in range(len(assign)):
key = Booking.columns.get_loc('Index')
i_t = answer[k][0]
j_t = int(Booking.iloc[answer[k][1], key])
# hold_ID
assign[k][0] = i_t
# L_port
assign[k][2] = Booking.iloc[j_t, Booking.columns.get_loc('LPORT')]
# D_port
assign[k][3] = Booking.iloc[j_t, Booking.columns.get_loc('DPORT')]
csv = []
for k in range(len(assign)):
csv.append(assign[k])
c_list = []
c_list.append("Hold_ID")
c_list.append("Load_RT")
c_list.append("LPORT")
c_list.append("DPORT")
assign_data = pd.DataFrame(csv, columns=c_list)
return assign_data
I4_ALL.append(I_ALL[i][3][k])
""" w = np.zeros((N,4,T))
w[:,1,:] = 1
for t in range(int(T/60)):
w[:,1,t*60:t*60+30]=1
w[:,1,t*60+30:(t+1)*60]=0 """
print(T)
m = gb.Model()
y = m.addVars(len(C_ALL), T, lb=0, vtype=gb.GRB.CONTINUOUS)
n = m.addVars(len(C_ALL), T + 1, lb=0, vtype=gb.GRB.CONTINUOUS)
w = m.addVars(N, 4, T, lb=0, ub=1, vtype=gb.GRB.BINARY) #binary variable
# w = m.addVars(N, 4, T, lb=0, ub=1, vtype=gb.GRB.CONTINUOUS)
new_w = m.addVars(N, T, lb=0, ub=1, vtype=gb.GRB.CONTINUOUS)
u = m.addVars(N, T - 1, lb=0, ub=1, vtype=gb.GRB.CONTINUOUS)
m.setObjective(
gb.quicksum(gb.quicksum(t * y[c, t] for c in D_ALL)
for t in range(T)) + alpha * gb.quicksum(
gb.quicksum((T - t) * y[c, t]
for c in list(set(C_ALL) - set(D_ALL)))
for t in range(T)), gb.GRB.MINIMIZE)
# m.setObjective(alpha * gb.quicksum(gb.quicksum((T-t) * y[c,t] for c in list(set(C_ALL)-set(D_ALL))) for t in range(T)), gb.GRB.MINIMIZE)
m.addConstrs(y[c, t] - n[c, t] <= 0 for c in C_ALL for t in range(T))
m.addConstrs(y[c, t] <= Q[c] for c in list(
set(C_ALL) - set(I1_ALL) - set(I2_ALL) - set(I3_ALL) - set(I4_ALL) -
set(D_ALL)) for t in range(T))
m.addConstrs(y[c, t] <= Q[c + 1] / beta[c, 0] for c in V_ALL for t in range(T))
m.addConstrs(y[c, t] <= Q[c + 2] / beta[c, 1] for c in V_ALL for t in range(T))
m.addConstrs(y[c, t] <= Q[c + 3] / beta[c, 2] for c in V_ALL for t in range(T))
m.addConstrs(y[c, t] - w[i, j, t] * Q[c] <= 0 for i in range(N)
for j in range(4) for c in I_ALL[i][j] for t in range(T))
m.addConstrs(y[c, t] + W * n[proc_all[c][0], t] <= W * jam[c]
def Gurobi_pDispersion(dij, p_facilities, total_facilities):
t1 = time.time()
facility_range = range(total_facilities) # range of total facilities
M = np.amax(dij) # Max Value in dij
mPDP = gbp.Model(" -- p-Dispersion -- ") # Instantiate a model
gbp.setParam('MIPFocus', 2) # Set MIP Focus to 2 for optimality
# Add Decision Variables
facility_variable = []
for destination in facility_range:
facility_variable.append(
mPDP.addVar(vtype=gbp.GRB.BINARY,
lb=0,
ub=1,
name='y' + str(destination + 1)))
# Add Maximized Minimum Variable
D = mPDP.addVar(vtype=gbp.GRB.CONTINUOUS,
lb=0,
ub=gbp.GRB.INFINITY,
name='D')
# Update Model Variables
mPDP.update()
# Set Objective Function
mPDP.setObjective(D, gbp.GRB.MAXIMIZE)
# Add Facility Constraint
mPDP.addConstr(gbp.quicksum(facility_variable[destination] \
for destination in facility_range) \
== p_facilities)
# Add Inter-Facility Distance Constraints ==> n(n-1)/2
for origin in facility_range:
for destination in facility_range:
if destination > origin:
mPDP.addConstr(
dij[origin][destination] \
+ M * 2 \
- M * facility_variable[origin] \
- M * facility_variable[destination] \
>= D)
else:
pass
# Optimize and Print Results
mPDP.optimize()
mPDP.write('path.lp')
t2 = round(round(time.time() - t1, 3) / 60, 5)
print '\n**********************************************************************'
selected_facilities = []
for v in mPDP.getVars():
if 'D' in v.VarName:
pass
elif v.x > 0:
var = '%s' % v.VarName
selected_facilities.append(var)
print ' | ', var
print ' | Selected Facility Locations ------------- ^^^^ '
print ' | Candidate Facilities [p] ---------------- ', len(
selected_facilities)
print ' | Largest Value in dij (M) ---------------- ', M
print ' | Objective Value (D) --------------------- ', mPDP.objVal
print ' | Matrix Dimensions ----------------------- ', dij.shape
print ' | Real Time to Solve (minutes)------------- ', t2
print '**********************************************************************'
print ' -- The p-Dispersion Problem Gurobi -- '
print '\n -- James Gaboardi, 2016 -- '
def opt_idle(av_fleet, customers, t):
idle_avs = list(j_veh for j_veh in av_fleet
if j_veh.status in ("idle", "relocating"))
len_idle_avs = len(idle_avs)
unassign_cust = list(i_cust for i_cust in customers
if i_cust.status == "unassigned")
len_custs = len(unassign_cust)
dist_assgn = [[0 for jj in range(len_idle_avs)] for ii in range(len_custs)]
x = [[0 for jj in range(len_idle_avs)] for ii in range(len_custs)]
count_pass = -1
for i_pass in unassign_cust:
count_pass += 1
count_veh = -1
cur_wait = t - i_pass.request_time
elapsed_wait_penalty = cur_wait * Set.gamma
for j_veh in idle_avs:
count_veh += 1
av_curb_wait = j_veh.curb_time_remain * Set.veh_speed
dist_assgn[count_pass][count_veh] = Distance.dist_manhat_pick(i_pass, j_veh) - \
elapsed_wait_penalty + av_curb_wait
t1 = time.time()
# Model
models = gurobipy.Model("idleOnly_minDist")
models.setParam('OutputFlag', False)
# Decision Variables
for i in range(len_custs):
for j in range(len_idle_avs):
x[i][j] = models.addVar(vtype=gurobipy.GRB.CONTINUOUS,
obj=dist_assgn[i][j],
name='x_%s_%s' % (i, j))
models.update()
# constraints
# if the number of unassigned travelers is less than the number of idle vehicles
# then make sure all the unassigned travelers are assigned a vehicle
if len_custs <= len_idle_avs:
for ii in range(len_custs):
models.addConstr(
gurobipy.quicksum(x[ii][j] for j in range(len_idle_avs)) == 1)
for jj in range(len_idle_avs):
models.addConstr(
gurobipy.quicksum(x[i][jj] for i in range(len_custs)) <= 1)
# else if the number of unassigned travelers is greater than the number of idle vehicles
# then make sure all the idle vehicles are assigned to an unassigned traveler
else:
for ii in range(len_custs):
models.addConstr(
gurobipy.quicksum(x[ii][j] for j in range(len_idle_avs)) <= 1)
for jj in range(len_idle_avs):
models.addConstr(
gurobipy.quicksum(x[i][jj] for i in range(len_custs)) == 1)
models.optimize()
if models.status == gurobipy.GRB.Status.OPTIMAL:
for m_pass in range(len_custs):
for n_veh in range(len_idle_avs):
if x[m_pass][n_veh].X == 1:
temp_veh_status = "base_assign"
win_cust = unassign_cust[m_pass]
win_av = idle_avs[n_veh]
Vehicle.update_vehicle(t, win_cust, win_av,
Regions.SubArea, temp_veh_status)
Person.update_person(t, win_cust, win_av)
break
else:
sys.exit("No Optimal Solution - idleOnly_minDist")
# print("Vehicles= ", len_veh, " Passengers= ", len_pass, " time=", time.time() - t1)
return
def Build(C_pop, T_pop, matches, weights):
#start the setup timer
setup_time = FNC.timerStart()
#set weights for covariates to their min
weight_base, mean_T_pop, dist = FNC.Pop_Calcuations(C_pop, T_pop)
weights = weights * weight_base
#start creating model elements
Ctrl = list(range(len(C_pop)))
C_pop.index = Ctrl
Treat = list(range(len(T_pop)))
T_pop.index = Treat
T_n = len(T_pop)
dist = pd.DataFrame(dist, index=Treat, columns=Ctrl)
Covar = list(T_pop)
weights = pd.Series(weights, index=Covar)
#define the model
m = Model('match')
m.Params.LogFile = ""
m.Params.LogToConsole = 0
#create variables
FNC.printMessage("Creating Gurobi Variables")
assign = [[
m.addVar(vtype=GRB.BINARY, name="%i, %i" % (t, c)) for c in Ctrl
] for t in Treat]
assign = pd.DataFrame(assign, index=Treat, columns=Ctrl)
z = m.addVars(Covar, vtype=GRB.CONTINUOUS, name="z")
m.update()
#objective fuction
FNC.printMessage("Creating Gurobi Objective Function")
m.setObjective((dist * assign).sum().sum() + quicksum(weights[i] * z[i]
for i in Covar),
sense=GRB.MINIMIZE)
#define the constraints
FNC.printMessage("Creating Gurobi Constraints")
c1_t = FNC.timerStart()
constr_1(m, assign, matches, Treat)
FNC.printMessage("Constraint 1 done")
c1_t = FNC.timerStop(c1_t, 3)
c2_t = FNC.timerStart()
#####################################################
threaded_constr_2(m, assign, Ctrl, 30)
#####################################################
FNC.printMessage("Constraint 2 done")
c2_t = FNC.timerStop(c2_t, 3)
c3_t = FNC.timerStart()
constr_3(m, assign, z, C_pop, matches, T_n, mean_T_pop, Treat, Covar)
FNC.printMessage("Constraint 3 done")
c3_t = FNC.timerStop(c3_t, 3)
c4_t = FNC.timerStart()
constr_4(m, assign, z, C_pop, matches, T_n, mean_T_pop, Treat, Covar)
FNC.printMessage("Constraint 4 done")
c4_t = FNC.timerStop(c4_t, 3)
m.update()
setup_time = FNC.timerStop(setup_time, 3)
Timings = [setup_time, c1_t, c2_t, c3_t, c4_t]
Timings = pd.Series(Timings, index=["Setup Time", "C1", "C2", "C3", "C4"])
return m, assign, Timings
def run(lin, relaxcapacity, verbose=False):
"""
Realiza Optimización de Despacho con Seguridad para
horizonte de tiempo sin considerar relajo de restricciones.
"""
M = 10000
NGen = 3
NLin = 3
NBar = 3
NHrs = 24
NEle = NGen + NLin
NFal = NGen + NLin
NSce = NFal + 1
numLin = range(NLin)
numBar = range(NBar)
numGen = range(NGen)
numFal = range(NFal)
numEle = range(NEle)
numSce = range(NSce)
numHrs = range(NHrs)
numScF = range(1, NSce)
GENDATA, DEMDATA, LINDATA, STODATA = ld.loadsystemdata()
#Generator data
genBar = GENDATA[:, 1]
CV = GENDATA[:, 2]
CRUP = GENDATA[:, 3]
CRDOWN = GENDATA[:, 4]
CENC = GENDATA[:, 5]
CAPG = GENDATA[:, 6]
RUPMAX = GENDATA[:, 7]
RDOWNMAX = GENDATA[:, 8]
PMAXGEN = GENDATA[:, 9]
PMINGEN = GENDATA[:, 10]
QMAXGEN = GENDATA[:, 11]
QMINGEN = GENDATA[:, 12]
#Demand data
DPMAX = DEMDATA[:, 1]
DQMAX = DEMDATA[:, 2]
#Line data
FMAX = LINDATA[:, 1]
XL = LINDATA[:, 4]
LENGTHLINE = LINDATA[:, 6]
#Demand data
DEMRES, DEMIND, DEMCOM = ld.loadcurvedata()
D = np.array([DPMAX[0] * DEMRES, DPMAX[1] * DEMIND, DPMAX[2] * DEMCOM])
#Scenarios data
A, B, PROB = ld.loadscendata()
# =============================================================================
# SIPS
# =============================================================================
#Storage data
CBAT = abs(STODATA[:, 2])
PMAXBAT = abs(STODATA[:, 8])
#Load Shedding
DEMLSMAX = DEMDATA[:, 4]
CLS = DEMDATA[:, 3]
# =============================================================================
# ADD MORE CAPACITY TO LINE TO RELAX CONSTRAINT
# =============================================================================
FMAXP = FMAX.copy()
if type(lin) is list:
for (l, rc) in zip(lin, relaxcapacity):
FMAXP[l] = rc * FMAXP[l]
elif type(lin) is int:
FMAXP[lin] = relaxcapacity * FMAXP[lin]
# =============================================================================
# FACTOR CRECIMIENTO GENERACIÓN, DEMANDA Y LINEAS
# =============================================================================
#crecimiento demanda anual
fdem = 1.05
#crecimiento capacidad en MW
fcap = 100
#crecimiento generacion
fgen = 1.07
# =============================================================================
# COSTOS DE INVERSION POR MW
# =============================================================================
CINVLIN = 740 * LENGTHLINE[0]
CINVBAT = 400000
print(CINVLIN * fcap)
# =============================================================================
# HORIZONTE PLANIFICACION
# =============================================================================
NYears = 20
numYears = range(NYears)
DEMY = np.ndarray((NYears, NBar, NHrs))
RUPMAXY = np.ndarray((NYears, NGen))
RDOWNMAXY = np.ndarray((NYears, NGen))
PMAXGENY = np.ndarray((NYears, NGen))
PMINGENY = np.ndarray((NYears, NGen))
for y in numYears:
DEMY[y, :] = (fdem**y) * D
PMAXGENY[y, :] = (fgen**y) * PMAXGEN[:]
RUPMAXY[y, :] = (fgen**y) * RUPMAX[:]
RDOWNMAXY[y, :] = (fgen**y) * RDOWNMAX[:]
PMINGENY[y, :] = PMINGEN[:]
# =============================================================================
# DEFINE VARIABLES
# =============================================================================
#Define problem
model = gp.Model("Model")
if not verbose:
model.setParam('OutputFlag', 0)
else:
model.setParam('OutputFlag', 1)
#Variables
p = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.CONTINUOUS)
x = model.addVars(NYears, NGen, NHrs, vtype=GRB.BINARY)
e = model.addVars(NYears, NGen, NHrs, vtype=GRB.BINARY)
a = model.addVars(NYears, NGen, NHrs, vtype=GRB.BINARY)
f = model.addVars(NYears,
NLin,
NSce,
NHrs,
vtype=GRB.CONTINUOUS,
lb=-GRB.INFINITY)
theta = model.addVars(NYears,
NBar,
NSce,
NHrs,
vtype=GRB.CONTINUOUS,
lb=-GRB.INFINITY)
rup = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.CONTINUOUS)
rdown = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.CONTINUOUS)
xrup = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.BINARY)
xrdown = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.BINARY)
#SIPS
pbat = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.CONTINUOUS)
demls = model.addVars(NYears, NGen, NSce, NHrs, vtype=GRB.CONTINUOUS)
#Investments Line
xf50 = model.addVars(NLin, vtype=GRB.BINARY)
#Battery
xbat = model.addVars(NBar, vtype=GRB.BINARY)
#Objective Function
Objective = quicksum(quicksum(quicksum(CV[g]*p[y,g,0,t] + CENC[g]*e[y,g,t] + CAPG[g]*a[y,g,t] for g in numGen) +
quicksum(PROB[s-1]*quicksum(CV[g]*p[y,g,s,t] + CRUP[g]*rup[y,g,s,t] + CRDOWN[g]* rdown[y,g,s,t]
for g in numGen) for s in numScF) for t in numHrs) +
quicksum(quicksum(PROB[s-1]*quicksum(CBAT[b]*pbat[y,b,s,t] + CLS[b]*demls[y,b,s,t] for b in numBar)
for s in numScF) for t in numHrs)for y in numYears) + \
quicksum(fcap*xf50[l]*CINVLIN for l in numLin) + \
quicksum(PMAXBAT[b]*CINVBAT*xbat[b] for b in numBar)
#Contraints
#Turn on and Turn Off machines
for y in numYears:
for g in numGen:
model.addConstr(x[y, g, 0] == e[y, g, 0] - a[y, g, 0])
for t in range(2, NHrs):
for g in numGen:
model.addConstr(x[y, g, t] == x[y, g, t - 1] + e[y, g, t] -
a[y, g, t])
for t in range(3, NHrs):
for g in numGen:
model.addConstr(
x[y, g, t] >= quicksum(e[y, g, k]
for k in range(t - 3, t)))
model.addConstr(
1 - x[y, g, t] >= quicksum(a[y, g, k]
for k in range(t - 3, t)))
for g in numGen:
model.addConstr(x[y, g, 0] >= e[y, g, 0])
model.addConstr(1 - x[y, g, 0] >= a[y, g, 0])
for g in numGen:
model.addConstr(x[y, g, 1] >= quicksum(e[y, g, k]
for k in range(2)))
model.addConstr(1 - x[y, g, 1] >= quicksum(a[y, g, k]
for k in range(2)))
for g in numGen:
model.addConstr(x[y, g, 2] >= quicksum(e[y, g, k]
for k in range(3)))
model.addConstr(1 - x[y, g, 2] >= quicksum(a[y, g, k]
for k in range(3)))
#Limits of Generators
for t in numHrs:
for s in numSce:
for g in numGen:
model.addConstr(
p[y, g, 0, t] + rup[y, g, s, t] <= PMAXGENY[y, g] *
x[y, g, t])
model.addConstr(
p[y, g, 0, t] - rdown[y, g, s, t] >= PMINGENY[y, g] *
x[y, g, t])
model.addConstr(
rup[y, g, s, t] <= RUPMAXY[y, g] * xrup[y, g, s, t])
model.addConstr(rdown[y, g, s, t] <= RDOWNMAXY[y, g] *
xrdown[y, g, s, t])
model.addConstr(xrup[y, g, s, t] + xrdown[y, g, s, t] <= 1)
#Baterias
model.addConstr(pbat[y, g, s, t] <= PMAXBAT[g] * xbat[g])
#Loas shedding
model.addConstr(demls[y, g, s, t] <= DEMLSMAX[g])
#Reserves are 0 for the base case scenario
for t in numHrs:
for g in numGen:
model.addConstr(xrup[y, g, 0, t] + xrdown[y, g, 0, t] == 0)
model.addConstr(demls[y, g, 0, t] == 0)
model.addConstr(pbat[y, g, 0, t] == 0)
#Nodal Balancing por scenarios pre and post failure
for t in numHrs:
for s in numSce:
model.addConstr(
p[y, 0, s, t] + pbat[y, 0, s, t] == DEMY[y, 0, t] +
f[y, 0, s, t] + f[y, 2, s, t] - demls[y, 0, s, t])
model.addConstr(p[y, 1, s, t] + pbat[y, 1, s, t] +
f[y, 0, s, t] == DEMY[y, 1, t] +
f[y, 1, s, t] - demls[y, 1, s, t])
model.addConstr(
p[y, 2, s, t] + pbat[y, 2, s, t] + f[y, 2, s, t] +
f[y, 1, s, t] == DEMY[y, 2, t] - demls[y, 2, s, t])
#Availability of elements
for t in numHrs:
for s in numSce:
#Availability of generators
for g in numGen:
model.addConstr(p[y, g, s, t] <= A[g, s] *
(p[y, g, 0, t] + rup[y, g, s, t]))
model.addConstr(p[y, g, s, t] >= A[g, s] *
(p[y, g, 0, t] - rdown[y, g, s, t]))
#Availability of lines
for l in numLin:
model.addConstr(f[y, l, s, t] <= FMAX[l] * B[l, s] +
fcap * xf50[l] * B[l, s])
model.addConstr(f[y, l, s, t] >= -1 * FMAX[l] * B[l, s] -
fcap * xf50[l] * B[l, s])
model.addConstr(f[y, 0, s, t] <= 1 / XL[0] *
(theta[y, 0, s, t] - theta[y, 1, s, t]) + M *
(1 - B[0, s]))
model.addConstr(f[y, 0, s, t] >= 1 / XL[0] *
(theta[y, 0, s, t] - theta[y, 1, s, t]) - M *
(1 - B[0, s]))
model.addConstr(f[y, 1, s, t] <= 1 / XL[1] *
(theta[y, 1, s, t] - theta[y, 2, s, t]) + M *
(1 - B[1, s]))
model.addConstr(f[y, 1, s, t] >= 1 / XL[1] *
(theta[y, 1, s, t] - theta[y, 2, s, t]) - M *
(1 - B[1, s]))
model.addConstr(f[y, 2, s, t] <= 1 / XL[2] *
(theta[y, 0, s, t] - theta[y, 2, s, t]) + M *
(1 - B[2, s]))
model.addConstr(f[y, 2, s, t] >= 1 / XL[2] *
(theta[y, 0, s, t] - theta[y, 2, s, t]) - M *
(1 - B[2, s]))
#Capacity of lines in scenario base 0
for t in numHrs:
for l in numLin:
model.addConstr(
f[y, l, 0, t] <= FMAX[l] * B[l, 0] +
fcap * xf50[l] * B[l, 0]) #scenario 0: must respect limits
model.addConstr(
f[y, l, 0,
t] >= -1 * FMAX[l] * B[l, 0] - fcap * xf50[l] * B[l, 0]
) #scenarios of contingencies can violate limits until a treshhold
#Capacity of lines relaxed for scenarios of contingencies
for t in numHrs:
for s in numScF:
for l in numLin:
model.addConstr(f[y, l, s, t] <= FMAXP[l] * B[l, s] +
fcap * xf50[l] * B[l, s])
model.addConstr(f[y, l, s, t] >= -1 * FMAXP[l] * B[l, s] -
fcap * xf50[l] * B[l, s])
# =============================================================================
# RUN MODEL
# =============================================================================
model.setObjective(Objective, GRB.MINIMIZE)
model.optimize()
print('Costos Totales:', model.objVal)
# =============================================================================
# SAVE RESULTS
# =============================================================================
psol = np.array([[[[p[y, g, s, t].X for t in numHrs] for s in numSce]
for g in numGen] for y in numYears])
xsol = np.array([[[x[y, g, t].X for t in numHrs] for g in numGen]
for y in numYears])
esol = np.array([[[e[y, g, t].X for t in numHrs] for g in numGen]
for y in numYears])
asol = np.array([[[a[y, g, t].X for t in numHrs] for g in numGen]
for y in numYears])
fsol = np.array([[[[f[y, l, s, t].X for t in numHrs] for s in numSce]
for l in numLin] for y in numYears])
thetasol = np.array([[[[theta[y, n, s, t].X for t in numHrs]
for s in numSce] for n in numBar]
for y in numYears])
rupsol = np.array([[[[rup[y, g, s, t].X for t in numHrs] for s in numSce]
for g in numGen] for y in numYears])
rdownsol = np.array([[[[rdown[y, g, s, t].X for t in numHrs]
for s in numSce] for g in numGen]
for y in numYears])
xrupsol = np.array([[[[xrup[y, g, s, t].X for t in numHrs] for s in numSce]
for g in numGen] for y in numYears])
xrdownsol = np.array([[[[xrdown[y, g, s, t].X for t in numHrs]
for s in numSce] for g in numGen]
for y in numYears])
demlssol = np.array([[[[demls[y, g, s, t].X for t in numHrs]
for s in numSce] for g in numGen]
for y in numYears])
pbatsol = np.array([[[[pbat[y, g, s, t].X for t in numHrs] for s in numSce]
for g in numGen] for y in numYears])
xf50sol = np.array([xf50[l].X for l in numLin])
xbatsol = np.array([xbat[b].X for b in numBar])
print('Inversion Lineas', xf50sol)
print('Inversion Baterias', xbatsol)
np.save('results\\ECOPSOL_SIPS_PLANIFICACION', psol)
np.save('results\\ECOXSOL_SIPS_PLANIFICACION', xsol)
np.save('results\\ECOESOL_SIPS_PLANIFICACION', esol)
np.save('results\\ECOASOL_SIPS_PLANIFICACION', asol)
np.save('results\\ECOFSOL_SIPS_PLANIFICACION', fsol)
np.save('results\\ECOTHETASOL_SIPS_PLANIFICACION', thetasol)
np.save('results\\ECORUPSOL_SIPS_PLANIFICACION', rupsol)
np.save('results\\ECORDOWNSOL_SIPS_PLANIFICACION', rdownsol)
np.save('results\\ECOXRUPSOL_SIPS_PLANIFICACION', xrupsol)
np.save('results\\ECOXRDOWNSOL_SIPS_PLANIFICACION', xrdownsol)
np.save('results\\ECODEM_SIPS_PLANIFICACION', D)
np.save('results\\ECOPDEMLSSOL_SIPS_PLANIFICACION', demlssol)
np.save('results\\ECOPBATSOL_SIPS_PLANIFICACION', pbatsol)
return model.objVal
def opt_idle_pick(av_fleet, customers, t):
unassign_cust = list(i_cust for i_cust in customers
if i_cust.status == "unassigned")
assign_cust = list(ii_cust for ii_cust in customers
if ii_cust.status == "assigned")
temp_av_fleet = av_fleet[:]
for j_av in temp_av_fleet:
i_cust = j_av.next_pickup
if i_cust.reassigned == 1:
assign_cust.remove(i_cust)
temp_av_fleet.remove(j_av)
idle_avs = list(j_veh for j_veh in temp_av_fleet
if j_veh.status in ("idle", "relocating"))
pick_avs = list(j_veh for j_veh in temp_av_fleet
if j_veh.status == "enroute_pickup")
# just want to get avs in the right order
idle_n_pick_avs = idle_avs + pick_avs
len_idle_n_pick_av = len(idle_n_pick_avs)
no_assign_or_pick_cust = unassign_cust + assign_cust
len_no_assign_or_pick_cust = len(no_assign_or_pick_cust)
dist_assgn = [[0 for j in range(len_idle_n_pick_av)]
for i in range(len_no_assign_or_pick_cust)]
x = [[0 for j in range(len_idle_n_pick_av)]
for i in range(len_no_assign_or_pick_cust)]
prev_assign = [0 for z in range(len_no_assign_or_pick_cust)]
count_pass = -1
for i_pass in no_assign_or_pick_cust:
count_pass += 1
count_veh = -1
cur_wait = t - i_pass.request_time
elapsed_wait_penalty = cur_wait * Set.gamma
for j_veh in idle_n_pick_avs:
count_veh += 1
if i_pass.status == "unassigned":
if j_veh.status in ("idle", "relocating"):
av_curb_wait = j_veh.curb_time_remain * Set.veh_speed
dist_assgn[count_pass][count_veh] = Distance.dist_manhat_pick(i_pass, j_veh) - elapsed_wait_penalty \
+ av_curb_wait
elif j_veh.status == "enroute_pickup":
dist_assgn[count_pass][count_veh] = Distance.dist_manhat_pick(i_pass, j_veh) - elapsed_wait_penalty \
+ Set.reassign_penalty
else:
sys.exit(
"Something wrong with AV state - idlePick_minDist")
elif i_pass.status == "assigned":
prev_assign[count_pass] = 1
if j_veh.next_pickup == i_pass:
if j_veh.status == "enroute_pickup":
dist_assgn[count_pass][
count_veh] = Distance.dist_manhat_pick(
i_pass, j_veh) - elapsed_wait_penalty
else:
sys.exit(
"Something wrong with current AV-customer match - idlePick_minDist"
)
elif j_veh.status in ("idle", "relocating"):
av_curb_wait = j_veh.curb_time_remain * Set.veh_speed
dist_assgn[count_pass][count_veh] = Distance.dist_manhat_pick(i_pass, j_veh) - elapsed_wait_penalty \
+ Set.reassign_penalty + av_curb_wait
elif j_veh.status == "enroute_pickup":
dist_assgn[count_pass][count_veh] = Distance.dist_manhat_pick(i_pass, j_veh) - elapsed_wait_penalty \
+ 2*Set.reassign_penalty
else:
sys.exit(
"Something wrong with AV state - idlePick_minDist")
else:
sys.exit(
"Something wrong with customer state - idlePick_minDist")
t1 = time.time()
# Model
models = gurobipy.Model("idlePick_minDist")
models.setParam('OutputFlag', False)
# Decision Variables
for i in range(len_no_assign_or_pick_cust):
for j in range(len_idle_n_pick_av):
x[i][j] = models.addVar(vtype=gurobipy.GRB.CONTINUOUS,
obj=dist_assgn[i][j],
name='x_%s_%s' % (i, j))
models.update()
# constraints
# Previously assigned passengers must be assigned a vehicle
for ii in range(len_no_assign_or_pick_cust):
models.addConstr(
gurobipy.quicksum(x[ii][j] for j in range(len_idle_n_pick_av)) -
prev_assign[ii] >= 0)
if len_no_assign_or_pick_cust <= len_idle_n_pick_av:
for ii in range(len_no_assign_or_pick_cust):
models.addConstr(
gurobipy.quicksum(x[ii][j]
for j in range(len_idle_n_pick_av)) == 1)
for jj in range(len_idle_n_pick_av):
models.addConstr(
gurobipy.quicksum(
x[i][jj] for i in range(len_no_assign_or_pick_cust)) <= 1)
else:
for ii in range(len_no_assign_or_pick_cust):
models.addConstr(
gurobipy.quicksum(x[ii][j]
for j in range(len_idle_n_pick_av)) <= 1)
for jj in range(len_idle_n_pick_av):
models.addConstr(
gurobipy.quicksum(
x[i][jj] for i in range(len_no_assign_or_pick_cust)) == 1)
models.optimize()
if models.status == gurobipy.GRB.Status.OPTIMAL:
for n_veh in range(len_idle_n_pick_av):
found = 0
for m_pass in range(len_no_assign_or_pick_cust):
if x[m_pass][n_veh].X == 1:
win_cust = no_assign_or_pick_cust[m_pass]
win_av = idle_n_pick_avs[n_veh]
if win_av.next_pickup != win_cust:
if win_cust.status == "unassigned":
Person.update_person(t, win_cust, win_av)
elif win_cust.status == "assigned":
win_cust.status = "reassign"
Person.update_person(t, win_cust, win_av)
if win_av.status in ("idle", "relocating"):
temp_veh_status = "base_assign"
elif win_av.status == "enroute_pickup":
temp_veh_status = "reassign"
else:
temp_veh_status = "wrong"
Vehicle.update_vehicle(t, win_cust, win_av,
Regions.SubArea,
temp_veh_status)
found = 1
break
if found == 0:
no_win_av = idle_n_pick_avs[n_veh]
if no_win_av.status == "enroute_pickup":
temp_veh_status = "unassign"
Vehicle.update_vehicle(t, Person.Person, no_win_av,
Regions.SubArea, temp_veh_status)
else:
sys.exit("No Optimal Solution - idlePick_minDist")
# print("Vehicles= ", tot_veh_length, " Passengers= ", len_no_assign_or_pick_cust, " time=", time.time()-t1)
return
def create_model_3(depot, hubs, clients, cost_matrix_1, cost_matrix_2):
"""
Model 3 = A simple 2E-CVRP with no tour decision variable
:param robots: an object with location and number
:param depot: location of the depot station, name of the depot
:param hubs: each hub has a location
:param clients: clients have names,
:param cost_matrix_1: a dict that uses the keys of locations to give you the distance (Depot - Hubs)
:param cost_matrix_2: a dict that uses the keys of locations to give you the distance (Hubs-clients, Hubs-clients)
(ex. cost_matrix_1[h.name,d.name] = distance between hub i and depot j)
:return: Gurobi model with the related constraints
"""
### Creating model:
model = gp.Model("2E-VRP")
#####################################################
# 1. sets & parameters (parameters should be already defined in __init__)
#####################################################
# a. define maximum number of robots:
# TODO: move this to create an object list of robots
n_c = len(clients) # number of customers
R_max = len(clients)
R = list(range(R_max)) # list of robots
Dep = list(range(len(depot))) # list of depots
C = [c.name for c in clients] # list of clients
D_c = [c.demand for c in clients] # list of demands
L = list(set([l[0] for l in cost_matrix_2.keys()])) # L = m+n
# TODO: transform L to be a set of str.
H = list(range(len(hubs))) # list of hubs (hubs 0,1,2)
Loc_c = [c.loc for c in clients] # list of locations
Loc_c_dict = {c.name: c.loc
for c in clients
} # dict of locations - will not be used most probably
R_cap = 50 # maximum robot capacity
R_dist = 200
V_cap = 300
# defining the number of trucks by calculating the total demand
total_demand = math.fsum(c.demand for c in clients)
t = math.ceil((total_demand / V_cap))
V = list(range(t)) # list of trucks
# fixed costs
c_hub = 50
c_truck = 50
c_robot = 10
# variable cost
c_truck_distance = 5
#####################################################
# 2. decision variables
#####################################################
# hub open/not-open
o = model.addVars(H, vtype=GRB.BINARY, name="o")
# robot-client-hub assignment
x = model.addVars(C, R, H, vtype=GRB.BINARY, name="x")
# robot assignment
# robot = model.addVars(R, vtype=GRB.BINARY, name="robot")
# van-hub assignment
# TODO: compare with the TSP formulation
# create a list of Depots + Hubs
# DH = [0]+H # in our case it is only one depot, the model is extendable to include multiple (if we..)
DH = list(range(len(hubs) + 1))
# TODO: remember in the distance matrix hubs are 0,1,2 and depots are 3
# y = model.addVars(V, DH, DH, vtype=GRB.BINARY, name="y")
y = model.addVars(
V, DH, DH, vtype=GRB.BINARY,
name="y") # TODO: you need the subtour elimination constraint
# TODO: set the variable to 0 when dh1 = dh2
# edges assignment to robots
# z = model.addVars(L, L, R, vtype=GRB.BINARY, name="z")
# a new decision variable to assign robots to hubs
w = model.addVars(R, H, vtype=GRB.BINARY, name="w")
#####################################################
# 3. constraints
#####################################################
# (2) - all client nodes are assigned to hubs
# implemented in constraint (D)
# (4) - if robot serves a client, the corresponding hub must be open
model.addConstrs(
(gp.quicksum(x[c, r, h] for r in R) <= o[h] for c in C for h in H),
name="no_hub_no_robot")
# (8) - all the trucks must return to the depot station
model.addConstrs((gp.quicksum(y[v, dh, len(H)]
for dh in DH[:-1] if dh != len(H)) == t
for v in V),
name="trucks1")
# (9) - all trucks must depart from the depart station
model.addConstrs((gp.quicksum(y[v, len(H), dh]
for dh in DH[:-1] if dh != len(H)) == t
for v in V),
name="trucks2")
# (10) - number of trucks leaving = number of trucks returning
# TODO: review indexing
model.addConstrs((gp.quicksum(y[v, dh1, dh2] for v in V
for dh1 in DH[:-1] if dh1 != dh2)
== gp.quicksum(y[v, dh2, dh1] for v in V
for dh1 in DH[:-1] if dh1 != dh2)
for dh2 in DH[1:]),
name="trucks3")
# (11) - sum of all trucks going to the same hub is 1 if hub is open
model.addConstrs(
(gp.quicksum(y[v, dh1, dh2] for v in V
for dh1 in DH[:-1] if dh1 != dh2) == o[dh2]
for dh2 in DH[:-1]),
name="if_truck_then_hub") # TODO: test with putting cstr 11 before 4
# (16) - maximum truck capacity - needs correction
#model.addConstrs(((gp.quicksum(D_c[int(c)]*x[c, r, h] for c in C for r in R for h in H))
# <= V_cap*(gp.quicksum(y[v, dh1, dh2] for v in V for dh1 in DH for dh2 in DH if dh1 != dh2))),
# name="truck_cap")
# (17) - adding constraint 17 from the article for testing
# model.addConstr((gp.quicksum(o[h] for h in H) == len(H)),name="constraint_17")
# we shouldn't need constraint 17 if the variable o is activated by a constraint x which is already written
# (A) - distance covered by every robot should be less than maximum allowed distance
#model.addConstrs((gp.quicksum(cost_matrix_2[c1, c2] * z[c1, c2, r] for c1 in L for c2 in L if c1 != c2)
# <= R_dist for r in R), name="robot_dist")
# forming a tour contstraint (22) :
# (B1) - ensure that the robot assigned to a client travels to another location after every client
#model.addConstrs((gp.quicksum(z[int(c1), int(c2), r] for c1 in C if c1 != c2) == gp.quicksum(x[c2, r, h] for h in H) for r in R for c2 in C),
# name="tour1")
# (B2) - same but from j --> i
#model.addConstrs((gp.quicksum(z[int(c2), int(c1), r] for c1 in C if c1 != c2) == gp.quicksum(x[c2, r, h] for h in H) for r in R for c2 in C),
# name="tour2")
# Another solution would be to add a very high cost to the same node.
# maybe will need to add constraint to the depot.
# (B3) - ensure that each robot forms a tour from and to the same hub
# Same hub constraints:
# TODO: compare between running with B3 and B4 and without
# another solution is to assign robots to hub prior to optimization
# model.addConstrs((gp.quicksum(z[int(c), n_c + h, r] for c in C) == w[r, h] for r in R for h in H),
# name="sameHub1")
# (B4) - same as (B3) but from j --> i
#model.addConstrs((gp.quicksum(z[n_c + h, int(c), r] for c in C) == w[r, h] for r in R for h in H),
# name="sameHub2")
# (C) - respect maximum robot capacity
model.addConstrs((gp.quicksum(D_c[int(c)] * x[c, r, h]
for c in C) <= R_cap * w[r, h] for r in R
for h in H),
name="robot_cap")
# (D) - each client is served by a single robot from a single hub
model.addConstrs(
(gp.quicksum(x[c, r, h] for h in H for r in R) == 1 for c in C),
name="Just_one_hub")
# (E) - same as (D) but summing over all clients and all hubs
model.addConstrs((x.sum(c, '*', h) <= 1 for c in C for h in H),
name="Just_one_robot")
# (D) is shadowed by (E)
# (WO) - 25/03 - set o == 1 if w == 1
model.addConstrs(
gp.quicksum(w[r, h] for r in R) <= R_max * o[h] for h in H)
# (R) - decision variable r constraints - needs revision
#model.addConstrs(robot[r] == gp.quicksum(z[int(c1), int(c2), r] for c1 in C for c2 in C if c1 != c2) for r in R)
# model.addConstrs((gp.quicksum(x[c, r, h] for c in C for h in H) <= robot[r] for r in R),
# name="no_robot_no_robot")
# TODO: check if we need constraints on w
# each robot is assigned to one hub
model.addConstrs((w.sum(r, '*') <= 1 for r in R),
name="Just_one_hub_for_robot")
# Additional constraints for "z"
# model.addConstrs((gp.quicksum(z[c1, c2, r] for r in R) <= 1 for c1 in L for c2 in L if c1 != c2), name="only_one_robot_z")
#####################################################
# 4. Objective function
#####################################################
# 1. truck distance cost
cost_truck_var = gp.quicksum(cost_matrix_1[dh1, dh2] * c_truck_distance *
y[v, dh1, dh2] for v in V for dh1 in DH
for dh2 in DH if dh1 != dh2)
# 2. truck fixed cost
cost_truck_fixed = c_truck * t # this will change in v2
# 3. robot distance cost
#cost_robot_var = gp.quicksum(cost_matrix_2[c1, c2] * z[c1, c2, r]
# for c1 in L for c2 in L if c1 != c2 for r in R)
# 4. robot fixed cost
cost_robot_fixed = gp.quicksum(c_robot * w[r, h] for r in R for h in H)
# 5. hub fixed cost
cost_hub_fixed = gp.quicksum(c_hub * o[h] for h in H)
model.setObjective(
cost_truck_var + cost_truck_fixed + cost_robot_fixed + cost_hub_fixed,
GRB.MINIMIZE)
#####################################################
# 5. Saving LP model (remember to change versions)
#####################################################
model.write("../output/lp_model/2EVRP_v2_truck.lp")
#### Optimize!
model.optimize()
status = model.Status
if status in [GRB.INF_OR_UNBD, GRB.INFEASIBLE, GRB.UNBOUNDED]:
print("Model is either infeasible or unbounded.")
sys.exit(0)
elif status != GRB.OPTIMAL:
print("Optimization terminated with status {}".format(status))
sys.exit(0)
#####################################################
# 6. analyzing solutions
#####################################################
print('Optimal solution found with total cost: %g' % model.objVal)
for v in model.getVars():
print('%s %g' % (v.varName, v.x))
# 1. analyzing hubs
print("Analyzing hubs")
# 2. analyzing robots
print("Analyzing Robots")
# 3. analyzing tours (echelon 1)
print("Analyzing tours (depot - hubs)")
# 4. analyzing tours (echelon 2)
print("Analyzing tours (hubs - clients)")
def build_objectif(self):
c, p = self.dimensions['c'], self.dimensions['p']
h, Q = self.constants['h'], self.constants['conflicts']
obj_room = gb.quicksum(self.vars['x'].itervalues())
dist = {(i, j): Q.get((i, j), 0) * gb.quicksum(h[l] * (self.vars['x'][i, l] - self.vars['x'][j, l]) for l in self.constants['available'][i])
for i in range(c) for j in range(c)}
obj = self.gamma * gb.quicksum([dist[i, j] * dist[i, j] for i in range(c) for j in range(c)]) - obj_room
self.problem.setObjective(obj, gb.GRB.MAXIMIZE)
return True