def branch_n_price(n, demands, capacity, distances, MasterProb):
queue = PriorityQueue()
MasterProb.RelaxOptimize()
obj_val = MasterProb.relax_modelo.ObjVal
queue.insert(obj_val, MasterProb)
best_int_obj = 1e3
best_relax_obj = 1e3
nodes_explored = 0
best_model = None
while not queue.isEmpty():
obj_val, MP_branch = queue.delete()
nodes_explored += 1
MP_branch.RelaxOptimize()
solution = MP_branch.getSolution()
duals = MP_branch.getDuals()
branch_cost = MP_branch.getCosts()
branch_routes = MP_branch.modelo.getA().toarray()
sol_is_int = all([float(round(s, 4)).is_integer() for s in solution])
# sol_is_int = all([False if i > 0.3 and np.abs(i - 1.0) > 0.3 else True for i in solution ])
if obj_val < best_int_obj and sol_is_int:
print(f"Best Integer Obj: {obj_val}")
print(f"Nodes explored: {nodes_explored}")
best_int_obj = obj_val
# print(f"best sol: {solution}")
best_model = copy_model(branch_cost, branch_routes, MP_branch)
if obj_val < best_relax_obj:
print(f"Best Relaxed Obj: {obj_val}")
print(f"Nodes explored: {nodes_explored}")
best_relax_obj = obj_val
# --- # --- # Column generation # --- # --- #
new_MP = column_generation(n, demands, capacity, distances, duals,
MP_branch)
if new_MP != None:
new_MP.RelaxOptimize()
branch_cost = new_MP.getCosts()
branch_routes = new_MP.modelo.getA().toarray()
if new_MP.relax_modelo.ObjVal <= best_relax_obj:
queue.insert(new_MP.relax_modelo.ObjVal,
copy_model(branch_cost, branch_routes, new_MP))
else:
# --- # If stopped col generation then branch if solution is not integer # --- #
if not sol_is_int:
# print("#--#--#--# Not integer solution ........Branching")
queue = branch(branch_cost, branch_routes, n, demands,
capacity, distances, duals, solution, MP_branch,
queue, best_relax_obj)
else:
# print(f"best sol: {solution}")
best_model = MP_branch
return best_model
