Storing gap and runtime

ユーザーの入力を待っています。

Min Lee

2023年05月18日 03:57

• 
• 

Good morning, I was hoping to as a question on storing gap, and runtime. I use the following command: 

#storing computation time and gap
def data_cb(model, where):
    if where == gp.GRB.Callback.MIP:
        cur_obj = model.cbGet(gp.GRB.Callback.MIP_OBJBST)
        cur_bd = model.cbGet(gp.GRB.Callback.MIP_OBJBND)
        cur_time = model.cbGet(GRB.Callback.RUNTIME)
        model._data.append([cur_time, cur_obj, cur_bd])

network_size = [200, 500]
running_count = 10  # number of different seeds used




for ns in network_size:
   for r in range(1,running_count+1):

        m.Params.TimeLimit = 6*60*60 #run for 6 hours maximum
        m.Params.MIPGap = 0
        # Solve the model

        m._obj = None
        m._bd = None
        m._data = []
        m._start=time.time()
        m.optimize(callback = data_cb)

        #m.terminate()

        # Output Reesults
        m.write('linear_model.lp')

        gap_and_bound = m._data[-1]
        final_gap = (abs(gap_and_bound[1] - gap_and_bound[2])/abs(gap_and_bound[1]))*100
        duration1 = gap_and_bound[0]

        if m.Status != GRB.Status.OPTIMAL:
            print('Optimization was terminated before the optimal with status ' + str(m.Status))
#            sys.exit(0)

        # Print number of solutions stored
        nSolutions = m.SolCount
        print('Number of solutions found: ' + str(nSolutions))

        m.params.ObjNumber = 0
        welfare_sheltering_occupancy_level = - m.ObjNVal
        print('Total welfare-oriented sheltering capacity =', welfare_sheltering_occupancy_level, end='\n')
        m.params.ObjNumber = 1
        Total_cost = round(m.ObjNVal, 2)
        print('Total cost =', Total_cost, end='\n')
        Overall_obj_value = W2*Total_cost - W1*welfare_sheltering_occupancy_level
        print('Overall obj value =', Overall_obj_value)

        model_output.loc[r-1] = summary_result
        print(model_output.loc[r-1])
        

    model_output.to_excel(my_folder+"/Outcome_Summary_Location" + str(ns) + ".xlsx", sheet_name = 'Network'+str(ns))

 

 

When I run the model, I get this error: list index out of range for m._data[-1]. It does not happen all the time, but it does  happen for some occasions. 

 

Any help would be appreciated!

 

Another question I have is that the r loop completely stops when there is no optimal solution. Is there a way to jump to the next iteration even when no solution can be found? 

 

Thank you so much for your time and help.

 

Sincerely,

 

Min

関連

• Callbacks

0

• 

  Jaromił Najman

  • Gurobi Staff

  2023年05月22日 08:51

  Hi Min,

  When I run the model, I get this error: list index out of range for m._data[-1]. It does not happen all the time, but it does  happen for some occasions. 

  It is possible that a model is solved in Presolve and thus will never reach the MIP callback. Could you show the log output of a case where this error occurs?

  In such cases you should always first check the size of \(\texttt{m._data}\) via \(\texttt{len(m.data)}\). If the length is \(0\) but the status is \(\texttt{GRB.OPTIMAL}\), you can still compute one data point via accessing the model attributes ObjVal, ObjBound, and RunTime.

  Another question I have is that the r loop completely stops when there is no optimal solution. Is there a way to jump to the next iteration even when no solution can be found? 

  I don't see why the loop should completely stop. The sys.exit line is commented out. Couldn't you just use the \(\texttt{continue}\) keyword in the non-optimal case?

  Best regards, 
  Jaromił

  0

サインインしてコメントを残してください。