Trying to work with maximum-value causes model is infeasible-error

Answered

Frank Diederichsen

April 16, 2023 21:00 Edited

• 
• 
• 

Hello,
I'm trying to solve an optimization problem by the gurobi optimizer, but I'm getting the model is infeasible message.
The following part causes the problem:

mf = {(i,j): opt_model.addVar(vtype=grb.GRB.CONTINUOUS,name="mf_{0}_{1}".format(i,j)) for i in set_I for j in set_J} 
                 .
                 .
                 .
        for j in range(0, num_tasks):
            # Constraint: start-time of every task in group i is greater than group-start
            opt_model.addConstr(
                ts_vars[i,j] >= gs_vars[i]
            )
            if j > 0:
                # Constraint: get the largest finish-time of preceding tasks as start-time for task j
               opt_model.addConstr(
                   mf[i,j] == grb.max_(tf_vars[i,j_dash] for j_dash in get_preceding_tasks(set_J,j))
               )
               opt_model.addConstr(
                    ts_vars[i,j] >= mf[i,j]
                )
            # Constraint: finish-time of every task is start-time + runtime of the task
            opt_model.addConstr(
                tf_vars[i, j] + eps >= grb.quicksum(
                    x_vars[i, j, k] * get_runtime(workloads[j], tasktypes[j], max_widths[j], k, i,
                                                  cores) for k in set_K) + ts_vars[i, j]
            )

ts_vars, tf_vars and gs_vars are of the type continous.
i and j are integers.
x_vars are of the type binary. For every j there is exactly one x[i,j,k], so e.g. the sum of x[i,j,k] and x[i',j,k] would be 1.
the get_runtime-function calculates the runtime of type float for every task (it can be 0.0).
gs_vars stands for the start-time of a group, it is calculated before.

The start-time of every task should be the maximum of the finish-times of the preceeding tasks in the group, so that tasks run one after the other.
Both constraints, 

# Constraint: get the largest finish-time of preceding tasks as start-time for task j and 
# Constraint: finish-time of every task is start-time + runtime of the task

do not work together. If I use only one of them, the program runs (of course, with unusable results).
I'm not sure if the mathematical approach is incorrect or if I simply did some programming errors, as I am very new to gurobi.
Any help would be much appreciated.

With kind regards
Frank

Related to

• Infeasible

0

• 

  Mario Ruthmair

  • Gurobi Staff

  April 17, 2023 09:44

  Hi Frank,

  Did you already try to follow the suggestions in this article about how to find out why some model is infeasible?

  Best regards,
  Mario

  0
• 

  Frank Diederichsen

  April 17, 2023 20:45

  • 
  • 
  • 

  Hi Mario,
  thanks for the immediate reply.
  Well, I read about computing an IIS but I thought this will be overdoing for my problem.
  Meanwhile I tried to set up an IIS and I'm examinating the result.
  
  Thanks again and best regards
  Frank

  0

Please sign in to leave a comment.