Adding Penalties

[Antonia Calvi]

I have a cumulative variable over a large horizon.

I would like to penalise if its modulus is outside of a certain range.

I was thinking of iterating through the different nodes and creating conditional expressions with the solver which would assign say an IntVar(0) if it is within the range, otherwise a penalty of 50.

I would then like another variable to store the sum of the overall penalties and then add that to minimise the finaliser - this however does not seem to change anything.

My output does not seem to take account of the penalties.

I am also unsure how to iteratively add to an IntVar... any suggestions?

I use python btw

[Laurent Perron]

My understanding

Horizon 100

Cumulative, safe level 8, max level 10

You want to penalize by 50 if you are above the safe level on each quarter of the horizon 

Create 4 optional intervals with fixed dated (start = 0, duration = 25, presence = lit0; start = 25, duration = 25, presence = lit1, ...). 

Add them to the cumulative.

Create negation of lit0, lit1, lit2, lit3

Add 50 * lit_i to the objective.

Does it solve your problem?

--Laurent

Laurent Perron | Operations Research | lpe...@google.com | (33) 1 42 68 53 00

--
You received this message because you are subscribed to the Google Groups "or-tools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to or-tools-discu...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Laurent Perron]

I meant add 50 * not_lit_i to the objective.

[Antonia Calvi]

I don't really follow your reasoning.. what I am having is a routing problem, my time horizon is within a few day, but I want to penalise if a vehicle arrives within closing hours at a specific day - hence taking the modulus.

right now what I have is something of the sort:

sl = solver.IntVar(0, 0)

    for cust in customers.customers:

        if (cust.tw_open is not None) and (cust.tw_close is not None):

       

            less1 = solver.ConditionalExpression(solver.IsGreaterOrEqualCstVar(((time_dimension.CumulVar(routing.NodeToIndex(cust.index)))%86400),28000).Var(), solver.IntVar(0,0), 50).Var()

         

            more1 = solver.ConditionalExpression(solver.IsLessOrEqualCstVar((time_dimension.CumulVar(routing.NodeToIndex(cust.index))%86400),28800).Var(), solver.IntVar(0,0), 50).Var()

            

            slack = solver.Max(less1, more1).Var()

            sl=solver.Sum([sl,slack]).Var()

            solver.Minimize(sl,1)

 

    routing.AddVariableMinimizedByFinalizer(sl)

[Antonia Calvi]

Also.. in routing problems how do you add directly to the objective? I have found difficulty doing this..

[Laurent Perron]

I was proposing a scheduling solution with the CP-SAT solver.

You want a feature of the routing library. Please forget my suggestion.

--Laurent

Laurent Perron | Operations Research | lpe...@google.com | (33) 1 42 68 53 00

[Antonia Calvi]

You don't happen to have suggestions do you?

[Laurent Perron]

No, I will let the routing expert answer you :-)

[Samir Sabri]

Thanks :)

--
Kind Regards,

--------------------------------------------- 

Samir Sabri
Software Architect& Developer

www.dcaclab.com

Jordan-Middle East