Pack constraint strange behavior

[AntonisKl]

Hello, I want to use the Pack() function (https://developers.google.com/optimization/reference/python/constraint_solver/pywrapcp#pack_1) of the Routing solver.

I read the documentation and I wrote some code based on the basic VRP example (https://developers.google.com/optimization/routing/vrp#entire_program1) to test this function.

In the above VRP example, I just added the following code before the solver starts to search for a solution:

    solver = routing.solver()

    all_vars_to_pack = [solver.IntVar(0,2,"a"), solver.IntVar(0,2,"a")]

    for var in all_vars_to_pack:

        routing.AddToAssignment(var)

    pack_constraint = solver.Pack(all_vars_to_pack, 2)

    usage_vars = [solver.IntVar(4,4,"a"), solver.IntVar(4,4,"a")]

    for var in usage_vars:

        routing.AddToAssignment(var)

    pack_constraint.AddSumVariableWeightsLessOrEqualConstantDimension(usage_vars, [3, 5])

    solver.Add(pack_constraint)

The problem is this: When the program is executed, I see that it packed both of the items (quantity 4+4=8) in the 2nd bin which is not feasible because the capacity of the 2nd bin is 5.

My questions are the following:

1. Is there a way to avoid the mixing of items?

2. Are there any examples of how to properly use the Pack class and Pack() function while trying to solve a VRP?

[lperron]

What are the values of all_vars_to_pack in the assignment ? Laurent Perron | Operations Research | lperron@google.com | (33) 1 42 68 53 00 Le dim. 28 juin 2020 à 20:58, AntonisK <notifications@github.com> a écrit :

…

Hello, I want to use the Pack() function ( https://developers.google.com/optimization/reference/python/constraint_solver/pywrapcp#pack_1) of the Routing solver. I read the documentation and I wrote some code based on the basic VRP example ( https://developers.google.com/optimization/routing/vrp#entire_program1) to test this function. In the above VRP example, I just added the following code before the solver starts to search for a solution: solver = routing.solver() all_vars_to_pack = [solver.IntVar(0,2,"a"), solver.IntVar(0,2,"a")] for var in all_vars_to_pack: routing.AddToAssignment(var) pack_constraint = solver.Pack(all_vars_to_pack, 2) usage_vars = [solver.IntVar(4,4,"a"), solver.IntVar(4,4,"a")] for var in usage_vars: routing.AddToAssignment(var) pack_constraint.AddSumVariableWeightsLessOrEqualConstantDimension(usage_vars, [3, 5]) solver.Add(pack_constraint) The problem is this: When the program is executed, I see that it packed both of the items (quantity 4+4=8) in the 2nd bin which is *not* feasible because the capacity of the 2nd bin is 5. My questions are the following: 1. Is there a way to avoid the mixing of items? 2. Are there any examples of how to properly use the Pack class and Pack() function while trying to solve a VRP? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2087>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACUPL3I6FVLGCK5J52SLMSTRY6HHFANCNFSM4OKU63RA> .

[AntonisKl]

Each var of all_vars_to_pack list has 1 as value in the assignment.