Item allocation in a knapsack with Choco Solver

Question

I try to implement a multidimensionnal knapsack problem with Choco Solver through JAVA. My idea is to assign 3 items in 2 knapsacks.

My item have a weight and knapsack a limit : int[] itemWeight = {2, 2, 2}; int[] knapsackLimit = {4, 4};

And my decision variable where 3 items have a knapsack between {0, 1} : int[] itemAllocation = {1, 1, 0};

I wrote this problem by using Choco Solver :

Model model = new Model("KnapsackProblem");

// Allocation variable
IntVar[] x = new IntVar[3];
for (int i = 0; i < itemNumber; i++) {
    x[i] = model.intVar("x"+i, 0, knapsackNumber-1);
}

// Knapsack capacities variables
IntVar[] limitVar = new IntVar[knapsackNumber];
for (int i = 0; i < knapsackNumber; i++) {
    limit[i] = model.intVar(knapsackLimit[i]);
}

IntVar[] itemWeightVar = new IntVar[itemNumber];
for (int i = 0; i < itemNumber; i++) {
    itemWeightVar[i] = model.intVar(0, 2);
    model.element(itemWeightVar[i], itemWeight,x[i]);
}

// Limit Cosntraints
for (int i = 0; i < knapsackNumber; i++) {
    model.sum(itemWeightVar, "<=", limit[x[i].getValue()]);
}

model.getSolver().solve();

Unfortunately, this method does not work. I always get the following allocation: [x0 = 0, x1 = 0, x2 = 0]

Thank you in advance.

Answer 1

This is a common mistake, you simply forget to POST your constraints so they are not taken into account. For instance, you should replace

model.element(itemWeightVar[i], itemWeight,x[i]);

by

model.element(itemWeightVar[i], itemWeight,x[i]).post();

Note that the post is not automatic as you may want to reify the constraint instead of posting it (which happens within ifThen statements for instance).

Best,

https://www.cosling.com/