This model was proposed in [LS02]. It consists in a set of preys (potential solution) and another one of predators which are freely moving in a lattice. Predators kill the preys they find and the preys will breed along the evolution. Hence, this method can be classified into the methods having a varying population density, since the number of prays will vary along the evolution. Basically, the algorithm works like follows:
1. Each prey has a chance to move, which is assigned to 0.5 in the paper. If a prey is allowed to move, then it moves to one free node among its 8 neighbors.
2. After they have move, they are allowed to breed with their neighbor preys.
3. Now it is the predator's turn. They look for a prey among their neighbors locations. If a predator does not find any pray in its neighborhood, it moves onto a neighbor cell; otherwise, it kills the weakest prey of its neighborhood and moves onto its cell.
4. While a number of required evaluations is not reached, go to step 1.
.