"... In PAGE 3: ...Our GP system was set up to be the same as given in [Koza, 1992, pages 147{155] except the populations were allowed to continue to evolve even after an ant succeeded in traversing the whole trail, programs are restricted to a maximum length of 500 rather than to a maximum depth of 17, each crossover produces one child rather than two, tournament selection was used and the ants were allowed 600 operations (Move, Left, Right) to complete the trail. The details are given in Table1 , parameters not shown are as [Koza, 1994, page 655]. On each version of the problem 50 independent runs were conducted.... In PAGE 11: ... While the expected size of crossover fragments depends in detail upon the trees selected as parents and the relative weighting applied to functions and terminals cf. Table1 , typically both the inserted subtree and the subtree it replaces consist of a function and its leafs. Since these subtrees are short together they produce a small change in total size.... ..."

"... In PAGE 7: ....4. Recruitment and nest moves We performed one experiment without recruitment, one in which the nest was left at the same position during the whole run and where ants never forgot about their hunting sites (T D1, Plocal D1), and finally one experiment with API without nest moves but with ants that forgot about hunting sites (T D1, Plocal D 50). In Table4 , the results of these experiments are compared to API using the default parameter settings. These experiments have shown the following: 1.... ..."

"... In PAGE 13: ... The example concerns multiple agents (the ants), each of which has input (to observe) and output (for moving and dropping pheromones) states, and a physical body which is at certain positions over time, but no internal mental state properties (they are assumed to act purely by stimulus-response behaviour). An overview of the formalisation of the state properties of this example is shown in Table1 . In these state properties, a is a variable that stands for ant, l for location, e for edge, and i for pheromone level.... ..."

"... In PAGE 4: ... In the case that there is no pheromone, the ant moves in a random way seeking the anthill or another pheromone path but leaving its pheromone to build the path for other ants. Table1 describes the cell state codification using a 5- digit natural number. Table 1.... ..."

"... In PAGE 7: ... Note that the database server(s) and the message server are not part of the business value and process model, so their impact on the costs cannot be assessed by evaluating a business value or process model in isolation. Table3 and Table 4 show costs for the scenario submit ad for all actors (except contact searchers). For the submit ad scenario, 4 paths can be identified (paths 1,2 for ads which are published and locally or remotely checked, paths 3,4 for ads which are rejected and locally or remotely checked).... ..."

"... In PAGE 3: ... The evolutionary system we use is identical to [LP97b] except the crossover operator is replaced by mutation. The details are given in Table1 , parameters not shown are as [Koz94, page 655]. On each version of the problem 50 independent runs were conducted.... ..."