Agent-Based Models and Optimal Control in Biology: A Discrete Approach - Mathematical Concepts and Methods in Modern Biology - page 163

Exercise 5.18. Explain the difference between roulette and tournament selection.
How would roulette selection be different if we were interested in schedules with the
highest cost?

Exercise 5.19. Explain the difference between uniform and 1-point crossover.
Describe a control problem for which uniform crossover is more appropriate, and
another for which 1-point crossover is more appropriate.

Exercise 5.20. Explain the mutation methods invert and neighbor swap .
Explain the advantages and disadvantages of each.

Exercise 5.21. Devise and explain methods for selection, crossover, and mutation
other than those available in the model. Describe advantages of each.

Exercise 5.22. The retention slider determines how many of the top solutions
should be carried over to the next generation. Describe a potential pitfall of setting
this value to 0. Describe a potential pitfall of setting this value too high.

Exercise 5.23. Notice that the default number of runs is 100. What benefit is there to
setting this value lower? What is the risk? What are the advantages and disadvantages
of setting this value higher?

Exercise 5.24. Click restore-defaults , then click setup . Toward the top
of the interface, make sure view updates is unchecked. Run the algorithm (this
may take a while). Make note of the best schedule found. Now run it again, only
changing the poison so that it degrades. Again, take note of the best schedule found
(note that this data should be saved in a .csv file in the folder where the model is
saved). What differences do you notice? Why do you think this is?

Exercise 5.25. Make sure that poison-degrade is set to On and that poison
-strength is set to 0.75. Choose values for the various sliders and methods on
the right-hand side of the interface tab (i.e., the algorithm settings) in an attempt to
minimize the cost. Use your intuition as a starting point, then revise the values based
on the output (make sure write-to-file? is turned on). What is the minimum
cost you are able to achieve? What pattern (if any) do you notice in the best schedule
found?

Exercise 5.26. Implement your ideas from Exercise 5.21 into the code and run
the genetic algorithm using them. Do they perform better or worse than the original
methods? 1

Exercise 5.27. Suppose the objective is simplified so that the goal is to simply
minimize the number of rabbits, without regard to the poison cost. What do you think
the best solution would be in this case? Alter the cost function in the code accordingly,
and run the GA. Does the outcome reflect your intuition?

Exercise 5.28. Determine a new objective and alter the cost function accordingly.
Run the GA. Discuss the pattern of the best schedule found, and state whether or not
this seems to make sense in light of your cost function.

1 This exercise requires knowledge of Netlogo programming.