One reviewer points out that we need to pursue additional test cases in order to make a more meaningful assessment of our results. Ideally we would apply our approach to problems with known optimum solutions, but such problems tend to be smaller and tend to use simplified design criteria. A second option would be to compare our results against other methods, and we would like to do this if we can find large scale benchmarks that use standard design criteria, and that have been made public. Our preferred plan for validating our algorithm is to compare our results against those produced by human experts in commercial building projects.