The paper describes a new algorithm for finding the pareto front of a multiobjective optimization problem.

Many optimization problems in design and engineering have more than one objective associated with them. For example an engineer designing a reactor would like to maximize safety while minimizing cost. The traditional approach converts these two objectives into a single objective and then uses standard single objective optimization procedures on the resulting problem. NSGA-II, on the other hand, finds an entire set of solutions representing various tradeoffs between safety and cost, which are optimal in the sense that no member of the solution set if better than any other member on both the objectives. What makes NSGA II significant is that it is able to do this search robustly and efficiently while delivering consistently good results. It was one of the first of a new generation of algorithms which were good enough to be used on real world problems.

Dr. Kalyanmoy Deb of the Kanpur Genetic Algorithms Laboratory at the Indian Institute of Technology in Kanpur has been involved with both genetic algorithms as well as multi-objective optimization for a while. He had Amrit Pratap and Mary (T. Meyarivan) working on some ideas for the second generation NSGA while I was working on an entirely different project. Over time I realized that I found what they were working on to be far more exciting than my own problem, and promptly proceeded to ignore my own work and dive into their project instead.