|William W. Cooper was a real gentleman, with a bright mind and a special skill for explaining his appealing ideas. One of the outstanding features of his charming character was his enthusiasm. Keenness for life and keenness for research. The intellectual couple integrated by Bill and Ab Charnes has been one of the most intriguing examples of joint research performance in the history of Mathematical Programming and Management Science. I first met Bill in 1981 in Austin, Texas. It was during an international meeting on Semi-infinite Programming. I met Bill for the second time in Aachen, during the 1991 EURO meeting. He was among the American people who came over to Europe to disseminate Data Envelopment Analysis (DEA). Knox Lovell was also there. I talk to both of them about common future DEA research and invited them to visit me in Spain. In 1994 Cooper was the keynote speaker in the meeting of the Spanish Statistical and Operations Research Society. I presented him my work about translation invariance DEA models as well as my critics about the MIP and MEP measure. As a consequence, we started talking about developing new efficiency measures based on the additive model. At that time, it was for me very stimulating to receive most of his DEA working papers. In 1995 my former University of Alicante distinguished him with a PhD Honoris Causa, and I was his protector. It was my opportunity for knowing him more deeply and for understanding what he was proud about. Basically he was proud of his contribution to the well-being of related people. He was also proud of his wife Ruth and of his common work with Abraham Charnes. Specifically, he mentioned several times the method of â€œcreating theory based on applicationsâ€. In late forties and working with the engineer Bob Mellon, they developed the first mathematical programming model applied to industry. It was a completely new blending model of gasolines for used in an integrated oil company. The corresponding paper, under the title Blending Aviation Gasolines, was published in 1952 in Econometrica. Curiously enough, they started trying to solve this problem the other way round, i.e., â€œapplying theory to solve applicationsâ€. They resorted to the â€œactivity analysisâ€ due to T.C. Koopmans, awarded later as Nobel Prize in Economy, but they failed.
As a consequence of their success they were invited to participate in other applied proyects. One of them is worth mentioning. It was the proposal of G.H. Simonds, the director of the research team of the EXXON refinery in Elizabeth, New Jersey. As a result a new approach was developed by replacing the stochastic coefficients in a linear programming model by random variables. In this way the data with variability out of control of the project direction could be modeled. The results were published as a 1958 paper in Management Science, coauthored by Charnes, Cooper and Simonds, under the title “Cost Horizons and Certainty Equivalents: An Approach to Stochastic Programming of Heating Oil”. Today we know this new approach as â€œchance constrained programmingâ€, a technique that has been used by many researchers within the stochastic programming literature.
Another relevant example of the â€œcreation of theory based on applicationsâ€ is goal programming, a research branch also created by Charnes and Cooper. For an interesting revision with several examples see their 1977 European Journal of Operational Research article entitled “Goal Programming and Multiple Objective Optimizations.
I have been working with Cooper and within a DEA framework since we met for the second time, in issues related to the additive model. Our first and largest paper was written jointly with Park and is devoted to introducing a new additive based efficiency measure baptized as RAM. It was published in 1999 in JPA, one year later than an application of RAM to the evaluation of water supply services in the Japanese region of Kanto. This second paper was joint work with Aida and Sueyoshi and was published in 1998 in OMEGA. Our sixth and last common paper was coauthored with Aparicio and BorrÃ¡s, devoted to decomposing profit efficiency and published in EJOR in 2011. I would like to close this writing with a basic thought of Bill Cooper. â€œProbably both DEA and goal programming need to be enlarged to develop new models so as to evaluate quality of life. This issue will require resorting both to â€œcreating theory based on applicationsâ€ and to â€œapplying theory to solve applicationsâ€ so that both approaches appear as complementary instead of antagonistic.â€
Jesus T. Pastor, Spain