A Personal History of ICS

By Harvey J. Greenberg

Let me set the stage. The OR/CS interface has its own history stemming from the 1950s. It was then that Martin Beale and Bill Orchard-Hays introduced sparse matrix techniques and associated matrix management. Eventually (about 1970), the numerical analysts latched on, and they were mostly within what was then the emerging computer science community. (There were no academic departments of computer science until the 1960s, and it was George Forsythe, a numerical analyst, who coined the term "computer science.")
IBM used linear programming to debug its model 7090, and Bill wrote Control Data Corporation's assembler (compass) so he could complete his linear programming system on schedule. Queueing theory and scheduling were also in their early stage of development and part of the O.R. foundations. Computer scientists used these to develop heuristics in designing operating systems.

In the mid-1970s, expert systems became the hype, and I remember hearing a few people talk at an ORSA meeting about the artificial intelligence/operations research (AI/OR) interfaces. I saw it as ironic because AI had such leaders as Herbert Simon, not only to become a Nobel Laureate in Economics, but also a recipient of the ORSA John von Neumann Prize for Theory. Seeing classical O.R. problems like scheduling approached in a "new" way struck me as some loss of history. It is perhaps reflective that the AI community does not now consider itself a subset of computer science, as it was then. Things evolve, and some interfaces get lost.

One of the most important contributions to the early development of the OR/CS interface (and to CS more generally) was a rigorous standard for numerical testing of algorithms. In 1973, Ric Jackson, John Tomlin, Larry Haverly and others met at Stanford's newly created Stanford Optimization Laboratory, which focused on computations. That meeting set a ball in motion that resulted in the first set of real test problems and guidelines.

About the same time there was a healthy competition between two groups: Californians Gordon Bradley, Jerry Brown and Glenn Graves and Texans Fred Glover, Darwin Klingman and several students (notably, Dick Barr and Dave Karney). They were each presenting network solvers that not only implemented algorithms smartly, but also applied advanced data structures that made a major difference. (They pushed speeds to where network flows could be solved about 100 times faster than using a commercial LP solver!) In the process, they raised the standards for numerical testing and comparative analysis. That influenced not only the O.R. community, but also a wider computer science community.

The OR/CS interface was alive and productive in the 1970s, but ACM began to distance itself from O.R. They had a special interest group in Mathematical Programming (SIGMAP), but that was eventually de-chartered (circa 1980). ACM also took simulation as its own, which was not only a loss to the O.R. identity, but also to the scope of simulation research. Further detachments in interfaces like algorithm design and analysis would add confusion to notation and related concepts. Perhaps one of the people who personified the OR/CS interface was Richard Karp. He received both the ACM Turing Award and the ORSA John von Neumann Prize for Theory — the only one to receive top awards in both computer science and O.R.

Building its own identity, ACM continued to distance itself from organizational units that explicitly embraced the OR/CS interface. As evidence of its continued dissociation with O.R., the ACM once had a department (what we call an "area") of operations research that is now gone. In summary, the CS side of the stage is that the premier society, ACM, was releasing ties with O.R., creating a void for a community who worked in the interfaces.

On the O.R. side, this was long before the merger of ORSA and TIMS. (In fact, ORSA just started to combine its meetings with TIMS in the mid-1970s.) ORSA had a few technical sections (TS), and they wanted to allow orderly growth. So, in response to our efforts to become a TS, they invented a two-tier system: begin as a SIG, then when you meet our criteria, you can petition to become a TS. As I recall, the main criterion was having at least 300 members for more than one year. A SIG could charge dues, hold special conferences and meet at the semi-annual ORSA-TIMS meeting. There were no clusters run by SIGs or TSs yet, so having some influence about sessions in the OR/CS interface was a big deal. A SIG could not carry over any money raised from dues or conferences from one year to the next. Also, a SIG could not have its own journal; that was a privilege reserved for a technical section.

As ACM distanced itself from O.R., ORSA embraced the interface. Even before the computer science SIG was official, we saw sessions at meetings, and in 1978, Operations Research published a special issue on the OR/CS interfaces (Vol. 26, No. 5), showing a significant range of interfaces that reflects the vision we all had. So, we became the CS SIG, with Sam Gorenstein elected by acclamation to be our first chair.

I became the second chair in 1977, followed by Gordon Bradley in 1978. Gordon introduced our newsletter, with much help from Jerry Brown.

During the 1970s, our membership grew to more than 500. A group of us, which included Gordon Bradley, Jerry Brown, Karla Hoffman, Ric Jackson Dick Nance and Dick O'Neill, prepared the proposal for the CS SIG to become a technical section, and CSTS was born officially in 1980, near the end of Ric Jackson's term as chair.

Working with the Mathematical Programming Society, Ric formed the Committee on Algorithms (COAL), and CS SIG members such as John Tomlin contributed to the first set of O.R. test problems and guidelines for testing algorithm implementations. During this early period, Ric recalls inspirational discussions with Darwin Klingman. Although Darwin did not serve as chair, he was a great supporter of CSTS. He received the TIMS Franz Edelman Award for an outstanding application of information technology, disciplined by his training in O.R.

Dick O'Neill became chair of CSTS in 1980, at which point we had 600 members from diverse areas of the OR/CS interface. His successor, Karla Hoffman, recalls productive discussions with John Mulvey who, like Darwin, did not hold an office, but was a great supporter. Our membership was holding stable at about 600. Jerry Brown succeeded Karla, and he introduced wine and cheese at our meetings. This was nothing at the time, as the ORSA establishment did not wholeheartedly approve. Of course, it soon became ubiquitous, which we all enjoy today.

In 1983, Joe Graves was chair. His successor, Phyllis Martin, served two terms as chair of CSTS, and I succeeded her (for a second term). I chaired a committee to plan the first CSTS Symposium: "Impacts of Microcomputers on Operations Research." (The proceedings were published by North-Holland, part of Elsevier Science, as Vol. 5 in the ORSA Publications in Operations Research Series, edited by Saul Gass.) The committee was composed of Saul Gass, Karla Hoffman, Warren Langley and me. It was held at the University of Colorado at Denver in 1985, and I believe that the proceedings papers are still insightful today. This evolved into what is now the ICS biannual symposium. The subsequent three symposia were held in Williamsburg, Va., at the suggestion of Bill Stewart, who handled local arrangements. It was perhaps an omen of our continued success that the weather was sunny at the first of these — most unusual for January.

In 1986, while Don Kraft was chair-elect of CSTS, we created the CSTS Prize for Research in the interfaces of O.R. & CS. I was fortunate to be its first recipient for my development of the ANALYZE software system, which provided an artificially intelligent environment for analyses of linear and integer programs.

I organized a committee to build the case for what became the ORSA Journal on Computing (now the INFORMS Journal on Computing). I was fortunate to have great help from the committee: Karla Hoffman, Bob Jeroslow, Don Kraft and Bill Pierskalla. They were a diverse group of outstanding researchers who provided important advice as we wrote the proposal to the ORSA Publications Committee. Meanwhile, Kluwer offered to publish the journal (without any further proposal writing), but I wanted the choice of publisher left to the CSTS membership at the next ORSA-TIMS meeting.

In 1987, under John Tomlin's leadership as the CSTS chair-elect, the attending members chose to have it be an ORSA journal, rather than be private (perhaps affiliated, like OR Letters). If the ORSA Council had chosen not to approve the journal at that meeting, we would have gone private. Thanks to Carl Harris, Tom Magnanti and Bill Pierskalla, ORSA did approve the inauguration of the JoC in 1987. I was its founding editor, and we published the first issue in 1989. I left the editorship in 1992, while Ramesh Sharda was still the CSTS chair, and Bruce Golden became the JoC's second editor. The third (and current) editor is David Kelton. Ramesh initiated the Kluwer OR/CS Interface Series. The first issue, published in 1993, was "A Computer-Assisted Analysis System for Mathematical Programming Models and Solutions: A User's Guide for ANALYZE," by Harvey J. Greenberg.

In 1989, Milt Gutterman succeeded John Tomlin as the CSTS chair. The years 1988-92 were devoted to building a greater membership base, but I focused on the JoC, so I cannot recall the details. Although I cannot remember everyone who led the CSTS during the 1990s, I think I have correctly identified the chairs. Membership peaked at more than 900 in 1996, and then dropped to about 600 in 1998. Here is where my memory fades because I was not as active as I had been. I do know that CSTS became ICS in 1998, at the very end of Dick Barr's term as chair. Dick was the architect of the petition and presented it to Council just before he handed over the gavel to Harlan Crowder.

I hope someone else adds to this personal history and that ICS keeps an ongoing archive. I probably missed some people who served well between 1976 and 1998, and I apologize to them for that omission.

In conclusion, there have been recent diversions from what I still think is an important, viable interface to pursue. I believe it is, and always has been, the ICS mission to articulate and lead the development of interfaces between operations research and computer science. It is not just a fact of history, but a matter of necessity, that these communities interact. O.R. has much to bring to the table of such new fields as computational biology, and we have seen too many opportunities slip away out of neglect. I hope current and future ICS members agree with this mission statement and that its leaders will actively pursue its goals.

I emphasize that from the beginning, we saw the interfaces broadly — applying O.R. methods like queueing and scheduling to operating system design; applying CS methods, like database theory and information structures, to O.R. problems; and developing special foundations and tools, like intelligent systems and simulation.






ICS Timeline



1974



Jerry Brown, Milt Gutterman and others begin talking about a special interest group





1975



Sam Gorenstein organizes meeting at Las Vegas ORSA-TIMS





1976



CS SIG born, Sam serves as chair





1977



Harvey Greenberg serves as chair





1978



Gordon Bradley serves as chair; newsletter born (edited by Gordon with help from Jerry Brown)





1979



Ric Jackson serves as chair; CSTS born





1980



Dick O'Neill serves as chair





1981



Karla Hoffman serves as chair





1982



Jerry Brown serves as chair; wine & cheese tradition starts





1983



Joseph Graves serves as chair





1984



Phyllis Martin serves as chair





1985



Phyllis Martin serves as chair (second term); first CSTS Symposium held at University of Colorado at Denver; organized by Saul Gass, Harvey Greenberg (chair), Karla Hoffman and Warren Langley





1986



Harvey Greenberg serves as chair (second term); first CSTS Prize awarded (to Harvey Greenberg)





1987



Don Kraft serves as chair; ORSA Journal on Computing born (Harvey Greenberg first editor) based on proposal developed by Harvey Greenberg (Chair), Karla Hoffman, Bob Jeroslow, Don Kraft and Bill Pierskalla





1988



John Tomlin serves as chair





1989



Milton Gutterman serves as chair; first issue of ORSA Journal on Computing is published; second CSTS Symposium held at Williamsburg, Va., organized by Ramesh Sharda (chair), Bruce Golden, Ed Wasil, Osman Balci and Bill Stewart





1990



Milton Gutterman serves as chair (second term)





1991



Ramesh Sharda serves as chair





1992



Ed Wasil serves as chair; Bruce Golden becomes second JoC editor; third CSTS Symposium held at William and Mary College, organized by Ramesh Sharda, Osman Balci (chair) and Stavros Zenios





1993



Osman Balci serves as chair; Kluwer OR/CS Interface Series inaugurated, edited by Ramesh Sharda. First issue is "A Computer-Assisted Analysis System for Mathematical Programming Models and Solutions: A User's Guide for ANALYZE" by Harvey J. Greenberg





1994



Bill Stewart serves as chair; fourth CSTS Symposium held at Williamsburg, Va., organized by Stephen Nash and Ariela Sofer





1995



John Hooker serves as chair; CSTS merges with ORSA AI section (new bylaws written)





1996



Robert Fourer serves as chair; fifth CSTS Symposium held at Southern Methodist University, organized by R. Barr (chair), R. Helgason and J. Kennington





1997



Richard Barr serves as chair; ICS born





1998



Harlan Crowder serves as chair













Published Volumes from CSTS Meetings


S.I. Gass, H.J. Greenberg, K.L. Hoffman, and R.W. Langley (eds.), "Impacts of Microcomputers on Operations Research," Elsevier Science Publishing Co., New York, N.Y., 1986.

R. Sharda, B. Golden, E. Wasil, O. Balci, and W. Stewart (eds.), "Impacts of Recent Computer Advances on Operations Research," Elsevier Science Publishing Co., New York, N.Y., 1989.

R. Sharda, O. Balci, and S. Zenios (eds.), "Computer Science and Operations Research: New Developments in their Interfaces," Pergammon Press, Oxford, U.K., 1992.

S.G. Nash and A. Sofer (eds.), "The Impact of Emerging Technologies on Computer Science and Operations Research," Kluwer Academic Press, Norwell, Mass., 1995.

R. Barr, R. Helgason, and J. Kennington (eds.), "Interfaces in Computer Science and Operations Research: Advances in Metaheuristics, Optimization, and Stochastic Modeling Technologies," Kluwer Academic Press, Norwell, Mass., 1997.

D.L. Woodruff (ed.), Advances in Computational and Stochastic Optimization, Logic Programming, and Heuristic Search Interfaces in Computer Science and Operations Research, Kluwer Academic Press, Norwell, Mass., 1998.










CSTS Prize Awards



1986



Harvey J. Greenberg for "A Fundamental Description of ANALYZE: A Computer-Assisted Analysis System for Linear Programming Models," ACM Transactions on Mathematical Programming, Vol. 9, No. 1, 1983, pp. 18-56.





1987



Alex Meeraus for his development and applications of the Generalized Algebraic Modeling Language (GAMS).





1989



Fred Glover and Darwin Klingman for results in "Layering Strategies for Creating Exploitable Structure in Linear and Integer Programs," Mathematical Programming, Vol. 40, No. 2 (1988), pp. 165-82. and to Marcel Neuts for his seminal works in computational probability, exemplified by his paper, "Computer Experimentation in Applied Probability," Journal of Applied Probability, Vol. 25A (Special Volume: A Celebration of Applied Probability, J. Gani, ed.), (1988), pp. 31-43.





1991



John N. Hooker for "Input proofs and rank one cutting planes," ORSA Journal on Computing, Vol. 1 (1989), pp. 137-145.





1992



Irvin J. Lustig, Roy E. Marsten and David F. Shanno for results in "Computational Experience with a Primal-Dual Interior Point Method for Linear Programming," Linear Algebra and its Applications, Vol. 152 (1991), pp. 191-222."





1993



Robert Fourer, David M. Gay and Brian W. Kernighan for their cumulative work, spanning over a decade, in the theory and practice of mathematical programming modeling languages."





1994



Fred Glover for development and extension of the tabu search metaheuristic.





1995



John Forrest and Donald Goldfarb for "Steepest-Edge Simplex Algorithms for Linear Programming," Mathematical Programming, Vol. 57, No. 3 (1993), pp. 341-374.





1996



Warren Adams and Hanif Sherali for their research on the Reformulation-Linearization Technique as documented in a series of articles dating from 1986.





1997



Dmitri P. Bertsekas and John N. Tsitsiklis for their book "Neuro-Dynamic Programming" and the research behind it.





1998



Ding-Zhu Du and Frank K. Hwang for "A proof of Gilbert-Pollack's conjecture on the Steiner ratio," Algorithmica, Vol. 7 (1992), pp. 121-135.












Harvey J. Greenberg (harvey.greenberg@cudenver.edu) is a professor of Mathematical Sciences with the Department of Mathematical Sciences, University of Colorado at Denver, where he is also an adjunct professor of Computer Science & Engineering and an adjunct professor of Preventive Medicine & Biometrics.






Acknowledgments and Dedications


I received help with this historical account from Dick Barr, Jerry Brown, Harlan Crowder, Bob Fourer, Saul Gass, Karla Hoffman, John Hooker, Ric Jackson, Don Kraft, Phyllis Martin, Dick Nance, Dick O'Neill, Bill Pierskalla, Ramesh Sharda, Bill Stewart and John Tomlin, and I received comments from Fred Murphy that made the writing clearer. I also thank Peter Horner (OR/MS Today) and Mary Magrogan (INFORMS), who rummaged through early records to help fill in the missing pieces. Of course, I remain responsible for the accuracy and completeness.

This article is dedicated to the memories of Samuel Gorenstein (1923-2005), Milton Gutterman (1929-2005), Carl Harris (1940-2000), Robert Jeroslow (1942-1988) and Darwin Klingman (1944-1989).