Autism and enthusiasms

The New York Times has an article on “Reaching an Autistic Teenager” describing an approach for teaching autistic kids, who generally are extremely self-absorbed and difficult to reach. The approach, where teachers and students alike work out how the day will go, with students learning to learn and to interact, sounds fantastic: I would have loved to have been part of that growing up.

Perhaps a reason for my enthusiasm comes from the following line in the article:

Children with autism — especially Asperger’s — are famous for all-consuming interests in Match-box cars, bus maps, train schedules, oscillating fans, Civil War battles, baseball statistics, black holes, dinosaurs, chess, or Star Wars.

Hmmm… for me would be “yes, yes, yes, no, yes, yes, yes, yes, yes, and yes”. And perhaps a bonus point for combining two and being fascinated with baseball schedules. You may draw your own conclusions.

Happy Birthday CPLEX

More than 20 years ago, Bob Bixby decided the world needed a better linear programming code. This wasn’t a particularly obvious decision. First, there were already existing linear programming codes. Second, linear programming implementation was not exactly a hot research topic. Third, Bixby was not particularly known in this area. A typical paper for Bixby would be “On Reid’s Characterization of the Ternary Matroids” (Abstract: In this paper the author proves a stronger version of a result of Ralph Reid characterizing the ternary matroids (i.e., the matroids representable over the field of 3 elements, GF(3))…). Hardly the sign of someone about to write a great linear programming solution code!

Fortunately, Bixby is really, really smart, so in short order, Bixby got a very good code together, and began to show the research issues involved in implementing linear programming. And the code got better and better, due in part to an eagerness to collect hard linear programming problems and to continuously improve the code based on them. You can read more about this history in the article Bob wrote for the 50th anniversary edition of Operations Research in 2002 in an article entitled “Solving Real-World Linear Programs: A Decade and More of Progress“.

CPLEX is now owned by ILOG, and ILOG is celebrating the 20th anniversary of CPLEX. They just put out a press release on the birthday. Yours truly has a part in the release, providing a quote, and, less obviously, an application:

United States Postal Service (USPS) has realized more than $10 million in savings to date by leveraging ILOG CPLEX for a strategic transportation management initiative. Developed by USPS and IBM, the Highway Corridor Analytic Program (HCAP) uses advanced technology to analyze USPS highway transportation scenarios and identify cost saving opportunities. HCAP model uses ILOG CPLEX to identify the best allocation of mail among transportation resources.

That is work I did with USPS, IBM Global Business Services, and Luis Zuluaga, now at the University of New Brunswick.

Michel Balinski IFORS Distinguished Lecture

The IFORS Distinguished Lecturer for the INFORMS meeting was Michel Balinski of Ecole Polytechnique and CNRS, Paris. Michel spoke on “One-Vote, One-Value: The Majority Judgement”, a topic close to my heart. In the talk, Michel began by discussing the pitfalls of standard voting (manipulation, “unfair” winners, and so on). He then spent most of his talk on a method he proposes for generating rankings and winners. For an election on many candidates (or a ranking of many gymnasts, or an evaluation of many wines: the applications are endless), have the electors (judges, etc.) rate each candidate on a scale using terms that are commonly understood. So a candidate for president might be “Excellent, Very Good, Good, Acceptable, Reject”. Then, the evaluation of a candidate is simply the median evaluation of the electors. The use of median is critical: this limits the amount of manipulation a voter can do. If I like a candidate, there is limited effect if I greatly overstate my liking: it cannot change the overall evaluation unless my evaluation is already under that of the median voter.

Michel then went on and discussed some tiebreaking rules (to handle the case that two or more candidates are, say “Very Good” and none “Excellent”). I found the tie-breaking rules less immediately appealing, but I need to think about these more.

Michel had done an experiment on this by asking INFORMS participants to do an evaluation of possible US Presidential candidates (not just Obama and McCain, but also Clinton, Powell, and a number of others). The result (on a small 129 voter sample) put Obama well ahead, but I do suspect some selection bias at work.

This work will be the basis of a book to be published at the end of the year, and there is a patent pending on the voting system (which I found a little strange: what would it mean to use a patented voting system?).

I didn’t get the URLs at the end of the talk.  If anyone got them, can you email me with them?  A quick web search only confused me more.

Thanks Ashutosh for this pointer.

Added Oct 20. Michel Balinski kindly wrote and provided the following references:

Michel Balinski and Rida Laraki, “Le jugement majoritaire : l’expérience d’Orsay,” Commentaire no. 118, été 2007, pp. 413-419.

One-Value, One-Vote: Measuring, Electing, and Ranking (tentative title), to appear 2009.

http://ceco.polytechnique.fr/jugement-majoritaire.html

Michel Balinski et Rida Laraki, A theory of measuring, electing and ranking,
Proceeding of the National Academy of Sciences USA, May 22, 2007, vol. 104, no. 21, pp. 8720-8725.

Michel Balinski et Rida Laraki, “Election by Majority Judgement: Experimental Evidence.”
Cahier du Laboratoire d’Econométrie de l’Ecole Polytechnique, December 2007, n° 2007-28

Operations Research as a path to academic administration

For most young researchers, administration is a word filled with horror.  Why would anyone want to be an academic administrator when you could spend your days exploring the wonder of operations research?  And many days, I (a not so young researcher) agree with them.  There is almost nothing better than spending time thinking, writing code, teaching, and doing all the wonderful things that make up the academic life.  However, after a decade or two, for some there comes a wish to have a broader effect.  Will yet one more paper on a better cutting plane or another game-theoretic analysis of a supply chain really affect very much?  Or perhaps after sitting through yet another unproductive, useless meeting, the thought comes “I can do better than these jackasses!”.  And the administrative path begins.

Some of the people I admire most in our field have done significant academic administration.  For instance, Patrick Harker was the editor of Operations Research when he was picked to be Dean of Wharton, one of the most important business school deanships around.  He is now President of the University of Delaware.

I think operations research is actually pretty good training for administration.  In OR, you learn to make decisions based on facts and data, rather than biases and preconceptions.  Within the business school, we often get to know many of the faculty as our methods can be used broadly (and are often considered “honest brokers” in conflicts between the bigger areas of finance and marketing).

This trend has grown strong enough that there is even a “Dean’s Meeting” at INFORMS.  From the excellent conference daily news:

Operations research trains a professional to become a better thinker, problem solver, and educator. As more and more members of INFORMS are discovering, the field also trains you to become a better university administrator.

Dozens of operations researchers across the country are becoming deans and provosts – and even presidents – of universities large and small, in the U.S. and throughout the world. Some examples: Prof. Arjang Assad, formerly of the Robert H. Smith School at the University of Maryland just became the dean of the University of Buffalo School of Management. Also this year, James Bean, the former president of INFORMS, became Senior Vice President and Provost of the University of Oregon.

I hadn’t realized that Jim Bean had moved up to Provost at Oregon.  Congratulations Jim!

Greetings from Washington

I am at the Washington INFORMS and have just come back from the first session (on experiences with COIN-OR).  Sixty-eight parallel tracks!  Looks like there might be 5000 people at this conference.  It is amazing how big the conference has become.  I was on the INFORMS Board around 2000 when the decision was made to go from two conferences/year (a holdover from the ORSA/TIMS and TIMS/ORSA conferences before the societies merged in 1994) to just one.  At the time, we were getting 1500-1800 people per conference.  The discussion was “Maybe we could get 2400 people if we ran just one conference”.  No one was bold enough to suggest our yearly count would reach 3000-3600, let alone grow enormously!  But INFORMS is the one place to go where you know you will see most or all of your colleagues every year.  While I prefer smaller conferences, this is still a must-go-to event, simply for the mix of people to see.

Weather in Washington is beautiful and the conference hotel, while huge, seems to work, so this is shaping up to be a great conference.

On the blogging front, the ICS (INFORMS Computing Society) blog has posted from the conference, and I expect others to appear shortly.  If you are at the conference, don’t forget to drop by the COIN-OR booth (Booth 504) this evening (Sunday) around 8 to say hi to the blog-ORs!  And stop off to see Steve Baker, author of the Numerati.

Gurobi goes live

I had previously written about the exit of Gu, Rothberg, and Bixby from Ilog. I was curious where they would end up: Google, IBM, Yahoo? Somehow, I thought they would be together, but where? That has now been answered. Gurobi Optimization has gone live today. From their page:

Gurobi Optimization is in the business of providing robust, high-performance optimization software based upon the latest linear and mixed-integer programming technologies.

The computational progress in linear and mixed integer programming over the last twenty years has been nothing short of remarkable, enabling business, scientific and other applications that literally would have been unapproachable just a few short years ago. The Gurobi founders have been at the forefront of these developments.

The Gurobi suite of optimization products represent completely new implementations, redesigned from the ground up to fully exploit the latest mathematical and “engineering” improvements in the underlying methodologies, as well as developments in modern desktop computing hardware and programming environments; moreover, the Gurobi team is committed not just to providing the best technology now available, but to continuing to push forward the frontier of optimization solution capability.

Focused not just on the technology, Gurobi will also be introducing several new, more flexible delivery models for optimization technology, models that we believe will better serve our customers and the market in general.

I have to say that this makes me very happy. I really like CPLEX (IBM) and I really like Dash XPRESS (FairIsaac) But I really, really, really like groups of very smart, ambitious people fighting it out to see what really can be done in the field of optimization. This is great news for our field and I look forward very much to next year’s release.

added October 13. I have been asked to note that CPLEX is sold by ILOG, not IBM.  You can see some past posts on the proposed acquistion of ILOG by IBM, but that has not yet been approved by the powers that be.

Update on the blog-or-sphere get-together

I have confirmation from bloggers Laura McLay (Punk Rock Operations Research), Aurelie Thiele (Thoughts on Business, Engineering, and Higher Education), and Bill Hart (William E. Hart’s Blog) that they will be at booth 504 (the Coin-OR booth) at 8PM on Sunday at the INFORMS DC meeting (or at least will try hard to do so). If you are blogger and will be there, let me know so I can add you to the list. And if you are an adoring (or not) fan, be sure to drop by and say hi!

Update Oct 10 3PM. Ian Frommer (Green OR) will be at the Blog-OR gettogether.  Five bloggers and counting!

Meet at INFORMS?

I had meant to do something a little more formal for the blogging world at INFORMS DC, but until recently it was unclear whether I could make it to the conference (I am teaching now, and with a final exam coming up, I wanted to be sure my classes were ready). But I will be at INFORMS, and would like to meet up with the rest of the people either blogging or reading the blogs. Given the late date, why don’t we plan on an informal get-together during the Sunday reception. I will be by the “COIN-OR” booth at the Sunday reception at around 8PM. It is booth 504. It would be great to see you there!

New blog and new journal

A new blog by Bill Hart of Sandia National Labs reminds me that there is also an exciting new journal about to begin.  From Bill’s blog:

I have recently joined the editorial board of the new journal Mathematical Programming Computation, which publishes original research articles that are at the intersection of math programming and computing. This journal reflects the growing role of computation in operations research, where real-world applications often require the application of complex software packages to analyze mathematical models.

I too am involved in the new journal, as one of the members of the advisory board.  If you have some computational work looking for a good home, be sure to consider this journal.  With Bill Cook of Georgia Tech as editor, I am expecting it to be a great success.