A tweet from Marco Luebbecke contained a provocative quote from George Nemhauser’s plenary talk at the EURO-INFORMS Joint International Meeting:
This needs more discussion than 140 characters. Integer programming is truly an amazing tool for optimally solving problems that are theoretically difficult to solve. It would be hard for me to come up with reasons for why IP would not be the most important tool. What do you think?
July 3rd, 2013 at 9:10 am
Cannot agree more
July 3rd, 2013 at 12:15 pm
There was considerable support for IP on twitter. See my tweets here and here for some of these.
July 3rd, 2013 at 12:54 pm
The catch here is all to do with problem structuring. Too often I see mixed IP/ IP thrown in a cavalier and often redundant fashion at very lazily understood problems.
Take the clock back to when computers were programmed by punch cards and located hours away from your office and the emphasis on problem structuring and economy of solution takes on its rightful significance.
July 3rd, 2013 at 2:56 pm
I couldn’t agree more!
The thing about integer programs is not only do they show up in a myriad of real-world application domains that require integer variables, but they also enable us to model decision yes/no binary variables. That allows us to mathematically make some serious “what to choose?” decisions
July 3rd, 2013 at 4:41 pm
As an IP guy, I want to agree. On the other hand, it would be interesting to compare the number of “important” (to the end user) IP models for which no combination of LP plus rounding, heuristics our dynamic programming produces a satisfactory result, versus the number of “important” queuing problems that would be unsolvable without simulation.
July 6th, 2013 at 6:42 am
The main tool is our brain. Using it wisely makes the need of other tools less important.