On 25/10/2013 18:12, Paul Cannongeyser wrote: > Trying to code 0-1 Multiple Knapsack Problem. > > I ran this example. My results don't agree with the answer in the book. ... > % From http://www.or.deis.unibo.it/kp/Chapter6.pdf entitled 0-1 Multiple knapsack problem, Example 6.2 ... > I now realize that I must add constraints that say that X1 and Y1 cannot both be 1 at the same time, > same for X2 and Y2, same for X3 and Y3, etc., in other words, the character packing the two knapsacks > has one of each item and can put it into the first knapsack or the second knapsack or leave it behind. Correct. Whenever you get a solution that is better than the one you expect, you are likely to have forgotten a constraint, in this case constraint (6.3) on page 157. > > I suppose I could do it this way: > > X1 + Y1 $=< 1, > X2 + Y2 $=< 1, > X3 + Y3 $=< 1, etc. > ... > Is this the correct (or generally accepted) approach? Yes, this is how you would model it with LP/MIP constraints. However, to save yourself a lot of typing, you should use an array for modelling, rather than working with individual variables. Then you have X[1,1] for X1 up to X[2,10] for Y10, and you can use loops to set up the constraints. -- JoachimReceived on Sun Oct 27 2013 - 01:13:03 CEST
This archive was generated by hypermail 2.2.0 : Tue Oct 29 2013 - 18:13:26 CET