[R] optimization: multiple assignment problem

[Hans W.Borchers]

Jean-Francois Chevalier <Jean-Francois.Chevalier <at> bisnode.com> writes:
> 

You have already given the answer yourself. You have binary variables x(j, i),
you need to set up the inequalities, and then apply one of the mixed-integer
linear programming solvers in R, for instance 'lpSolve', 'Rglpk', 'Rsymphony'.

Setting up the inequalities may be slightly involved. You have not provided
reproducible code, so no concrete answer possible.

Hans Werner

> Hello,
> I'm trying to solve a multiple assignment problem.
> I found a package Adagio and its function mknapsack which
> maximize vstar = p(1)*(x(1,1) + ... + x(m,1)) + ... ... + 
>                  p(n)*(x(1,n) + ... + x(m,n))
> subject to w(1)*x(i,1) + ... + w(n)*x(i,n) <= k(i) for i=1,...,m
> x(1,j) + ... + x(m,j) <= 1 for j=1,...,n x(i,j) = 0 or 1 
>     for i=1,...,m , j=1,...,n ,
> It's close to what I'm trying to do except that
> 1)k(i) = k for any I (not an issue)
> 2)p is dependent of the item AND the knapsack
> 3)each item must be assigned
> maximize vstar = p(1,1)*x(1,1) + ... + p(m,1)*x(m,1) + ... ... + 
>                             p(1,n)*x(1,n) + ... + p(m,n)*x(m,n)
> with p(j,i) profit of assigning item i to knapsack j
> subject to w(1)*x(i,1) + ... + w(n)*x(i,n) <= k for i=1,...,m
> x(1,j) + ... + x(m,j) = 1 for j=1,...,n x(i,j) = 0 or 1 
>      for i=1,...,m , j=1,...,n ,
> It would be really helpful if you could indicate me any package, 
> function that would solve my problem?
> Thanks in advance,
> Best regards,
> Jean-François