clustering and SOS constraints

Answered

Permanently deleted user

• December 03, 2021 19:07

Consider a clustering problem; grouping N points into K clusters:

x = m.addVars(N, K, vtype=GRB.BINARY)

 

Should I use this constraint

m.addConstrs((x.sum(i, '*') == 1 for i in range(N)))

or the SOS1 constraint?  

 

Thanks.

0

• 

  Jaromił Najman

  Gurobi Staff

  • December 06, 2021 09:50
  • Edited

  It depends.

  Do you want that each point is in exactly 1 cluster? Then go for the equality constraint.

  Do you want that each point is in at most 1 cluster? Then go for the SOS1 constraint.

  Best regards,
  Jaromił

  0
• 

  Permanently deleted user

  • December 06, 2021 11:52

  

  

  Each point must end up exactly in 1 cluster. Equality constraint it is then.  

  Thank you!

  Dudley

  0

Please sign in to leave a comment.