How to define a list of n tuples, each tuple contains 2 variables

Answered

Fei Wang

• April 23, 2023 21:10
• Edited

I want to define a list of n tuples, each tuple represents a point in a plane, so each tuple contains two variables, what gurobi API should I use? I am using Python 

 For example, if n = 5, then my variables are
(x1, y1), (x2, y2),(x3,y3),(x4,y4), (x5, y5)

I tried x = model.addVars(n, 2, vtype=GRB.CONTINUOUS, name="x")

But then I don't know how to extract a row from x 

Related to

• Modeling

1

• 

  Mario Ruthmair

  Gurobi Staff

  • April 24, 2023 09:06

  Hi,

  As you defined the names of your variables, I would suggest to use two variable dicts, e.g.,

  x = model.addVars(n, vtype=GRB.CONTINUOUS, name="x")
  y = model.addVars(n, vtype=GRB.CONTINUOUS, name="y")

  Would this be uncomfortable to use for you?

  Best regards,
  Mario

  1
• 

  Fei Wang

  • April 24, 2023 19:18

  

  

  

  Hi Mario,

  Thanks for the answer, but I want to generalise it to  define a vector which represent a point in r dimensional space, If I use this approach, I would have to modify the code to define z if I want to define a point in R^3, instead of just changing r in the input. 
  
  The reason I want to extract a row from x is that I want to add some constraints regarding the distances between two points which is a second order cone constraint. 
  For example,  something like d @ d <= t @ t where d =  [x[i] - x[ j], y[i] - y[j], z[i] - y[j],...., ] (there are r component in d), I am not sure how to do this. Instead of x,y,z I should use x_1,x_2,....,x_r?  

  0
• 

  Mario Ruthmair

  Gurobi Staff

  • April 25, 2023 17:55

  It seems you could use the gurobipy Matrix API to handle your vars and constraints. Here is some material:

  • Tutorial video
  • Small example

  Basically, you can add a matrix variable like this:

  x = model.addMVar((n,r), vtype=GRB.CONTINUOUS, name="x")

  Then, you can slice this matrix variable similar to how you would do it in numpy to build your constraints.

  1
• 

  Fei Wang

  • April 27, 2023 19:46

  

  

  

  Thanks! That's the method I am looking for

  0

Please sign in to leave a comment.