confusion about adding MVar.fromlist()

Answered

Thomas Leveringhaus

• August 10, 2023 12:50

Dear Gurobi-Team,

I tried to use MVar.fromlist() (https://www.gurobi.com/documentation/10.0/refman/py_mvar_fromlist.html) but the given example does not work:

import gurobipy as gp
model = gp.Model("model")
x0 = model.addVar()
x1 = model.addVar()
x2 = model.addVar()
x3 = model.addVar()
x_1d = MVar.fromlist([x0, x1, x2, x3])  # 1-D MVar
x_2d = MVar.fromlist([[x0, x1], [x2, x3]])  # 2-D MVar

I use Spyder and it already warns before runnig the script, that MVar is undefined. That's obviously true, so I added a "dummy MVar":

import gurobipy as gp
model = gp.Model("model")
x0 = model.addVar()
x1 = model.addVar()
x2 = model.addVar()
x3 = model.addVar()
MVar = model.addMVar(0)
x_1d = MVar.fromlist([x0, x1, x2, x3])  # 1-D MVar
x_2d = MVar.fromlist([[x0, x1], [x2, x3]])  # 2-D MVar

That seems to work, but is that the intended way of use?

Is it a problem, that the dummy MVar has a shape of zero and the following MVars built on it have different shapes?

Does the dummy need to remain in the model, or can I delete it after declaring the really wanted MVars?

Can I reuse the dummy as often as I want, like I did before, or do I need a dummy for each MVar in the model, as in the following?

import gurobipy as gp
model = gp.Model("model")
x0 = model.addVar()
x1 = model.addVar()
x2 = model.addVar()
x3 = model.addVar()
MVar1 = model.addMVar(0)
x_1d = MVar1.fromlist([x0, x1, x2, x3])  # 1-D MVar
MVar2 = model.addMVar(0)
x_2d = MVar2.fromlist([[x0, x1], [x2, x3]])  # 2-D MVar

Best regards

Thomas

Related to

• Matrix API

0

• 

  Riley Clement

  Gurobi Staff

  • August 10, 2023 13:00

  Hi Thomas,

  Our example is incomplete in the sense that it doesn't show the imports.  It's working on the assumption that the following import exists:

  from gurobipy import *

  You can make it work in your example like this:

  x_1d = gp.MVar.fromlist([x0, x1, x2, x3])  # 1-D MVar
  x_2d = gp.MVar.fromlist([[x0, x1], [x2, x3]])  # 2-D MVar

  - Riley

  0
• 

  Thomas Leveringhaus

  • August 10, 2023 13:09

  

  

  

  Thanks, that works :-)

  0

Please sign in to leave a comment.