General expression : min_ or max_

Answered

Sung Ook Hwang

• October 29, 2020 20:23

Hi,

 

I wanted to use min_ expression in the rhs of constraint (please see the below expression, schedule[i] is decision variable).

model.addConstr(quicksum(100*schedule[i] for i in index) <= 100 - min_(100, quicksum(100 for i in index if i.id==0)))

 

However I had TypeError which is unsupported operand type for -: 'int and 'GenExpr'.

 

Thus, I changed the above expression as below.

model.addConstr(quicksum(-100 - 100*schedule[i] for i in index) >= min_(100, quicksum(100 for i in index if i.id==0)))

Now I have GurobiError that General expressions can only be used for equality constraint.

Would you please help me how I can solve this issue?

Thanks,

 

0

• 

  Eli Towle

  Gurobi Staff

  • October 30, 2020 00:04
  • Edited

  The min_() helper function can only be used to set one variable equal to a group of variables and/or constants. This would be used if, e.g., you want to add a constraint like \( x = \min\{y, z, 2\} \), where \(x\), \(y\), and \(z\) are variables.

  In your case, the elements of \( \texttt{index} \) are not variables. Thus, the minimum of \( \texttt{100} \) and \( \texttt{sum(100 for i in index if i.id == 0)} \) is a constant value that doesn't depend on any variables. This means you can use Python's built-in \( \texttt{min()} \) function:

  model.addConstr(quicksum(100*schedule[i] for i in index) <= 100 - min(100, sum(100 for i in index if i.id == 0)))

  Note that I used Python's built-in \( \texttt{sum()} \) function instead of Gurobi's quicksum() function. The latter returns a LinExpr object, which won't work well with \( \texttt{min()} \).

  0
• 

  Sung Ook Hwang

  • October 30, 2020 03:47

  

  

  

  Thank you Eli for your help!

  0

Please sign in to leave a comment.