Documentation mismatch for range constraints

回答済み

Matthias Walter

2025年08月11日 19:44

• 
• 

According to the documentation, adding a range constraint for L ≤ f(x) ≤ U produces an equation f(x) - s = L for a new continuous nonnegative slack variable s ≤ U-L. However, the following Python code

from gurobipy import *
m = Model("")
x = m.addVar()
y = m.addVar()
m.update()
m.addRange(x + y, 5, 20)
m.write('debug.lp')

yields the following debug.lp file contents:

\ LP format - for model browsing. Use MPS format to capture full model detail. 
Minimize 
 
Subject To 
R0: C0 + C1 + RgR0 = 20 
Bounds 
RgR0 <= 15 
End

So it apparently adds the equation f(x) + s = U. The same problem arises using the C API.

関連

• C

0

• 

  Riley Clement

  • Gurobi Staff

  2025年08月11日 23:16

  Hi Matthias,

  Thank you for reporting this.

  Both are correct implementations, and the value from one slack variable can be derived from another using the equation

  s_1 + s_2 = U - L

  I'm not sure if there are circumstances where sometimes one implementation is used, and in other circumstances the other is used, but I'll check and propose a change to the documentation.

  It's likely that at some point we made a simple change between the implementations and assumed it was an internal detail that wasn't publicly documented.

  Thanks,
  Riley

  0

サインインしてコメントを残してください。