Potential Numerical Issue in MIP Solving

Answered

Xintong Zhou

January 26, 2026 22:03

• 
• 
• 

Hi,

I am working on this small instance:

Minimize
OBJ: 162.98 x0 + 441.38 x1 - 419.67 x2 - 219.21 x3
Subject To
c0: 71.79 x0 + 32.46 x1 + 98.05 x2 - 2.8 x3 >= -342.3374
c1: - 0.35 x0 + 6.47 x1 - 7.13 x2 - 33.29 x3 >= -445.57
c2: - 68.64 x0 + 68.14 x1 + 45.6 x2 - 67.49 x3 <= -159.87
Bounds
 -100 <= x0 <= 100
 -100 <= x1 <= 100
 -100 <= x2 <= 100
 -100 <= x3 <= 100
Generals
x0 x2
End

Gurobi returns the following solution with obj = -80372.02..:

x0 = -4.0
x1 = -100.0
x2 = 99.0
x3 = -27.212376088915594

While other major solvers (including scip, highs, cbc) return a better solution with obj = -80579.46:

x0 = -3.0
x1 = -100.0
x2 = 100.0
x3 = -27.4370681886452

I am not sure if this is a possible tolerance issue or a bug. Any suggestions are appreciated.

Gurobi version:

Gurobi Optimizer version 13.0.0 build v13.0.0rc1 (linux64 - "Ubuntu 22.04.5 LTS")
Copyright (c) 2025, Gurobi Optimization, LLC

 

0

• 

  Maliheh Aramon

  • Gurobi Staff

  January 26, 2026 22:34

  Hi, 

  Thank you for reporting this to us. It seems it is a bug. I can replicate it with version 13.0.0, but versions 12.0.3 and 11.0.3 gave -80579.5 as the optimal solution. I will transfer this post to a ticket so that we can investigate it further. 

  Best regards,

  Maliheh

   

   

  0

Please sign in to leave a comment.