Model not feasible while solution does exists

Answered

Huib Meulenbelt

November 11, 2022 12:42

• 
• 
• 

Hi,

I create model as follows

def create_model() -> gurobi.Model:
    m = gurobi.Model()
    m.Params.LogToConsole = 1
    m.Params.PoolSearchMode = 1
    m.Params.PoolSolutions = 1
    m.Params.PoolGap = 0
    m.Params.NonConvex = 2

    return m

Next, 

I introduce a number of variables and constraints:

proj_matrix = [[-0.47686108766245655, 0.8789748939150358, -0.002576614140793262], 
               [-0.8789639728247254, -0.4768680170660497, -0.00438506278966113], 
               [-0.005083064976417589, 0.00017368519028203956, 0.9999870660583067]]

shot_x = m.addVar(vtype=gurobi.GRB.CONTINUOUS, name='shot_x')
shot_y = m.addVar(vtype=gurobi.GRB.CONTINUOUS, name='shot_y')
shot_z = m.addVar(vtype=gurobi.GRB.CONTINUOUS, name='shot_z')

condition_x = shot_x * proj_matrix[0][0] + shot_y * proj_matrix[1][0] + shot_z * proj_matrix[2][0]
condition_y = shot_x * proj_matrix[0][1] + shot_y * proj_matrix[1][1] + shot_z * proj_matrix[2][1]
condition_z = shot_x * proj_matrix[0][2] + shot_y * proj_matrix[1][2] + shot_z * proj_matrix[2][2]

proj_z = -24.008363727141393
proj_x = -0.2297853284073595
proj_y = -35.20627970757762

m.addConstr(gurobi.quicksum([condition_x]) == proj_x)
m.addConstr(gurobi.quicksum([condition_y]) == proj_y)
m.addConstr(gurobi.quicksum([condition_z]) == proj_z)

m.update()
m.optimize()
print(f'number of solutions: {m.SolCount}')

The above results in no solution. However. I know that the correct solution is:

x = -30.77400000000125
y = 17.09599999999955 
z = -24.013000000000005

What am I doing wrong? Thanks in advance!

0

• 

  Jaromił Najman

  • Gurobi Staff

  November 11, 2022 13:18

  The default lower bound for variables is \(0\). You have to explicitly state that a variable is allowed to attain negative values.

  shot_x = m.addVar(lb=-gurobi.GRB.INFINITY, vtype=gurobi.GRB.CONTINUOUS, name='shot_x')
  shot_y = m.addVar(lb=-gurobi.GRB.INFINITY, vtype=gurobi.GRB.CONTINUOUS, name='shot_y')
  shot_z = m.addVar(lb=-gurobi.GRB.INFINITY, vtype=gurobi.GRB.CONTINUOUS, name='shot_z')

  Best regards, 
  Jaromił

  0
• 

  David Torres Sanchez

  • Gurobi Staff

  November 11, 2022 13:28

  Hi,
  
  The default lower bound of variables is 0.0  Model.addVar().
  You can change this when you create the variables:

  shot_x = m.addVar(lb=-100, vtype=gurobi.GRB.CONTINUOUS, name='shot_x')

  Additionally, you don't need to add the \(\texttt{quicksum}\) expression as your \(\texttt{condition_*}\) are already LinExpr.
  
  Cheers, 
  David

  0
• 

  Huib Meulenbelt

  November 11, 2022 13:40

  • 
  • 
  • 

  Thank you Jaromil and David!

  0

Please sign in to leave a comment.