Unable to convert argument to an expression

Answered

Zahra Hoobakht

March 11, 2024 17:19

• 
• 
• 

Hi 

I am trying to define my objective function but I am getting "Unable to convert argument to an expression" error. I am using the Sympy library because I need to get Hessian of some matrices based on some variables and then optimize them based on the constraints I have 

trJinv=sp.trace(Jinv)
trace_Jinv=sp.expand(trJinv)
from gurobipy import Model, GRB
import gurobipy as gp
# Create a model
m = Model("Minimize expression")
m.addVar(lb=0.0, ub=GRB.INFINITY, name="R1")
m.addVar(lb=0.0, ub=GRB.INFINITY, name="R2")
constraints = []
variableNames = ["R1","R2"] # save all variable names
R1 = sp.Symbol("R1")
R2 = sp.Symbol("R2")

obj=trace_Jinv.subs({(x * x.conjugate().T)[0,0]: R1,
    (x * x.conjugate().T)[1,1]: R2})
constraints.append(R1+R2==5)
m.setObjective(obj, gp.GRB.MINIMIZE)

my objective is to minimize (29*𝑅1+17*𝑅2)/(300*𝜋**2*𝑅1**2+627*𝜋**2*𝑅1*𝑅2+12*𝜋**2*𝑅2**2)

I would appreciate it if you could help me

0

• 

  Jonasz Staszek

  • Community Moderator

  March 11, 2024 19:04

  • 
  • 
  • 

  Hi Zahra,

  first of all, I would suggest you post a minimal, reproducible example of the problem you are struggling with.

  Further, take a look at those two links (here and here). Perhaps they will guide you a little bit in formulating your objective.

  As a general remark: as stated in the documentation, model.setObjective() only accepts LinExpr or QuadExpr (or arguments which can be converted to these). Your obj object:

  obj=trace_Jinv.subs({(x * x.conjugate().T)[0,0]: R1,
      (x * x.conjugate().T)[1,1]: R2})

  does not meet these criteria. You may be better-off rewriting the objective directly, the way you wrote it down at the end of your question.

  Also, you may want to store your variables as R1 and R2:

  R1 = m.addVar(lb=0.0, ub=GRB.INFINITY, name="R1")
  R2 = m.addVar(lb=0.0, ub=GRB.INFINITY, name="R2")

  as otherwise Gurobi will see them as sp.Symbol objects and you'll get an error.

  Hope this helps.

  Best regards
  Jonasz

  0
• 

  Zahra Hoobakht

  March 11, 2024 21:31

  • 
  • 
  • 

  Hi Jonasz

  thanks for your response. I narrow down my objective function to the numerator of trace which is 29*R1+17*R2 in which R1 and R2 are symbols coming from sympy library. I want to solve a minimization problem like this but still receiving the same error:

  Min 29*R1+17*R2

  constraints: R1+R2=5

  R2>=3*R1-3

   

  R1,R2>0
  
  my code for the objective function is :

  from gurobipy import Model, GRB
  import gurobipy as gp
  import sympy as sp
  R1,R2 =sp.symbols('R1 R2')
  objective=R1+R2
  # Create a model
  m = Model("Minimize expression")

  r1 = m.addVar(lb=0.0, ub=gp.GRB.INFINITY, name="r1")
  r2 = m.addVar(lb=0.0, ub=gp.GRB.INFINITY, name="r2")
  obj=objective.subs({ R1:r1,
      R2: r2})

  
  m.setObjective(obj, gp.GRB.MINIMIZE)

  ------------------------------------------------------------------------------------------

  GurobiError                               Traceback (most recent call last)
  Cell In[2], line 15
       10 r2 = m.addVar(lb=0.0, ub=gp.GRB.INFINITY, name="r2")
       11 obj=objective.subs({ R1:r1,
       12     R2: r2})
  ---> 15 m.setObjective(obj, gp.GRB.MINIMIZE)

  File src\\gurobipy\\model.pxi:1488, in gurobipy.Model.setObjective()

  File src\\gurobipy\\util.pxi:44, in gurobipy._simpleexpr()

  GurobiError: Unable to convert argument to an expression

  0
• 

  Jonasz Staszek

  • Community Moderator

  March 12, 2024 11:54

  • 
  • 
  • 

  Hi Zahra,

  as I wrote earlier, you cannot use Sympy objects as arguments to your objective function.

  Perhaps this:

  m.setObjective(r1 + r2, gp.GRB.MINIMIZE) 

  is what you are looking for?

  Best regards
  Jonasz

  0

Please sign in to leave a comment.