Invalid argument to Model.addConstr

Answered

Yutian He

August 27, 2024 00:22

• 
• 
• 

Dear,

I am getting the following error: Invalid argument to Model.addConstr.

I put my code downstairs, could you please help me to fix this error? Thank you!

Here is my code:

import random
import numpy as np
import gurobipy as gp
from gurobipy import GRB
from gurobipy import *
from gurobipy import quicksum



m=15
n=10
m_a1 = (m + 1) // 2
m_a2 = m - (m + 1) // 2
np.random.seed(1)

#Define x^0
x0_first_part = np.random.uniform(-1, 0, n // 2)

# Generate the next ⌊n/2⌋ - 1 components from Unif(0, 10)
x0_second_part = np.random.uniform(0, 10, n // 2 -1)

# Set the last component to zero
x0_last_part = np.array([0])

# Combine all parts to create the vector x^0
x0 = np.concatenate([x0_first_part, x0_second_part, x0_last_part])
x0



xi = np.zeros((m, n))

# Loop through the first ceil(m/2) data points
for i in range((m + 1) // 2):
    # Assign the sensitive attribute as -1
    xi[i, -1] = -1
    # Assign the first (n-1) features from Unif(0, j)
    for j in range(1, n):
        xi[i, j-1] = np.random.uniform(0, j)

# Loop through the remaining data points
for i in range((m + 1) // 2, m):
    # Assign the sensitive attribute as 1
    xi[i, -1] = 1
    # Assign the first (n-1) features from Unif(0, j+2)
    for j in range(1, n):
        xi[i, j-1] = np.random.uniform(0, j + 2)



E_xi = np.mean(xi, axis=0)

# Calculate the dot product of E[xi]^T and x0
dot_product = np.dot(E_xi, x0)

# Generate the noise from Unif(-0.1, 0.1)
uniform_noise = np.random.uniform(-0.1, 0.1, size=m)

# Final noise for each data point is uniform_noise * dot_product
noise = uniform_noise * dot_product

y_tilde = np.dot(xi, x0) + noise
var=list(range(m))
var2 = list(range(n))



m0 = gp.Model("Knapsack")

x = m0.addVars(var, vtype = GRB.CONTINUOUS, name = "x")
z = m0.addVars(var, vtype=GRB.CONTINUOUS, name="z")


obj_expr1 = gp.quicksum(z[i] for i in var)

# First half: sum of z[i] for i from 0 to 7 (indices 1 to 8)
obj_expr2 = gp.quicksum(z[i] for i in range(8))

# Second half: sum of z[i] for i from 8 to 14 (indices 9 to 15)
obj_expr3 = gp.quicksum(z[i] for i in range(8, 15))


objective = (obj_expr1 / m) + (obj_expr2 * m_a2 / (m_a1 * m)) + (obj_expr3 * m_a1 / (m_a2 * m))

for i in var:
    expr = sum(xi[i] * x[i] for i in var) - y_tilde[i]  # This creates a LinExpr using Gurobi's variable x[i]
    
   
    m0.addConstr(z[i] >= expr)
    m0.addConstr(z[i] >= -expr)

m0.setObjective(objective, sense=GRB.MINIMIZE)

m0.optimize()

Related to

• Python

0

• 

  Riley Clement

  • Gurobi Staff

  August 27, 2024 03:47

  Hi Yutian,

  I'll try and help you debug, rather than give you the answer.

  The full error gives you the exact place where the error occurs:

  GurobiError                               Traceback (most recent call last)
  Cell In[4], line 89
       85 for i in var:
       86     expr = sum(xi[i] * x[i] for i in var) - y_tilde[i]  # This creates a LinExpr using Gurobi's variable x[i]
  ---> 89     m0.addConstr(z[i] >= expr)
       90     m0.addConstr(z[i] >= -expr)
       92 m0.setObjective(objective, sense=GRB.MINIMIZE)
  
  File src/gurobipy/model.pxi:3675, in gurobipy.Model.addConstr()
  
  File src/gurobipy/model.pxi:3467, in gurobipy.Model.__addConstr()
  
  GurobiError: Invalid argument to Model.addConstr

   

  The error is saying the addConstr method does not like something about z[i] >= expr.  If you put the following lines of code just before the line which causes the error, then it will print the value of z[i] and expr to screen.

      print(f"{z[i]=}")
      print(f"{expr=}")

  Does the output look like what you would expect?

  - Riley

   

  0
• 

  Yutian He

  August 27, 2024 04:05

  • 
  • 
  • 

  Hi Riley,

  Thank you for the two-line code! You are right, looks like it is not what I want, I will check it later and give you more feedback. Thank you again!

  0

Please sign in to leave a comment.