stack MLinExpr and add constraints

回答済み

ANH TU VU

2022年06月05日 17:16

• 
• 
• 

m = gp.Model('LP_model')

expected_cost = m.addMVar(70, lb=0, ub=float('inf'), vtype='C', name='expected_cost')

m.setObjective(expected_cost.sum(), GRB.MINIMIZE)

 

# expected_cost.shape = (70,), ST_mat.shape = (70,70), R_mat.shape=(70,70)

 

futil_maintanance = expected_cost @ ST_mat

futil_replace = expected_cost @ R_mat

 

futility = np.vstack((futil_maintanance, futil_replace)).T

 

I want to stack two matrices vertically in Gurobi, but it doesn't work now

 

EV_ful = EV_myopic + beta*futility

 

# futil_maintanance.shape = (70,), futil_replace.shape=(70,), futility.shape=(70,2), EV_ful.shape=(70,2)

 

I want to add constraints that each value in expected_cost is greater than the two values in the same row in matrix EV_ful. Like expected_cost[i] > EV_ful[i,j] for j in range(2) for I in range(70).

 

Can you please walk me through the issue? I used addConstrs as below but it did not work

 

for i in range(70):

for j in range(2):

le = EV_ful[i,j]

m.addConstrs(expected_cost[i] >= le)

0

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2025年11月16日 23:35

  This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum, or try Gurobot, our chatbot interface offering instant, expert-level support.
• 

  Maliheh Aramon

  • Gurobi Staff

  2022年06月08日 06:29

  Hi, 

  We cannot pass Gurobi MLinExpr() objects to the numpy.vstack() method. This method only accepts numpy arrays. To implement the constraints as you explained, I think the script below should do the job:

  import gurobipy as gp
  from gurobipy import GRB
  import numpy as np
  
  if __name__ == "__main__":
       m = gp.Model("LP_model")
      expected_cost = m.addMVar(
          70, lb=0, ub=float("inf"), vtype="C", name="expected_cost"
      )
      m.setObjective(expected_cost.sum(), GRB.MINIMIZE)
  
      ST_mat = np.random.random((70, 70))
      R_mat = np.random.random((70, 70))
      beta = 0.001
      EV_myopic = np.random.random((2, 70))
  
     for col in range(70):
         m.addConstr(
              expected_cost @ ST_mat[:, col] * beta 
              + EV_myopic[0, col] 
              - expected_cost[col]
              <= 0
        )
  
     for col in range(70):
         m.addConstr(
             expected_cost @ R_mat[:, col] * beta
             + EV_myopic[1, col]
             - expected_cost[col]
             <= 0
        )

  Let me know if something is not clear.

  Best regards,

  Maliheh

  0
• 

  ANH TU VU

  2022年06月08日 08:04

  • 
  • 
  • 

  Hi Maliheh, thanks for your support. I follow the code you provided, but it doesn't seem to work now. Please see the code below and help me to go over the problem. Thank you so much for your time and effort

   

  def gurobi_LP(S, p, MF, params, beta=0.75, threshold=1e-6, suppr_output=False):

  
  

  ST_mat = np.zeros((S, S))

  R_mat = np.vstack((np.ones((1, S)),np.zeros((S-1, S))))

   

  EV_myopic = myopic_costs(S, MF, params, p)

  
  

  # Create a new model

  m = gp.Model('LP_model')

  
  

  # Create variables

  expected_cost = m.addMVar(S, lb=0, ub=float('inf'), vtype='C', name='expected_cost')

  
  

  # Set objective

  m.setObjective(expected_cost.sum(), GRB.MAXIMIZE)

  
  

  # Add constraints

  for i in range(S):

  # m.addConstr(expected_cost[i] <= EV_myopic[:,0][i] + expected_cost @ ST_mat[:,i]) * beta

  # m.addConstr(expected_cost[i] <= EV_myopic[:,1][i] + expected_cost @ R_mat[:,i]) * beta

  m.addConstr(expected_cost[i] <= EV_myopic[0,i] + expected_cost @ ST_mat[:,i]) * beta

  m.addConstr(expected_cost[i] <= EV_myopic[1,i] + expected_cost @ R_mat[:,i]) * beta

  
  

  # Run LP model

  m.optimize()

  
  

  # Extract outputs

  exp_cost = np.array([expected_cost[i].X for i in range(S)])

  # exp_cost = np.array(exp_cost)

  
  

  futil_maintanance = np.dot(exp_cost, ST_mat)

  futil_replace = np.dot(exp_cost, R_mat)

  futility = np.vstack((futil_maintanance, futil_replace)).T

   

  exp_cost_new = EV_myopic + beta*futility

  
  

  return (exp_cost, choice_prob(exp_cost_new), choice_prob(EV_myopic))

  0
• 

  Maliheh Aramon

  • Gurobi Staff

  2022年06月09日 05:25

  Hi, 

  I was able to run the snippet that I sent before. Could you run that snippet?
  
  I cannot run your code snippet because the input data is missing. Do you get the error "Invalid argument to Model.addConstr"? If so, the constraints should be implemented exactly as below:

  m.addConstr(
      EV_myopic[0, i] + expected_cost @ ST_mat[:, i] * beta - expected_cost[i]
      >= 0
  )
  m.addConstr(
     EV_myopic[1, i] + expected_cost @ R_mat[:, i] * beta - expected_cost[i] 
     >= 0
  )

   In your implementation, the LHS is a Gurobi Var object and the RHS is a Gurobi MLinExpr object. As shown in the documentation of a Linear Matrix Constraint in the form \(\texttt{expr1 sense expr2}\), either both \(\texttt{expr1}\) and \(\texttt{expr2}\) are MLinExpr objects or one is MLinExpr object and the other one is a constant. 

  If this explanation does not resolve the issue, please share the code snippet with input data such that I can run it and also include the exact error message you get. 

  Best regards,

  Maliheh

  0

投稿コメントは受け付けていません。