ValueError: Mismatching inner dimensions

Answered

Honglin An

• July 15, 2024 13:33

import numpy as np
import gurobipy as gp
from gurobipy import *

D = 10 
F = 2 
N = 5 

H = np.random.random((N,D))-0.5
R = np.random.random(N) - 0.5
R = np.matmul(R.T,H)
H = np.matmul(H.T,H)
I = np.identity(F)

m = gp.Model()
A = m.addMVar(shape=(D,F),vtype=GRB.CONTINUOUS, name='A')
beta = m.addMVar(shape=F,vtype=GRB.CONTINUOUS,name = 'beta')
v = m.addMVar(shape = D, vtype=GRB.CONTINUOUS)

#obj = (H @ A @ beta - R) @ (H @ A @ beta - R)
obj = v @ H @ v - 2 * R @ v

m.setObjective(obj,GRB.MINIMIZE)

m.addConstr(v == A @ beta)
m.addConstrs(A @ A == I)

m.setParam('NonConvex', 2)
m.optimize()

After running the above code, I got the error message on the following line

m.addConstrs(A @ A == I)

saying that ValueError: Mismatching inner dimensions 2 vs. 10.

Does anyone have any suggestions on how to resolve this issue? Many thanks in advance.

Related to

• Matrix API

1

• 

  Marika Karbstein

  Gurobi Staff

  • July 15, 2024 13:45

  When you do matrix multiplication, the dimensions need to match. You need to use MVar.transpose().
  Respecting the dimension of matrix I, the line should be

  m.addConstr(A.transpose() @ A == I)

  0
• 

  Honglin An

  • July 15, 2024 13:51

  

  

  I see, thank you for the help! 

  0

Please sign in to leave a comment.