Variable values return 'None' after solving the problem

[BrianN92]

Describe the bug
After solving the problem, variables u & y returned "None"

To Reproduce

import cvxpy as cp
import numpy as np
from numpy.linalg import pinv

# Define the state space matrices
A = np.array([[ 1.96745861e-05,  8.13485217e-05,  5.77344499e-04,
         2.50156980e-03, -4.20390862e-04],
       [ 1.83218791e-05, -6.64167786e-05, -2.59408881e-03,
         4.48058026e-04,  3.18793055e-03],
       [-1.89169954e-05,  5.12847259e-05, -5.48305222e-04,
        -3.06533107e-04,  8.44923166e-03],
       [-1.20247997e-05, -6.27915617e-07, -9.10863713e-05,
        -8.94478349e-04, -5.40108892e-04],
       [ 8.84586089e-06,  1.63172443e-05,  2.90230873e-04,
         8.18201672e-05, -4.44945551e-03]])
B = np.array([[-3.76965442e-04,  9.39303000e-06, -3.85775209e-04],
       [-6.18910044e-05, -2.51123269e-05,  2.66895587e-03],
       [-2.55411165e-05, -1.04126946e-05,  3.07196176e-03],
       [-9.61991011e-05, -2.35462956e-06, -1.32939899e-04],
       [ 2.69436292e-07,  5.11188160e-06, -1.64164895e-03]])
C = np.array([[-0.49404405, -0.08989964,  0.0388238 ,  0.73830449,  0.03432347],
       [-0.09379532,  0.52031726,  0.69049302, -0.01848544, -0.48960013]])
P = np.array([[-2.06760151e-05, -3.21313691e-05, -1.30601012e-04,
         -1.94565573e-03, -3.02356442e-04],
        [-9.48289848e-06, -1.47534791e-05, -1.92291562e-03,
         -2.36686530e-04,  9.72073962e-03],
        [ 2.21965970e-08, -9.95765465e-09,  2.32493205e-07,
          1.68952239e-06,  1.19896873e-06],
        [ 1.24753321e-07,  6.03565609e-08,  3.92995927e-06,
          1.56616827e-06, -5.94143681e-05],
        [-1.38540647e-11,  3.28933541e-11,  1.05254161e-10,
         -1.43334351e-09, -4.32397068e-09],
        [-6.15186801e-10, -7.62775488e-10, -1.96259023e-08,
         -7.12773739e-09,  2.96860790e-07],
        [-2.26745719e-14, -6.75690148e-14, -1.27542998e-12,
          8.76124070e-13,  2.10134794e-11],
        [ 3.05688294e-12,  3.84253319e-12,  9.91919233e-11,
          3.48000947e-11, -1.48501598e-09]]) 
Qc =  np.array([[-8.45801844e-05,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00],
        [ 4.35226485e-03,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00],
        [ 2.76036183e-07, -8.45801844e-05,  0.00000000e+00,
          0.00000000e+00],
        [-2.18694183e-05,  4.35226485e-03,  0.00000000e+00,
          0.00000000e+00],
        [-1.51381989e-09,  2.76036183e-07, -8.45801844e-05,
          0.00000000e+00],
        [ 1.09514976e-07, -2.18694183e-05,  4.35226485e-03,
          0.00000000e+00],
        [ 7.70546269e-12, -1.51381989e-09,  2.76036183e-07,
         -8.45801844e-05],
        [-5.48482155e-10,  1.09514976e-07, -2.18694183e-05,
          4.35226485e-03]])
Qd = np.array([[ 1.19794929e-04, -4.35021797e-06,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00],
        [-1.28349584e-05, -2.35965438e-05,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00],
        [ 2.00207348e-07,  5.00827964e-09,  1.19794929e-04,
         -4.35021797e-06,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00],
        [ 7.89893960e-08,  7.05527333e-08, -1.28349584e-05,
         -2.35965438e-05,  0.00000000e+00,  0.00000000e+00,
          0.00000000e+00,  0.00000000e+00],
        [-1.75896686e-10,  1.88459231e-13,  2.00207348e-07,
          5.00827964e-09,  1.19794929e-04, -4.35021797e-06,
          0.00000000e+00,  0.00000000e+00],
        [-3.17811134e-10, -3.48672309e-10,  7.89893960e-08,
          7.05527333e-08, -1.28349584e-05, -2.35965438e-05,
          0.00000000e+00,  0.00000000e+00],
        [ 1.37219715e-13, -2.07807729e-14, -1.75896686e-10,
          1.88459231e-13,  2.00207348e-07,  5.00827964e-09,
          1.19794929e-04, -4.35021797e-06],
        [ 1.54604756e-12,  1.75203552e-12, -3.17811134e-10,
         -3.48672309e-10,  7.89893960e-08,  7.05527333e-08,
         -1.28349584e-05, -2.35965438e-05]])
Ud = np.array([[0,0,0,0],[87.78,87.85,87.88,87.94]])
Y = np.array([[1.549219073455385898e+00],[3.605849153580493294e+00]])

# Prediction/control horizen
Nc = 4
Np = 4

n1,n_in = B.shape
B_d = B[:, 0:n_in-1]
B_c = B[:, n_in-1:n_in]
n_d = B_d.shape[1]
n_c = B_c.shape[1]

# Define weight matrices
weights_y = np.array(([0.8,0.2])) # Arbitrary weights
W_y = weights_y * np.identity(Y.shape[0])
weights_u = np.array(([0.05,0.05,0.9]))
W_u = weights_u * np.identity(B.shape[1])

# Initialize the problem
u = cp.Variable((n_c,Nc))  # Decision variable
umin = np.zeros(n_c) # lower bound
umax = np.ones(n_c)
y = cp.Variable((Y.shape[0],Nc+1))
ymin = np.zeros(Y.shape[0])
ymax = np.array([3.0,5.0])
y_init = cp.Parameter(Y.shape[0])
x = cp.Variable((n1,Nc+1))
x_init = cp.Parameter(n1)
U = cp.atoms.affine.vstack.vstack((Ud,u))
objective = 0
x_init.value =  np.dot(pinv(C),Y).reshape((-1,))
y_init.value = Y.reshape((-1,))
constraints = [x[:,0]==x_init,y[:,0]==y_init]
for k in range(Nc):
    objective += cp.quad_form(y[:,k], W_y) + cp.quad_form(U[:,k],W_u) # Objective function J=Y'QY + U'RU
    constraints += [x[:,k+1] == A@x[:,k] + B_d@Ud[:,k] + B_c@u[:,k]]
    constraints += [y[:,k+1] == C@x[:,k+1]]
    constraints += [ymin <= y[:,k], y[:,k] <= ymax]
    constraints += [umin <= u[:,k], u[:,k] <= umax]
prob = cp.Problem(cp.Minimize(objective), constraints)
prob.solve(solver=cp.OSQP, warm_start=True)

Version

• OS: Windows 10
• CVXPY Version: '1.0.25'

Additional context
Did anyone run into this issue? I am new to cvxpy, not sure how to address it... Thank you!

[rileyjmurray]

Before trying to access variable values, you should always check if prob.value is finite. In your case, you have a minimization problem, and prob.value == np.inf after calling prob.solve(). This means your problem is infeasible.

Diagnosing infeasibility issues is a common task when using optimization models in practice. Usually you will find either a bug in your code, or you will see that the abstract mathematical model can be infeasible (even if coded up perfectly). Helping users identify which situation they are in is beyond the scope of GitHub issues, so I'll be closing this issue. However, you can ask for help on StackOverflow with a cvxpy tag.