module import : NameError (name 'A' is not defined )

Question

I have created a module which calculates QR decomposition and tests .It contains several different functions. I have saved this file as Numla.py . I want to import these functions in Jupyter Notebook but I keep having this error . I have also tried other importing methods e.g from Numla import qr , from Numla import * , import Numla as num . But I keep jhaving the same name error .

import Numla Numla.qr(A)

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
Input In [11], in <cell line: 2>()
      1 import Numla
----> 2 Numla.qr(A)

NameError: name 'A' is not defined

import numpy as np 


def qr(A):
  #source: #https://rosettacode.org/wiki/QR_decomposition#Python
    m, n = A.shape
    Q = np.eye(m)
    for i in range(n - (m == n)):
        H = np.eye(m)
        #calculates Householder matrix i: rows and i: columns from A i: rows and ith column
        H[i:, i:] = make_householder(A[i:, i])
        Q = Q@H
        A = H@A
    return Q, A

def make_householder(a):
    #finds orthogonal vector to mirror
    u = a / (a[0] + np.copysign(np.linalg.norm(a), a[0]))
    u[0] = 1
    H = np.eye(a.shape[0])
    #finds Householder projection
    H -= (2 / np.dot(u, u)) * u[:, None] @ u[None, :]

    return H

def compose(qr, make_householder):
    def comp(arg):
        return qr()(make_householder(arg))
    return comp

def test_compose(qr, make_householder):
    m=3
    n=3
    A = np.random.rand(m, n)
    Q, R = qr(A)
    Q.round(3)
    R.round(3)

    Q_np, R_np=np.linalg.qr(A)
    np.allclose(Q, Q_np)
    np.allclose(R, R_np)
    
    print (" QR-Decomposition and test ")
    print ('\n')
    print ("Q vs Q_np")
    print (Q.round(8))
    print (Q_np)
    print ('\n')
    print ("R vs R_np")
    print (R.round(8))
    print (R_np)
    print ('\n')
    
    if np.allclose(Q, Q_np) == True:
        print('ok')
        if np.allclose(R, R_np) == True:
            print ('ok')
            print('\n')
            
            return 
        
        
#test_compose(qr, make_householder)



def back_substitution(U, y):
    
    n = U.shape[0]
    x = np.zeros_like(y, dtype=np.double);
    x[-1] = y[-1] / U[-1, -1]
    
    for i in range(n-2, -1, -1):
        x[i] = (y[i] - np.dot(U[i,i:], x[i:])) / U[i,i]
    return x

def linsolve_qr():
    
    #QR decomposition with qr function 
    Q, R = np.linalg.qr(A)
    y = Q.T @ B
    #solve for x using back substitution method
    K=back_substitution(R, y)
    #using np solver to prove back substitution+QR method works
    L=np.linalg.solve(R, y)
    
    return 

def test_linsolve_qr():
    A = np.array([
        [6., 7., 8.],
        [2., 4., 3.],
        [8., 1., 2]])
    B = np.array([1., -2., 3.])
    
    #QR decomposition with qr function 
    Q, R = np.linalg.qr(A)
    y = Q.T @ B
    K=back_substitution(R, y)
    L=np.linalg.solve(R, y)

    print (" QR-Decomposition implementation to solve LGS ")
    print ('\n')
    print (back_substitution(R, y))
    print (np.linalg.solve(R, y))
    
    if np.allclose(K, L) == True:
        print('ok')
        return
    
#test_linsolve_qr()


if __name__ == "__main__":
    test_compose(qr, make_householder)
    test_linsolve_qr()

Look into variable scope if you haven't. There is no way that `A` would be able to reach where you want it to — Nicholas Hansen-Feruch, Nov 05 '22 at 23:12

score 0 · Answer 1 · answered Nov 05 '22 at 23:22

To import a function, just use the function name without the arguments:

from Numla import qr

You can then use the function qr in your notebook.

I've tested this locally and it works. It will work fine as long as Numla.py is in a directory in your Python path. If you continue having issues with Colab you might want to consult this answer.

module import : NameError (name 'A' is not defined )

1 Answers1