Error while implementing a MATLAB code in Python

Question

I am trying to implement this code in python. It works fine in MATLAB But when I use same logic the python interpreter gives a different answer.

I am having problem in finding the correct function or method in python. Need help regarding this. Code is given below

Code:

Sensors = 10; %number of receivers
samples = 10000; %number of set of signals
p = 10; %number of signals
theta1 = [10 20 40 60 100 120 140 80 90 110];
angles = theta1*(pi/180);
X = zeros(Sensors,samples);
for k=1:p-1
    Array_factor = exp(j*pi*cos(angles(1,k))*(0:Sensors-1 ));
    br =ones(1,samples); %creating real values between 1-10000
    temp = rand(1,samples); %generate random numbers b/w 1 and 10000
    br(find(temp<.5))=-1; %find real numbers
    bi =ones(1,samples); %creating imaginary values between 1-10000
    temp = rand(1,samples); %generate c omplex random numbers b/w 1 and 10000
    bi(find(temp<.5))=-1; %find imaginary numbers
    b = br+j*bi;
    X = X +Array_factor'*b; %complete Input signal combined with array factor
end

The above is a MATLAB code Now I have converted it in Python

import numpy as np

#number of receivers
num_sensors = 10

#Total samples to scan
samples = 10

#number of signals
p = 10

theta1 = [10,20,40,60,100,120,140,80,90,110]

#angles in degrees
angles = [i*(np.pi/180) for i in theta1]

#Input Signal
X = np.zeros((num_sensors,samples), dtype = complex)


for k in range(1,p-1):
    array_factor = np.exp(1j*np.pi*np.cos(angles[k])*np.arange(sensors))
    
    #creating real values 
    
    br = np.ones(samples)

    temp = np.random.rand(samples)

    br[temp < 0.5] = -1
#        br[temp<0.5] = -1

    bi = np.ones(samples)

    temp = np.random.rand(samples)

    bi[temp<0.5] = -1

    b = br + 1j*bi

    X += np.dot(array_factor,b)

print(X)

Output expected in Python

X = 10 x 10000 double b = 1 x 10000 double

Answer 1

The NumPy equivalent for Array_factor'*b expression is:

np.atleast_2d(array_factor).conj().T @ np.atleast_2d(b)

---

• np.atleast_2d(array_factor).conj().T converts array_factor from row vector to column vector as described in the following answer.
  .conj().T is the NumPy equivalent of ' operator.
• np.atleast_2d(b) converts b from 1D array to 2D array.
  Note that in MATLAB rows vectors are 2D arrays (all vectors are matrices) and in NumPy the row vector is 1D array.
• The @ operator is used for matrix multiplication (matches * MATLAB operator).
  We may also use dot product: np.dot(np.atleast_2d(array_factor).conj().T, np.atleast_2d(b))

---

Note:
Since Array_factor is complex we have to use .' for transposing.
' operator produces complex conjugate transpose.

---

There is another bug in the loop indexing:
It should be for k in range(p-1) instead of for k in range(1,p-1).

---

Updated Python code:

import numpy as np

#number of receivers
num_sensors = 10

#Total samples to scan
samples = 10000

#number of signals
p = 10

theta1 = [10,20,40,60,100,120,140,80,90,110]

#angles in degrees
angles = [i*(np.pi/180) for i in theta1]

#Input Signal
X = np.zeros((num_sensors,samples), dtype = complex)


for k in range(p-1): #for k in range(1,p-1):
    array_factor = np.exp(1j*np.pi*np.cos(angles[k])*np.arange(num_sensors))
    
    #creating real values 
    
    br = np.ones(samples)

    temp = np.random.rand(samples)

    br[temp < 0.5] = -1
#        br[temp<0.5] = -1

    bi = np.ones(samples)

    temp = np.random.rand(samples)

    bi[temp<0.5] = -1

    b = br + 1j*bi

    #X += np.dot(array_factor, b)
    X += np.atleast_2d(array_factor).conj().T @ np.atleast_2d(b)

print(X)