Element-wise subtraction of two stem plots to form an array

I am having trouble subtracting one stem plot from another of the same size. Whenever I run my code due to the subtraction of x[n]-xhat[n] I recieve a different array every time, and therefore calculations like error signal and mean squared error MSE change every time I run the program. Is this because Matlab does not recognize that the stem plots are discrete? Here is my code:

a = [1 -1/4];
b = [1]; 
Number_of_samples = 200;
Time_length = 20;
Ts = Time_length / (Number_of_samples - 1);
n = 0 : Ts : Time_length;
%real signal, non-periodic
x = ((0.9) .^ n) .* cos(2 * pi .* n); 
subplot(4, 2, 1);
stem(n, x, 'filled');
title('Input Signal x[n]');
xlabel('n');
ylabel('x[n]'); 

d= sqrt(0.04)*randn(size(y))+1;%Gaussian noise
subplot(4,2,3)
stem(n,d,'filled');
title('Gaussian Noise d[n]');
xlabel('n');
ylabel('d[n]');

e=[1];
f=[1 -1/4];
xhat=filter(e,f,yhat_new);
subplot(4,2,6);
stem(n,xhat,'filled');
title('Output Signal xhat[n]');
xlabel('n');
ylabel('xhat[n]');

error = x-xhat;              %Error between two signals 
sqrd_err = error.^2;         %Taking element-wise sum of the error vector 
MSE = mean(sqrd_err,'all');

energy_signal = sum(error.^2); %Energy Signal

Again, I am having trouble with element-wise subtraction of two stem plots. I am able to get a correct stem plot of the subtraction, but I need specific values like error signal and MSE.

Answer 1

[Cris LaPierre]

Subtraction is always elementwise.

Your numbers change each time because you are using random numbers

d= sqrt(0.04)*randn(size(y))+1; % Gaussian noise
              ^^^^^

This signal ultimately ends up in xhat, hence the constantly changing result.

[Samantha Bartos]

Good catch, thanks! I need to add Gaussian noise to my signal, so is there a way around the random function?

[Cris LaPierre]

That will depend on what you need to do. Gaussian noise is random.