RMS of matrices in a cell or array

Hi all,

If I have an n x n matrix, P, I find the rms by RMS_P = sqrt(mean(Prv(:).^2));.

Please, how can I find the rms of the ith elements in an array, P(;,:,i), along the ith dimension? Thanks.

This is what I have that outputs the rms of a matrix:

ang = 0:90:270;
low_lim = ang - 1;
up_lim = ang + 1;

%The input, inp_file is a 1001 x 1001 matrix, and it's attached.
for i = 1:length(ang)
    p_rot = zeros([1001 1001 4]);
    ang_var = low_lim + (up_lim - low_lim).*rand(1,length(ang));
    p_rot(:,:,i) = imrotate(inp_file,ang_var(i),'crop');
    mean_p_rot = mean(p_rot,3);
    P = (p_rot(:,:,1)) - mean_p_rot; 
    P(isnan(P)==1)=0; 
    RMS_P = sqrt(mean(P(:).^2));
end

I'm trying to repeat this loop multiple times. What I tried is:

for ii = 1: 5
    for i = 1:length(ang)
        p_rot = zeros([1001 1001 4]);
        ang_var = low_lim + (up_lim - low_lim).*rand(1,length(ang));
        p_rot(:,:,i) = imrotate(inp_file,ang_var(i),'crop');
        mean_p_rot = mean(p_rot,3);
        P = (p_rot(:,:,1)) - mean_p_rot; 
        P(isnan(P)==1)=0; 
        RMS_P = sqrt(mean(P(:).^2));
    end
    P_new(:,:,ii) = P;
    RMS_P_new(ii) = sqrt(mean(P_new(:).^2));
end

I'm having trouble putting the first loop in another loop, and finding the rms of each ii in P_new(:,:,ii).

Please help. Thanks!

[dpb]

Why not

P(isnan(P))=0;   % no need to explicitly compute ==1 isnan() is already logical
RMS_P = rms(P(:);  % use builtin RMS instead of sqrt(mean(P(:).^2));

In second why not just same above inside inner loop and

P_new(:,:,ii) = P;
RMS_P_new(ii) = rms(P(:));

?

[Mike]

Thanks, but I've tried the builtin RMS before...I kept getting the error: "Undefined function or method 'rms' for input arguments of type 'double'".

[dpb]

Hmmm....oh, I'd forgotten rms is in Signal Processing toolbox. Try

sqrt(mean(x.*x))

instead or if x can be imaginary

sqrt(mean(x .* conj(x)))

Think should be noticeably faster than .^2.

But, that aside, does rms(P(:)) solve the issue however you compute it?

[Mike]

Thanks! I stored each rms result from the inner loop, as an array element in the second loop.

end
    P_new(:,:,ii) = P;
    RMS_P_new(:,:,ii) = RMS_P;
end

I want click "accept your answer" but there's no button for that. I think it's because you replied as a comment, not as an answer. Thanks dpb.

Answer 1

[John BG]

Mike

have you noticed that rms has an option to aim on the dimension you want to sweep along?

P=randi(10,3,3,3)
P(:,:,1) =
          2.00          8.00          1.00
          5.00         10.00          9.00
         10.00          7.00         10.00
P(:,:,2) =
          7.00          4.00          8.00
          8.00          7.00          1.00
          8.00          2.00          3.00
P(:,:,3) =
          1.00          7.00          1.00
          1.00          4.00          5.00
          9.00         10.00          4.00
rms(P,1)
ans(:,:,1) =
          6.56          8.43          7.79
ans(:,:,2) =
          7.68          4.80          4.97
ans(:,:,3) =
          5.26          7.42          3.74
rms(P,2)
ans(:,:,1) =
          4.80
          8.29
          9.11
ans(:,:,2) =
          6.56
          6.16
          5.07
ans(:,:,3) =
          4.12
          3.74
          8.10
rms(P,3)
ans =
          4.24          6.56          4.69
          5.48          7.42          5.97
          9.04          7.14          6.45

If you find this answer of any help solving your question, please click on the thumbs-up vote link,

thanks in advance

John

[e-mail]

[Mike]

Thanks John. That'd have worked if I could use the built-in rms function. I get an error message when I try that. I stored each rms result from the inner loop, as an array element in the second loop.

I clicked the thumbs up...thanks!

end
    P_new(:,:,ii) = P;
    RMS_P_new(:,:,ii) = RMS_P;
end