Cumulative RMS in Matlab

Hi,

I'm trying to find the cumulative RMS of a vector. (The data happens to be model error - labelled devCoM for deviation of centre of mass. The calculation should get less accurate as time proceeds, so cumulative error is important.)

I can find the cumulative sum by taking

cumsum(devCoM)

...but this fluctuates about zero.

I can find the RMS of the entire dataset by taking

sqrt(sum(devCoM.^2)

...but this doesn't reveal the expected increase in error over time.

Is there some way I could combine these methods?

Thanks,

Frederick

Answer 1

[Jan]

Start with a mathematical definition of what you want to achieve.

I guess you want this:

x = rand(1, 100);  % Your data
% "Cumulative root mean square"
y = sqrt(cumsum(x.*x) ./ (1:numel(x)))

So the cumulative concerns the mean of the squared elements only, correct?

[Frederick Wells]

Great, this does just want I wanted it to! Thanks very much :)

Had some trouble for a while since I was using a column vector and the denominator produces a row, so the code that ultimately worked for me was

y = sqrt(cumsum(x.^2) ./ (1:numel(x))');