xcorr results: c written in long format, lags in wide format

I want to calculate the cross-correlation between ankle and hip movement.

My data are organised in a structure: model.filt.modelOutputs with 12500x3 matrices.

I attached an example file with only the filtered Ankle and Hip angles.

The data contains several movement cycles which can be identified by the starting frames stored in tStart.

When I use [c, lags] = xcorr, I get c as 1301x1 double and lags as 1x1301 double.

Can anyone explain, why this happens? When I use different example vectors, both results have the same orientation.

This is the part of my code, which reproduces the results::

load modelExample.mat
tStart = [1,611,1261];
[c, lags] = xcorr(modelExample.filt.RAnkleAngles(tStart(2):tStart(3),2), modelExample.filt.RHipAngles(tStart(2):tStart(3),2), 'normalized');

Answer 1

[DGM]

I don't see it documented anywhere, but lags is always a row vector, regardless of the orientation of the input.

If you want it to match the orientation, you can always either transpose or reshape it as needed.

[Torsten Pohl]

Hi DGM,

thank you for your answer. I already have a working solution for my code. I just wanted to understand, if there is a specific reason for it.

[DGM]

I'm really not the one to ask for reasons. If you look in xcorr, lags is simply calculated as

lags = -maxlag:maxlag;

which is a row vector. That's all I know.