About the delay in crp toolbox
Posted: Thu Jan 12, 2012 16:50
what to do if the delay is smaller than 1? how can we get a recurrence plot of time series with embedding delay less than 1?
Hello,
it seems that there is misunderstanding. As your data is discretised (because it is measurement), your time series consists of N data values, each can be indexed by natural number i. Now the delay is just the natural number of steps between two of your values, say i+tau. Tau cannot be smaller than 1. I suppose you would like to use the delay in terms of a measured real time t, what means it could be real number instead of natural number. But this can be used by simply dividing your real valued delay by the sampling time delta_t (where delta_t can be real number). Here is the relation between the time and the index in the discrete time series: t = delta_t * i.
Best regards
Norbert