Hello guys!

I'm having problems calculating the K2 invariant. I sampled the X observable from the Rossler system with a = b = 0.2, c = 5.7 and sampling rate 0.2 (as in Marwan 2007), in order to have a time series with length 10,000. But I found 0.060 +-0,06 instead of 0,069 +-0.03, and the plot of K2 x epsilon fluctuates havelly. The plot of "total number of lines at least" x Length is not so smooth as in Marwan 2007 too, and it has a 3th scaling region between L=1 and L=10.

After 2 months I don't know what can I be doing wrong. Perhaps it is the calculation of the cummulative histogram Pc(l). So, I'd like to ask you a question. Imagine a RP with 8 isolated points, 4 lines with length 2 and 2 lines with length 4. So, the histogram will be P(L) = {8, 4, 0, 2}. Now, I want to find the cummulative histogram Pc(L). I have two interpretations of how to compute it:

1st - Pc(L) = {14, 6, 2, 2}

2nd - Pc(L) = {24, 10, 4, 2}, because the line L=4 have 3 lines with L=3 on it etc (that is my 2nd interpretation).

Is the correct Pc(L) the 1st one? Or none is correct?

I will be very gratefull for everyone who could help.

## Problems calculating K2

- Norbert
- Expert
**Posts:**183**Joined:**Wed Jan 4, 2006 11:03**Affiliation (Univ., Inst., Dept.):**Potsdam Institute for Climate Impact Research, Germany**Location:**Potsdam, Germany**Location:**Potsdam Institute for Climate Impact Research, Germany

### Re: Problems calculating K2

Yes, your second guess is correct. I thought that it would be clear from our explanations.

I don't know where the problem could be. I have to look for our code. Perhaps we can provide it here for reproducing the results.

I don't know where the problem could be. I have to look for our code. Perhaps we can provide it here for reproducing the results.