Hi All, I am not sure if this is the right section to post this question but it does involve probability..so please redirect me if necessary. I am currently looking at the Robinson et al. (2013) paper on rank vector entropy in MEG (doi: 10.3389/fncom.2012.00101). Due to my lack of mathematical knowledge I am struggling to understand what role the 'leaky integrator' performs in this algorithm (or what a leaky integrator does in general?). Essentially a state histogram is produced counting all the occurrences of the rank vector states and the probability of a state occurring is calculated...however, this just produces the absolute entropy across time once Shannon's entropy has been applied. Am I right in thinking by introducing a time constant using the leaky integrator we are able to measure the relative fluctuations in entropy over time as opposed to the absolute entropy? And if so, how does the introduction of a time constant or leaky integrator achieve this? (I'm not sure mathematically how this gives us temporal information). Apologies if this is posted in the wrong section or exposes my complete lack of knowledge! Many thanks.