TSTOOL home page | TSTOOL documentation page | TSTOOL link page |

Fast, but crude auto mutual information of a scalar timeseries for the timelags from zero to maxtau. The input time series should be much longer than maximal timelag maxtau. The algorithm uses equidistant histogram boxes, so results are bad in a mathematical sense. However, a fast algorithm based on ternary search trees to store only nonempty boxes is used.

**Syntax:**

`a = amutual(ts, maxtau, partitions)`

**Input arguments:**

`ts`- vector holding time series data`maxtau`- maximal time lag`partitions`- number of partitions for the one-dimensional histogram

**Output arguments:**

`a`- vector of length maxtau+1, holding auto mutual information

Copyright © 1997-2009 DPI Göttingen