Parallel architecture is becoming the norm in big computing.  Hopefully this will mean more ensemble data sets in the future.  We made The First Eigenvalue Series Method for identifying interesting times in these data sets.  The method is fast, flexible, and transparent.

Presentation of The First Eigenvalue Series Method

An introduction to the First Eigenvalue Time Series Method for ensembles of time indexed model output.

For More Information

For a full discussion of this and other feature identification methods see the associated PhD thesis