Computational Time Series Analysis


This project aims at developing efficient algorithms for advanced time series analysis. We are particularly interested in the identification of different temporal regimes in time series, feature extraction localized within such regimes, trend analysis, signal classification, and change point detection. Our mathematical tools span a wide range from advanced hidden Markov models via low-dimension manifold approximation and parameter estimation for stochastic processes to advanced Bayesian statistics in signal classification, model discrimination, and clustering. In recent years we have considered applications of our methods to time series originating from molecular dynamics and spectroscopy, the geo-sciences, sociology, and in medical diagnostics.


The project has been funded by MATHEON within project A4 and by Microsoft Research.

Selected Publication

  • Horenko, I. and Schütte, Ch. (2010) On metastable conformational analysis of non-equilibrium biomolecular time series. Multiscale Modeling & Simulation, 8 (2). pp. 701-716. ISSN 15403467


  • Horenko, I. and Klein, R. and Dolaptchiev, S. and Schütte, Ch. (2008) Automated Generation of Reduced Stochastic Weather Models I: Simultaneous Dimension and Model reduction for Time Series Analysis. Mult. Mod. Sim., 6 (4). pp. 1125-1145.


  • Meerbach, E. (2009) Off- and Online Detection of Dynamical Phases in Time Series. PhD thesis, Free University of Berlin.


Christof Schütte

Project Researchers:

Juan Latorre, Marco Sarich, Hao Wu


Cooperation: Illia Horenko