Interdisciplinary Data Analytics and Model Design at ICS-ULisboa

Rui Perdigão’s “Interdisciplinary Data Analytics and Model Design” is now also available as a semester doctoral course at the University of Lisbon, offering 6 ECTS for students enrolled in partner programs.

Objectives

  • Acquisition of fundamental competences in data analysis, its relevance and implementation in the conceptualization and formal analysis of systems in an interdisciplinary perspective;
  • Learning fundamental techniques for information retrieval, analysis and treatment along with its uncertainties, from data acquisition to model design;
  • Acquisition of new competences in scientific research, development and communication at the interface between natural and social sciences;
  • Special emphasis on interdisciplinary challenges of climate change and decision support towards sustainable development.

Program

  • Beneath Data, there is a Story: Fundamental principles behind the nature, geometry and dynamics of information across natural, social and joint systems;
  • Retrieving the Story: Fundamental methods for data analytics and model design. From spatiotemporal geostatistics to broader dynamic information tools for data mining, pattern recognition, causal analysis and model design;
  • Quality-checking the Story: Techniques for quality check, uncertainty assessment and data processing towards strengthening information reliability;
  • Sharing the Story: Techniques for data visualization, information sharing and overall communication of scientific results;
  • GeoSys Operation: Operational real-world examples for a) data mining and machine learning in large satellite datasets; b) nonlinear analytics and model design for earth system dynamics; c) early warning and automated decision support systems in natural (e.g. hydro-meteorological, geophysical) hazards;
  • Frontier Operation: early warning detection and adaptive decision support of critical transitions and extremes in the earth system under climate change;
  • Hands-On: Simple analytical and computational examples on the prior points.

 

Complex System Dynamics at ICS-ULisboa

Rui Perdigão’s Complex System Dynamics is now available as a semester doctoral course at the University of Lisbon, offering 6 ECTS for students enrolled in partner programs.

Objectives

  • Acquisition of fundamental competences in complexity sciences, their relevance and implementation in the conceptualization, systematization, modeling and formal analysis of the complex dynamics underlying climate change;
  • Learning fundamental principles that allow to formulate the dynamics of complex systems, including emergence of extreme phenomena, in an elegantly simple and effective way without loss of rigor nor generality;
  • Deepening scientific research, development and communication at the interface between natural and social frontier sciences.

Program

  • Fundamental notions on the dynamics of complex systems, principles and underlying mechanisms in dynamic systems theory and physical information;
  • Methods of systematization of dynamic systems: simple conceptual structures representing complex natural, technical and social phenomena;
  • Fundamentals of the dynamics of the Earth system and the emergence of regimes, critical transitions and extreme events in the context of complexity sciences;
  • Coevolutionary models of climate change in a holistic perspective involving dynamics of the oceans, atmosphere, geosphere, biosphere and society;
  • Dynamic methods of extraction and analysis of information related to the dynamics of complex systems, from empirical and computational records;
  • Detection of patterns of spatial and temporal climatic variability from data of the dynamics of the Earth system and attribution to underlying mechanisms;
  • Methods of evaluating uncertainty and predictability in complex system dynamics, for representative model optimization and decision support.

Block Course at Karlsruhe Institute of Technology (KIT)

Combining Modern Statistical Methods and Thermodynamic Principles
for Dynamic System Analysis in Environmental Sciences

Analysing complex and dynamic environmental systems often reaches technical and theoretical limits of statistical standard tools. This is the case as most classical statistical approaches fundamentally require assumptions about statistical independence and stationarity, which are especially problematic as soon as we deal with non-linearity and coevolution. Many promising approaches founded in scientific fields ranging from theoretical physics to information theory have been developed. Still, their application to applied climate and environmental sciences remains challenging.

This 4-day block course will introduce a set of modern tools for rigorous analyses of dynamic systems. Starting off with examples from geophysics and fluid mechanics the course will guide you during an excursion through stochastic physics, information theory, phase-state analyses and applications with big data and scaling. We especially invite you to bring own data and application questions to provide hands-on utilisation of the concepts and tools within your field of research.

 

Programme

Day 1 – From fluid mechanics to dynamical systems

The first day ensures a common basis for all participants. With fundamentals from classical fluid dynamics, thermodynamics, stability and scaling laws the foundation is laid.
The day will consist of four teaching units with practical analytical and numerical examples across the earth sciences.

Day 2 – Coevolutionary dynamical systems

The second day will extend the classical fluid dynamics with stochastic physics and information theory. With this, complexity will be rigorously treated in a simple and coherent framework providing the physical background to coevolutionary dynamics and organisation.

The day will consist of two teaching units and two real-world application cases.

Day 3 – Mastering conjugated dimensions

The third day will redefine the state-space system analyses with physical princi- ples and thermodynamic limits as true phase-state analyses towards dynamic system understanding without the requirement for single attractors, finite pha- ses and fixed scales. It also includes conjugated pairs and a new breed of non- paired conjugates.

The day will consist of two teaching units, one unit with real-world applications and one unit transferring the approaches to own analyses of the participants. Evening: Preparation of applications to own data.

Day 4 – Presentations and frontier topics

The fourth day will start with two units presenting first results of applications to own data. Subsequently frontier topics of scaling and big data applications will be tackled.

Evening: Discussion of possible collaboration and emerging aspects.

 

Lecturer: Rui A. P. Perdigão

[See related courses by Rui Perdigão e.g. on Information Physics]

Schedule: 27 February – 2 March, 2018

Hosted by:

GRACE – Graduate School for Climate and Environment
Engesserstr. 6, 76131 Karlsruhe – T+49-721-608-43676 – F+49-721-608-48475

www.GRACE.kit.edu

Karlsruhe Institute of Technology (KIT)

www.kit.edu