Direkt zum Inhalt | Direkt zur Navigation

Benutzerspezifische Werkzeuge
Sektionen

Forschungsprojekt: MoSiLLDe

  Modeling and Simulation of Linked Lives in Demography

Zusammenfassung: Entwicklung einer domänenspezifischen Modellierungssprache und effizienter Simulationsalgorithmen zur Erstellung von Modellen sowohl auf daten- als auch hypothesengetriebene Weise. In Zusammenarbeit mit dem Max-Planck-Institut für demografische Forschung werden dabei Modelle zur Analyse von Migration zwischen Afrika und Europa ausgearbeitet.

  Projektinformationen
Laufzeit: 01.11.2014 bis 30.10.2016
Wissenschaftl. Bearbeiter: Dipl.-Inf. Alexander Steiniger; M.Sc. Tom Warnke
Studentische Hilfskräfte: Sebastian Plath; Oliver Reinhardt
Auftraggeber: DFG

Continuous-time microsimulation is an important method for predicting demographic changes
in the next decades. However, these modeling and simulation studies face various challenges.
One is to integrate in an adequate manner dynamics at different organizational levels, i.e. at
micro level (e.g., individuals), and at macro level (e.g., population), and linking those dynamics
in continuous time and discrete space. During the last decade, demography has started moving
from data- to hypothesis-driven development of individual-based models. This has an impact on
processes and methods used for validating these models. The project aims at addressing chal-
lenges of continuous time micro simulation in demography and will provide contributions in the
area of (simulation) modeling languages, efficient simulation, and guidance and documentation
of simulation studies.
• A modeling language shall be developed that allows a succinct modeling of life courses
of individuals and populations, and link those life courses in continuous time and discrete
space. It will be realized within the modeling and simulation framework James II.
• Due to enabling a compact description of an individuals life course and decisions over
time and space and the demand to simulate millions of individuals, efficient execution
algorithms for the modeling language are needed. To address the need for an efficient
simulation a variety of approaches to speed up model execution shall be explored, i.e.,
from simulator configuration on demand, approximative schemes, to parallel algorithms.
• The modeling language shall allow to define hypotheses about what determines the life
course of individuals in a flexible manner. To validate these hypotheses experiments
different from the typical data based ones are needed. Based on the portfolio of methods
provided in James II workflows shall be defined that support data- and hypothesis-driven
modeling and experimentation.
• A set of smaller case studies shall be developed to evaluate the methods. In addition,
in cooperation with partners from demography, the modeling language will be used to
describe, explain and predict international migration using individual-level data.




Projektstruktur