Regensburg 2022 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 5: Social Systems, Opinion and Group Dynamics

SOE 5.3: Vortrag

Montag, 5. September 2022, 13:00–13:15, H11

Multiscale Causal Structure in Armed Conflict — •Niraj Kushwaha and Edward Lee — Complexity Science Hub Vienna, Josefstädter Straße 39, 1080 Vienna, Austria

Armed conflict is a major and ongoing problem around the world today. The very features of conflict that makes it important, its multiscale and multidimensional impact, render it difficult to understand quantitatively. Here, we introduce a first principles approach to conflict by clustering sequences of conflict events into causal conflict avalanches. We rely on armed conflict data from the ACLED project, occurring from the years 1996-2021 in Africa. We investigate different spatial and temporal scales with a systematic coarse-graining procedure. For space, we tile the region with semi-regular bins that constitute our level of resolution, and for time we group days into discrete intervals. This formalism bridges the gap between microscopic and macroscopic descriptions of armed conflict. To infer causal relationships between different spatial bins we use a standard nonlinear measure of statistical dependency called transfer entropy. Using transfer entropy, we extract a directed causal network that links adjacent geographic locations. We then leverage this causal structure to join two conflict events if they are adjacent in time and belong to adjacent and causally connected spatial bins. We call the resulting sequences of conflict events, conflict avalanches. Further statistical and dynamical analysis reveal that many conflict avalanche features follow power-law distributions. Our work paves way for future investigation on classes of models which can explain the emergence of scaling in conflict across different scales.