Comparing data assimilation and likelihood-based inference on latent state estimation in agent-based models
PNAS Nexus · 30.04.2026 · 5 (5)
DOI: 10.1093/pnasnexus/pgag161 ↗Kurz gefasst
Fachartikel in der Zeitschrift PNAS Nexus, online erschienen am 19.05.2026. Verfasst von Blas Kolic und 3 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Big Data (IBiDat), Universidad Carlos III de Madrid, Ronda de Toledo, 1, Madrid 28005, Spain. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
In this article, we present the first systematic comparison of data assimilation (DA) and likelihood-based inference (LBI) in the context of an agent-based model (ABM). These models generate observable time series driven by evolving, partially latent microstates. Latent states must be estimated to align simulations with real-world data, a task traditionally addressed by DA, particularly in continuous and equation-based models used in weather forecasting. However, the nature of ABMs poses challenges for standard DA methods. Solving such issues requires adapting previous DA techniques or using ad hoc alternatives such as LBI. DA approximates the likelihood in a model-agnostic way, making it broadly applicable but potentially less precise. In contrast, LBI provides more accurate state estimation by directly leveraging the model’s likelihood, but at the cost of requiring a hand-crafted, model-specific likelihood function, which may be complex or infeasible to derive. We compare the two methods on the bounded-confidence model, a well-known opinion dynamics ABM, where agents are affected only by others holding sufficiently similar opinions. We find that LBI better recovers latent agent-level opinions, even under model misspecification, leading to improved individual-level forecasts. At the aggregate level, however, both methods perform comparably, and DA remains competitive across levels of aggregation under certain parameter settings. Our findings suggest that DA is well-suited for aggregate predictions, while LBI is preferable for agent-level inference.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Institute of Big Data (IBiDat), Universidad Carlos III de Madrid, Ronda de Toledo, 1, Madrid 28005, Spain.
- CENTAI, Corso Inghilterra, 3, Turin 10138, Italy.
- Intesa Sanpaolo Innovation Center, Corso Inghilterra, 3, Turin 10138, Italy.
ORCID-Profile: Blas Kolic, Corrado Monti, Gianmarco De Francisci Morales, Marco Pangallo
Zitieren
APA
Kolic, B., Monti, C., De Francisci Morales, G., & Pangallo, M. (2026). Comparing data assimilation and likelihood-based inference on latent state estimation in agent-based models. PNAS Nexus, 5(5). https://doi.org/10.1093/pnasnexus/pgag161
Vancouver
Kolic B, Monti C, De Francisci Morales G, Pangallo M. Comparing data assimilation and likelihood-based inference on latent state estimation in agent-based models. PNAS Nexus. 2026;5(5). doi:10.1093/pnasnexus/pgag161
BibTeX
@article{Kolic2026,
author = {Kolic, Blas and Monti, Corrado and De Francisci Morales, Gianmarco and Pangallo, Marco},
title = {Comparing data assimilation and likelihood-based inference on latent state estimation in agent-based models},
journal = {PNAS Nexus},
year = {2026},
volume = {5},
number = {5},
doi = {10.1093/pnasnexus/pgag161}
}
EUVIMED