Identifiability of linear stochastic state-space models with application to ecology

Frédéric Barraquand, Julien GibaudInstitute of Mathematics of Bordeaux, University of Bordeaux,CNRS, Bordeaux INP, Talence, France · 2025

Read Identifiability of linear stochastic state-space models with application to ecology by Frédéric Barraquand, Julien GibaudInstitute of Mathematics of Bordeaux, University of Bordeaux,CNRS, Bordeaux INP, Talence, France (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-08714.txt.

Description

Barraquand and Gibaud settle a theory question that hangs over ecological modeling: when a state-space model fails to fit, is the model itself at fault, or just the data? These models pair a hidden process, such as true animal numbers, with imperfect observations of it. Earlier checks for whether parameters can be recovered used expected values only, leaving the noise parameters untested. The authors build a complete summary from the spectrum of the observed process, capturing means and variances together.

Applying their test to a range of ecological models, they find the models are generally identifiable in theory. Problems arise when whole compartments of the system go unobserved. The practical moral is that most fitting failures are likely practical — too little or too noisy data — rather than proof the model can never work. For ecologists, the paper separates two kinds of trouble that are often confused.

Catalog record

CREATORFrédéric Barraquand, Julien GibaudInstitute of Mathematics of Bordeaux, University of Bordeaux,CNRS, Bordeaux INP, Talence, France
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE59 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-08714

Browse the full library · About the Tower of Babel