Uncertainty in Biology

Volume 17 of the series Studies in Mechanobiology, Tissue Engineering and Biomaterials pp 271-299


Sloppiness and the Geometry of Parameter Space

  • Brian K. MannakeeAffiliated withGraduate Interdisciplinary Program in Statistics, University of Arizona
  • , Aaron P. RagsdaleAffiliated withGraduate Interdisciplinary Program in Applied Mathematics, University of Arizona
  • , Mark K. TranstrumAffiliated withDepartment of Physics and Astronomy, Brigham Young University
  • , Ryan N. GutenkunstAffiliated withDepartment of Molecular and Cellular Biology Email author 

* Final gross prices may vary according to local VAT.

Get Access


When modeling complex biological systems, exploring parameter space is critical, because parameter values are typically poorly known a priori. This exploration can be challenging, because parameter space often has high dimension and complex structure. Recent work, however, has revealed universal structure in parameter space of models for nonlinear systems. In particular, models are often sloppy, with strong parameter correlations and an exponential range of parameter sensitivities. Here we review the evidence for universal sloppiness and its implications for parameter fitting, model prediction, and experimental design. In principle, one can transform parameters to alleviate sloppiness, but a parameterization-independent information geometry perspective reveals deeper universal structure. We thus also review the recent insights offered by information geometry, particularly in regard to sloppiness and numerical methods.


Sloppiness Hessian Experimental design Bayesian ensembles Cost functions Information geometry