Welcome to the ACBICI documentation!

ACBICI, A Configurable BayesIan Calibration and Inference package is designed to provide a modular, compact, and reconfigurable python package that can be employed to calibrate (complex) models and quantify their uncertainty. The development of this package started in 2023 at the Computational Solid Mechanics group <https://www.materials.imdea.org/groups/csm/> of the IMDEA Materials Institute and it is still under development.

The calibration strategy behind ACBICI follows closely the seminal work of Kennedy and O’Hagan (2001). In a nutshell, this method uses Bayesian inference to determine the most likely probability distribution of one or more parameters in a known model, given some experimental data and prior knowledge. More specifically, the framework systematically exploits Gaussian processes as surrogates of expensive models and also as regressors of discrepancy errors.

In this manual, we describe the ideas behind ACBICI, we provide details on how to calibrate models, and provide examples of the most important types of calibration.

License

ACBICI is distributed under a BSD-3 licence.

Contact

ACBICI has been developed by Christina Schenk and Ignacio Romero, at IMDEA Materials Institute, Spain. The library is provided as is and no warranty is offered on the results obtained. Questions, comments, suggestions, etc. can be addressed to christina.schenk@imdea.org or ignacio.romero@imdea.org or by creating an issue on the GitLab repository.

Contents:

• Installation
• Model calibration: general concepts
  • Models
  • Calibrating a model
  • GUI
  • Outputs from the calibration
  • Convergence
  • Multi-Output Calibration
• Preparing a calibration for ACBICI
  • Defining the model
  • Experimental data
  • Prior information
  • Collecting experimental data
  • Prior information
  • Selecting calibration type
• Gaussian processes
  • Definition
  • Covariance and kernel
  • The use of Gaussian processes in calibration
• Implementing a calibration process
  • Defining a model
  • Defining a calibration process
• Examples
  • Example 0: using ACBICI for the first time
  • Example 1: a one parameter model
  • Example 2: a two-parameter model
  • Example 3: Cobb-Douglas Production model
  • Example 4: Traction test
• src
  • src package

Indices and tables

• Index

• Module Index

• Search Page

External references

• Kennedy, M. C. and O’Hagan, A. (2001). Bayesian calibration of computer models, Journal of the Royal Statistical Society: Series B. (Statistical Methodology), 63, 425–464.

• Carmassi, M. and Barbillon, P. and Chiodetti, M. and Keller, M. and Parent, E. (2019). Bayesian calibration of a numerical code for prediction, Journal de la Société Française de Statistique, 160, 1-30.

• Higdon, Dave and Kennedy, Marc and Cavendish, James C. and Cafeo, John A and Ryne, Robert D. (2004). Combining Field Data and Computer Simulations for Calibration and Prediction, SIAM Journal on Scientific Computing, 26, 448–466.

• de Pablos, J.L., Sabirov, I. and Romero, I. (2023). A methodology for the calibration of complex material models: application to thermo-elasto-plastic materials for high-velocity impact simulations, Archives of Computational Methods in Engineering, 30, 2859-2888.

• Rasmussen, C. E. and Williams, C. K. I. (2006). Gaussian Processes for Machine Learning, MIT Press, Cambridge, Massachusetts.