Use cases

Below are some expected surmise use cases that we support (or are working to support) and plan to have examples of:

  • A user wants to emulate a computationally expensive model at a limited number of points in parameter space and interpolate those results through the whole space. Then, a user can use one of the methods in the \emulationmethods directory.

  • A user wants to emulate a computationally expensive model at a limited number of points in parameter space and interpolate those results through the whole space using a user-developed emulation method. Then, a user can drop the script of the method into the \emulationmethods directory, and use it within the surmise framework.

  • A user wants to perform principled uncertainty quantification that calibrates the models against data. Then, a user can use one of the methods in the \calibrationmethods directory.

  • A user wants to perform principled uncertainty quantification that calibrates the models against data via a user-developed calibration method. Then, a user can drop the script of the method into the \calibrationmethods directory, and use it within the surmise framework.

  • A user wants to test a new sampler within the surmise framework. Then, a user can drop the script of the method into the \utilitiesmethods directory, and use it within the surmise framework.

Reference to examples

Below is a non-exhaustive list of references utilizing surmise (in alphabetical order):

[1]

Moses Y-H Chan, Matthew Plumlee, and Stefan M Wild. Constructing a simulation surrogate with partially observed output. Technometrics, pages 1–13, 2023. To Appear. doi:10.1080/00401706.2023.2210170.

[2]

Pablo Giuliani, Kyle Godbey, Edgard Bonilla, Frederi Viens, and Jorge Piekarewicz. Bayes goes fast: uncertainty quantification for a covariant energy density functional emulated by the reduced basis method. Frontiers in Physics, 10:1212, 2023. doi:10.3389/fphy.2022.1054524.

[3]

Stephen Hudson, Jeffrey Larson, John-Luke Navarro, and Stefan M. Wild. libEnsemble: a library to coordinate the concurrent evaluation of dynamic ensembles of calculations. IEEE Transactions on Parallel and Distributed Systems, 33(4):977–988, 2022. doi:10.1109/tpds.2021.3082815.

[4]

Dananjaya Liyanage, Özge Sürer, Matthew Plumlee, Stefan M Wild, and Ulrich Heinz. Bayesian calibration of viscous anisotropic hydrodynamic simulations of heavy-ion collisions. arXiv preprint arXiv:2302.14184, 2023. doi:10.48550/arXiv.2302.14184.

[5]

Özge Sürer, Filomena M Nunes, Matthew Plumlee, and Stefan M Wild. Uncertainty quantification in breakup reactions. Physical Review C, 106(2):024607, 2022. doi:10.1103/PhysRevC.106.024607.

[6]

Özge Sürer, Matthew Plumlee, and Stefan M Wild. Sequential Bayesian experimental design for calibration of expensive simulation models. Technometrics, pages 1–26, 2023. To Appear. doi:10.1080/00401706.2023.2246157.

[7]

Eugene Wickett, Matthew Plumlee, Karen Smilowitz, Souly Phanouvong, and Victor Pribluda. Inferring sources of substandard and falsified products in pharmaceutical supply chains. IISE Transactions, pages 1–16, 2023. doi:10.1080/24725854.2023.2174277.