post image August 31, 2022 | 2 min Read

IAMG 2022 Poster Presentation

Date: 29 Aug - 3 Sept, 2022

Venue: Nancy, France

alt text

Title: CoFI - Linking geoscience inference problems with tools for their solution


Inference problems in the geosciences vary significantly in size and scope, from 1D regression of geochemical data, to building 3D models of Earth’s interior structure. As such, the combination of datasets and the approaches used for solving are often highly problem-specific. Some researchers develop inversion and parameter estimation algorithms that suit their needs and wonder where else they could be applied. Others focus on developing new and flexible forward models of Earth phenomena and want a complementary inversion algorithm that fits their needs. Many more of us have data and questions to answer and need both.

Our aim with CoFI, the Common Framework for Inference, is to build a community collaboration using open-source platform agnostic infrastructure to interface geoscience inference problems with the tools required to solve them. In addition by creating an extensive set of demonstration problems and worked examples, we hope to support education at all levels in this space. CoFI will initially take the form of a Python package able to interface with inference algorithms from different sources. In addition, we are building an Inversion Test Suite which collects examples from many disciplines within geoscience and elsewhere. The connection points between the two is a common standard that is flexible and capable enough to accommodate a variety of inference problems which can be recognised by various solvers that comply with this same common framework.

Having the basic setup and several examples in place already, we hope to encourage more forward examples and inversion algorithms to be developed and be contributed to this modular designed framework, with the lowest possible barriers to participation.

Begin exploring the development version of CoFI by heading to Colab