CoFI Roadmap#
This page contains both short-term and long-term goals of the CoFI project, and is a live snapshot of what we plan to invest our time and effort in.
Solver libraries#
Expanding the breadth of our supported solver libraries is one of the top priorities. Our initial targets are listed in the solver’s tree you’ve already seen in the home page.
Example gallery#
This is another prioritised task to be done. Here’s a list of examples we’d like to add within the next few months:
Shaw problem
Receiver function
Forward simulation with PyGIMLI (2D ERT, 2D DCIP, etc.)
Forward simulation with SimPEG (gravity, electromagnetics, etc.)
(to be continued)
User interface#
We aim to find a relatively stable public API that guarantees user experience and
extensibility. The current status of CoFI includes three objects that represent
inference problem (BaseProblem), inference options (InversionOptions) as well as a
“running engine” (Inversion). Additionally, the actual library wrappers
(subclasses of BaseInferenceTool) are exposed to more advanced users for them to make it
easier to plug in their own solver or do some hacking.
The details of the APIs mentioned above will be tested by first group of users and adjusted according to their feedbacks.
Suggesting system#
For the time being, CoFI’s InversionOptions has a minimal suggesting system that
categorises all the backend tools with a layer of solving method. However, such a
flattened list of solving methods does not reflect much about the relevance and
difference between different tools.
Thoughts about a better suggesting system is under an initial stage, and we want to suit three types of audience: people who are certain about which solving approach they want to use, people who just wants to use CoFI’s wrapper on a certain tool, and others who are not sure about how to solve their inversion problem so need more advice.
Documentation#
The following documentation pages are planned to be written or expanded:
Tutorials section, where guides are step-by-step and compact enough to present users with common use cases and best practices with CoFI.
Inference tools library section - this will be a brand new part of CoFI’s documentation, starting from the solver’s tree you’ve seen in home page to expand the details of each inference methods. This part will focus on theories instead of technical details.
Frequently asked questions (FAQs)
Education materials#
Workshops and talks are to be arranged ideally more often in 2024, once we have a relatively stable API and richer set of examples established. These will be in the field of both geoscience and more general scientific open source community.
Additionally, a set of course materials on inference theory is going to be developed with CoFI as its main tool of code demonstration.