Explanation

The concepts behind the interfaces, their notation, and the reasons for the choices visible in the API. The pages assume linear algebra and numerical computing, not MRI reconstruction or convex optimization; terms from either are introduced where first used. Exact interfaces are in API reference, and complete workflows in the examples.

Page	Question
Interfaces and execution	Which interface fits a task, and what runs underneath it?
Data layout and conventions	How are arrays, axes, trajectories and Fourier transforms laid out, why, and how is data from elsewhere converted into that layout?
Inverse problems and their solvers	What is estimated, from what, and by which algorithm?
The MRI encoding operator	What does the MRI forward operator consist of, and how is it represented?
Non-Cartesian sampling	How is the Fourier transform computed off the Cartesian grid, and what is its normal operator?
Nonlinear forward models	What changes when the forward operator is nonlinear in the unknowns?
Differentiation through reconstruction	How do gradients pass through operators, solvers and unrolled iterations?
Learned reconstruction	Where does a network enter a reconstruction, and how is it trained and applied to large data?