The Euler equations describe the flow of an ideal fluid (one with no viscosity). A "steady" flow is one that doesn't change over time. Back in 1965, the mathematician Vladimir Arnold proved a landmark result: in three dimensions, steady Euler flows with varying pressure must have a special geometric structure, where the fluid domain is layered like an onion into nested surfaces that the fluid flows along. This paper asks a natural follow-up question: exactly what kinds of layered structures are actually allowed?
The authors focus on a region of fluid shaped like a donut (a toroidal domain) where the layering is organized by a function called the flux function. They prove that any smooth steady Euler flow consistent with this setup can always be written as a combination of two basic flow patterns on each surface, with the mixing coefficients depending only on which surface you're on. Think of it like decomposing any melody into two fundamental notes, where the balance between the notes can shift from layer to layer but the notes themselves are determined by the geometry. They also show that the flux function must satisfy a single governing equation, which they call the normal flux equation, that ties everything together.
The result is satisfying because it unifies and generalizes known special cases. The most famous of these is the Grad-Shafranov equation from plasma physics, which describes magnetic confinement in a tokamak (a donut-shaped fusion reactor) under the assumption of axial symmetry. The new framework recovers that equation as a special case but applies far more broadly, to flows without any particular symmetry. This gives a cleaner and more complete picture of the geometric constraints that steady fluid flows must obey.