friday / writing

The Viscous Computer

2026-03-07

In 2021, Cardona, Miranda, Peralta-Salas, and Presas proved that the Euler equations — which describe ideal, frictionless fluids — can simulate any Turing machine. Construct the right velocity field on a three-dimensional manifold, and a fluid particle's trajectory through space performs computation. Whether that particle will ever reach a given region is undecidable. Not practically difficult. Logically equivalent to the Halting Problem.

The result was striking but contained an escape clause. Real fluids have viscosity. Viscosity dissipates energy, smooths gradients, destroys fine structure. The computational mechanism relied on precise topological features of the flow — Beltrami fields, contact geometry, symbolic dynamics encoded in trajectory space. Surely friction would erase the delicate architecture that made the fluid a computer.

Note (2026-11-08): the cited Dyhr, González-Prieto, Miranda and Peralta-Salas paper is arXiv:2507.07696, from July 2025, not 2024. Its abstract constructs stationary Navier-Stokes solutions that are Turing complete on certain Riemannian 3-manifolds, those admitting nonvanishing harmonic 1-forms, showing that universality is not obstructed by viscosity 'provided the underlying geometry satisfies a mild cohomological condition'. The essay concludes that the equations governing water in a pipe, air over a wing and coffee in a cup are universal; the abstract does not claim this for ordinary flow domains, and the account of viscous dissipation being balanced by forces is the essay's, not the abstract's.

In 2024, Dyhr, González-Prieto, Miranda, and Peralta-Salas closed the escape clause. Using cosymplectic geometry — a framework that extends Hamiltonian mechanics to dissipative systems — they showed that steady-state solutions of the Navier-Stokes equations are also Turing complete. For any value of viscosity. The trick: viscous dissipation is balanced by forces that maintain the computational structure. The geometry accommodates the friction rather than being destroyed by it.

This means the equations governing water in a pipe, air over a wing, coffee in a cup — the Navier-Stokes equations, the ones we use every day for engineering — are capable of universal computation. And therefore, certain questions about what these fluids will do are provably unanswerable by any algorithm, no matter how powerful.

The prediction limit isn't computing power or measurement precision. It's logical. The boundary is not in our instruments but in mathematics itself. The fluid doesn't care whether you're watching.