Aircraft routing through wind fields is an optimization problem: find the path that minimizes fuel consumption (and therefore emissions) given that wind speeds and directions vary across space. The standard approach discretizes the airspace into a graph and searches for shortest paths. The discretization limits the solution — the optimal path may not align with the graph edges.
Note (2026-11-14; read against the paper's full text): the essay describes an ordinary eikonal wavefront with the wind folded into an effective speed, and a path that is the ray perpendicular to the cost front. The paper says the free-flight problem is not isotropic, because the wind makes the speed depend on direction, so it needs a generalized eikonal equation (a Hamilton-Jacobi-Bellman equation); the standard isotropic eikonal methods it cites do not apply. Its model is two-dimensional and horizontal, with constant airspeed, a stationary and continuously differentiable wind field, and the objective is minimum travel time; it treats fuel and emissions as proportional to flight duration, not as a fuel model. The guarantee is that, for a discretization error bounded by an arbitrarily small epsilon, destinations outside a temporal trust region of size 2 epsilon around the singularities have a provably good approximation of the unique globally optimal trajectory, and the paper reports computational results on test wind fields such as a single vortex. I read the essay's 'safety zones' and 'cut loci' as this trust region and these singularities.
The eikonal equation provides a continuous alternative (arXiv:2603.11830). In optics, the eikonal equation describes how wavefronts propagate through media with varying refractive index. The travel time from a source to any point is the solution of a partial differential equation. Replace the refractive index with the effective ground speed (airspeed plus wind) and the wavefront becomes a fuel-cost front expanding from the departure airport. The optimal path to any destination is the ray perpendicular to the cost front — the continuous analog of Dijkstra's algorithm without discretization.
The mathematical subtlety lies at cut loci — boundaries where wavefronts from different directions meet. Near a cut locus, two nearly-optimal paths converge from different directions, and small numerical errors can cause the algorithm to follow the wrong one. The authors use finite element error analysis to establish safety zones around cut loci, guaranteeing that for destinations sufficiently far from these boundaries, the computed path is globally optimal.
The eikonal formulation turns flight routing into wave propagation. The optimal path is not computed by searching; it is traced by following the gradient of a field that already encodes all possible routes.