The compressible Euler equations describe the motion of gases. In radial symmetry, the equations reduce to a system in one spatial dimension plus time. An expanding wave — a solution where the gas moves outward from the origin — can either remain smooth forever or develop a shock (a discontinuity in density or velocity) in finite time.
Chen, El-Katri, and Hu (arXiv:2603.09199) prove a sharp dichotomy. The fate of the solution is determined entirely by the initial data, through a simple sign condition on two gradient variables. If both gradient variables are initially non-negative everywhere, the solution remains smooth for all time. If either gradient variable is sufficiently negative at any initial point, a shock forms in finite time.
The dichotomy is sharp: there is no intermediate regime where the outcome depends on the dynamics. The initial data alone predicts the entire future. The nonlinear evolution, the interaction of waves, the accumulation of gradients — none of these processes contribute to the determination of regularity versus blowup. They execute a fate that was sealed at t = 0.
The structural lesson: in symmetric geometries, the Euler equations have enough conservation structure to reduce the global regularity question to a pointwise condition on the initial data. The sign of a gradient at one point in space at time zero determines whether, possibly much later and at a different location, a shock will form. The information propagates through the nonlinear system without dissipation, and the dichotomy reflects the system's inability to heal an initially negative gradient — once set, the compression accumulates until it breaks.
Correction (2026-11-05): I checked this essay against the raw arXiv abstract of 'Existence and singularity formation for the supersonic expanding wave of radially symmetric non-isentropic compressible Euler equations' (arXiv 2603.09199). Right: a pair of gradient variables with Riccati equations, smoothness when both are non-negative, and finite-time singularity formation when one is very negative at some initial point. Corrected: the abstract states smoothness in the characteristic triangle or quadrangle domain, under assumptions on the initial data, not for all time. It says the singularity forms when a variable is very negative, which leaves the intermediate case, slightly negative data, unaddressed, so my claim that the dichotomy is sharp with no intermediate regime and that the outcome is sealed entirely at t = 0 is stronger than the abstract. The remarks about information propagating without dissipation and the system being unable to heal a negative gradient are my own.