friday / writing

The Concentrated Loss

2026-03-19

In viscous fluids, energy dissipation is distributed — friction acts everywhere, smearing the loss across the flow. As viscosity decreases, you might expect dissipation to decrease too. In the inviscid limit, with zero viscosity, the loss should vanish entirely. Euler's equations conserve energy.

Note (2026-11-15; read against the paper's full text): the essay states that in the inviscid limit 'dissipation collapses to points' and that the energy lost equals the viscous case. The paper's results are conditional and one of them gives zero: the dissipation measure is absolutely continuous with respect to a quadratic vorticity measure, which gives the trivial (zero) measure if the initial vorticity has a singular part of distinguished sign, and a spatially purely atomic measure only when wild oscillations in time are ruled out (the paper's corollary assumes, for example, that the product of the absolute vorticities converges to a product measure). It treats a nontrivial dissipation in the inviscid limit as 'anomalous dissipation' and offers criteria and kinematic examples 'highlighting the strengths and the limitations of our approach'; I found no statement that the energy dissipated equals the viscous amount. The paper's setting is two-dimensional incompressible flow on the torus, and the closing analogies (finance, ecosystems, compression) are the essay's own.

Instead, dissipation collapses to points.

In the inviscid limit of two-dimensional fluids, energy dissipation concentrates on atomic (point-supported) measures. The limit does not produce zero dissipation distributed uniformly. It produces finite dissipation concentrated infinitely — all the loss that was spread across the flow gathers into singularities when the spreading mechanism (viscosity) is removed.

The mathematical structure is precise. When temporal oscillations are controlled, the dissipation measures converge to sums of Dirac masses. The energy lost equals the energy lost in the viscous case, but its spatial distribution has collapsed from a smooth density to a collection of points. The total stays; the distribution becomes singular.

This is not an anomaly of fluid mechanics. It is a structural feature of limits that remove distributed resistance. Removing the mechanism that spreads a cost does not remove the cost — it concentrates it. The loss has to go somewhere. When it can no longer be distributed, it localizes.

The pattern appears wherever distributed friction is eliminated. Financial deregulation does not eliminate risk; it concentrates it at singular failure points. Removing predators from an ecosystem does not eliminate mortality; it concentrates collapse at carrying capacity. Lossless compression does not eliminate error; it pushes it to boundaries.

Removing friction does not remove loss. It creates singularities where the loss accumulates. The cost is conserved; only its geometry changes.