Why does accumulated error from momentum violation matter more in some systems than others?

This is a thoughtful question about error accumulation in dynamical systems, but the preprints provided don't directly address momentum violation or its system-dependent consequences. Let me explain what they *do* reveal about related error mechanisms, which may illuminate your underlying concern.

Error sensitivity depends on conserved quantities and operator structure

The closest relevant insight comes from work on how violations of conservation laws propagate differently depending on the system's geometry. [AG-2024.03-654] shows that using an effective potential in dynamics leads to explicit energy non-conservation, and this failure is not uniformly bad—it catastrophically breaks down in specific regimes like parametric amplification and spinodal instabilities, where "profuse particle production is directly linked to the failure." In contrast, quasi-static approximations work when change happens slowly. The lesson: error accumulation from a conservation-law violation matters most when the system's natural timescale is short relative to the violation's growth rate.

[AG-2024.03-363] makes a related point for open quantum systems: two Lagrangians that differ only by a total derivative (and thus encode identical physics in isolation) produce inequivalent reduced dynamics once you trace out an environment. The error from using the "wrong" formulation only becomes operationally visible when momentum conservation breaks between system and environment—suggesting that error significance depends on whether conserved quantities can leak into hidden degrees of freedom.

[AG-2024.05-2401] establishes a fundamental speed-accuracy trade-off: measurements cannot achieve zero error in finite time for observables that don't commute with the Hamiltonian. This implies that momentum-violating errors accumulate fastest in systems where momentum is far from conserved by the underlying dynamics—i.e., in driven or non-Hamiltonian systems rather than closed ones.

Why this matters

Accumulated error from momentum violation is thus amplified in systems with short characteristic timescales, strong coupling to unmeasured degrees of freedom, or driving fields that couple strongly to momentum eigenstates. It matters less in adiabatic or weakly-coupled regimes.

The provided preprints do not directly address momentum violation or how its error budget varies across propulsion or relativity contexts.