Answers are grounded in this paper and cite their sources.
The mechanism by which a lattice index changes is not in dispute: when an eigenvalue of the Hermitian kernel crosses zero the overlap charge moves with it, and this spectral-flow picture has been standard since the operator was introduced [5, 6]. What has not been measured is how the amplitude at which this happens depends on system size. This paper measures it, and reports what the measurement refuses to support. Holding one phase pattern fixed and raising the link-noise amplitude in steps of 0.01π, we follow eigenvalues rather than sample configurations: over 40 patterns and 152 crossings the charge moves by one unit in 145 events and by two in 7, the median first crossing sitting at 0.350π against an ensemble half-way amplitude of 0.349π at the same size. At eight sizes from L=8 to L=64 the half-way amplitude obeys ε50=(0.2913±0.0023)π+(0.919±0.057)/L with χ2 of 0.95 per degree of freedom, against 2.17 for a logarithmic form in L and 2.54 for a logarithmic form in the level spacing; the extrapolated limit is 0.291π. A direction check removes the natural competing scale: the clean band edge rises by 42.5 per cent over the same range while the threshold falls by 24.9, so no clean spectral scale of this model tracks it. Two further measurements narrow what the object is. First, delivery does not fall to zero: at L=16 the fraction still returning the clean value is 0.15, 0.18 and 0.23 at 0.45π, 0.475π and 0.50π (100 realisations each), a rise that is 2.1σ by exact binomial test (uncorrected), while at fixed amplitude the fraction falls monotonically with size. The half-way amplitude therefore marks a crossing of one half, not a threshold. Second, the standard discriminator between a crossover and a critical point fails: a two-parameter logit collapse with the fitted midpoint is rejected at χ2 of about 106 per degree of freedom over 48 points, so neither reading is supported by these data. Five negative results are recorded. The claims are: a unit granularity of failure, an inverse-size drift with a finite extrapolated limit, a residual fraction that is not a floor, and a set of structural dead ends. The claims that are not made: no thermodynamic limit, no universality, no statement about the order of any transition, and no new mechanism.
No takes yet. Share an insight, caveat, or question.
GenMin Li (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: