Answers are grounded in this paper and cite their sources.
For the configuration model with a degree sequence whose third moment converges, write εₙ=1-νₙ, where νₙ→1 from below is the size-biased mean minus one. Van der Hofstad asked to show that in the barely subcritical regime the largest component has order maxₘₐₓ/εₙ, εₙ⁻²log(nεₙ³)\, and predicted that it has size (dₘₐₓ/εₙ)(1+oP(1)) when εₙ tends to zero sufficiently slowly. We show that neither holds in this generality. We use degree sequences with vertices of degree one and three and a polynomial number of vertices of a large common degree d. Paths of K such hubs joined by direct edges then appear, and the component of such a path is a tree with about Kd/εₙ vertices. Taking $K=2$ gives L₁≥(2-η)dₘₐₓ/εₙ with high probability, for εₙ tending to zero as slowly as we like; letting K→∞ gives L₁/maxₘₐₓ/εₙ,εₙ⁻²log(nεₙ³)\→∞ in probability. In all examples the degree of a uniform vertex converges in law to a variable on \1,3\ with all three moments, and nεₙ³→∞.
No takes yet. Share an insight, caveat, or question.
Yitzchak Shmalo (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: