Answers are grounded in this paper and cite their sources.
The Lorentzian no-boundary path integral with Dirichlet data at zero size weights cosmological perturbations with an inverse Gaussian, and the defect is contour-independent; implementations that fix an initial Euclidean momentum instead render the integrals manifestly convergent and stable, at the price of abandoning the sum over compact geometries off shell. We proposea third option: remove a = 0 from the ledger of data surfaces altogether. Fluctuation data is posed at the de Sitter throat, q0 = 3/Λ — the unique member of the Dirichlet family at which, in the review’s own saddle formula, the conjugate lapse pair coalesces onto the real axis. The point a = 0, a branch point of the perturbation equation, becomes an interior pointof a smooth Euclidean cap that functions as state preparation; regularity there is a normalizability condition, not endpoint data, and it selects Bunch–Davies over the enhanced mode by eleven orders of magnitude. Compactness is retained as a property of the geometry, regularity as a property of the state, and no surface carries two conditions. The one open question — manifest convergence of the lapse integral at the coalescence fold — is stated in the terms of the programme equipped to answer it.
No takes yet. Share an insight, caveat, or question.
James Laurence Williams (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: