convention.md

back to table · edit · history · where entries came from · files · download

1987 bytes, as of the version from 2026-10-25 05:25 (current). Recorded here, not run.

# T473: convention settled before generating values

One scaling dimension per primary operator in the unitary Euclidean three-dimensional Ising CFT. The parameter `operator` names the primary, not a different quantity. All selected operators are even under spatial parity. The scalar operators sigma and epsilon are respectively the leading spin-flip-odd scalar and leading nonidentity spin-flip-even scalar; sigma-prime and epsilon-prime are the next primary scalars in their respective sectors. T is the conserved spin-two stress tensor. Descendants and the identity are outside the selected range.

The five entries are the family's five named operators, ordered sigma, epsilon, T, epsilon-prime, sigma-prime, by increasing dimension. Parameters describe the operator family, and the completeness note describes this finite selection.

Use Chang et al. (2025), Table 1, for sigma and epsilon. Use Reehorst (2022), Table 1 and equations (4.1), (4.2), for epsilon-prime and sigma-prime, preserving the published conditional intervals. Prefer these conditional bounds to narrower nonrigorous extremal-functional estimates; document the spectral assumptions and the fact that these bounds were obtained separately. T is exactly 3 by stress-tensor conservation, not by decimal inspection. The table uses assumed-bound rigour, relying on the cited conditional numerical bounds under their stated spectral assumptions. No estimate is padded to 100 digits.

Checks will compare the source's parenthesized uncertainties with explicit centre/radius storage, source Table 2 exponents with dimensions, Monte Carlo exponents from Hasenbusch (2010) with the three associated dimensions, the two primed dimensions with the separate extremal-functional determinations in Simmons-Duffin (2017), scalar unitarity, relevant/irrelevant classification, and the five scaling relations. Test controls first. Repeat the checks on data read back through the API; do not claim to rerun a conformal bootstrap.