Conditional Continuum Limit of 4d SU(Nc) Yang-Mills Theory via Two-Layer Architecture, RG-Cauchy Uniqueness, and Step-Scaling Confinement

Strengths

• Exemplary transparency regarding conditionality: Section 1.5, Section 8.1, Section 8.2, and Section 9 clearly document all unproved assumptions, imported hypotheses, and errata from earlier versions.
• Honest identification of the summability obstruction in Remark 3.6 and Appendix B, demonstrating explicitly that the naive 2-loop asymptotic-freedom rate g_k^2 ~ c/k diverges logarithmically upon summation and thus cannot discharge Assumption 3.5.
• Lemma 1.4 provides a clean, quantitative verification that for a fixed physical box size L, the dyadic lattice size L/a_k = 2^k L/a_0 grows slower than the exponential volume window exp(c_w/g_k^2) along the 2-loop trajectory.
• Clear distinction in Appendix A (Remarks A.1 and A.6) between the spectral gap of Glauber/Langevin dynamics controlled by log-Sobolev inequalities and the Euclidean transfer-matrix spectral gap required for the physical mass gap.

Weaknesses

• Once the six assumptions of this paper and the six imported lattice hypotheses are postulated, the main theorems (Proposition 3.8, Theorem 3.9, Theorem 4.11) reduce to immediate one-line consequences of the assumptions (telescoping of an assumed summable series and preserving a lower bound >= sigma_0 under an assumed limit).
• The bridge in Appendix B only yields a non-summable O(g_k^2) bound (Eq. 35), leaving Assumption 3.5 unsubstantiated even conditionally on the companion lattice chain.
• Layer 2 (Section 4) does not use any output of Layer 1 (Section 3); the existence of the dyadic continuum limit of Creutz ratios is simply assumed in Assumption 4.4 rather than derived from the RG-Cauchy framework.
• Proposition 5.3 on SO(4) rotational invariance is purely heuristic and defers renormalized Schwinger functions, leaving the continuum state with only hypercubic symmetry.

Potential errors

• In Lemma 6.1 and Lemma A.3, the uniform clustering rate m_0 along the lattice coordinate axes is asserted to hold unchanged for arbitrary Euclidean separation d via a chessboard estimate, whereas along a 4D lattice diagonal the coordinate separation is d/2, reducing the axis transfer-matrix decay rate by up to a factor of 2 unless full SO(4) rotational invariance is already established.
• In Lemma A.3, the transfer matrix T_a acts on a single spatial slice Sigma_a, but observables O_1, O_2 in A_ell have physical diameter up to ell (spanning ell/a time slices); the spectral expansion of time-separated correlators does not address the multi-slice temporal support of O_1, O_2.
• In Corollary 3.3, the unrenormalized field Phi_a(varphi) in Eq. (7) lacks vacuum subtraction and wave-function renormalization; as a -> 0 with g(a)^2 -> 0, (1/N_c) Re Tr U_p -> 1, so Phi_a(varphi) converges trivially to the deterministic constant integral of varphi.

Strong novelty candidates

• Lemma 1.4 (window-compatibility margin ratio c_w * 2 b_0 approx 4.5 comparing dyadic continuum scaling against the multiscale volume window along the 2-loop asymptotic-freedom trajectory).

Unresolved material objections

• Undefined cross-scale observable map in Layer 1: Definition 2.4 defines A_ell as functions of finitely many links on Lambda_{a,L}, whose link set changes with a. No embedding map identifying O in A_ell at spacing a with an observable at spacing a/2 is defined for Assumption 3.5 or Theorem 3.9. If O fixes a lattice-unit loop (e.g. a single plaquette), <O>_{a_k,L} -> 1 trivially as g_k -> 0; if O fixes a macroscopic loop of physical size <= ell by link subdivision, its unrenormalized expectation vanishes in 4D due to UV perimeter divergence.
• Invalid infinite-time OS semigroup and Laplace-transform argument on a finite-time torus (Corollary 5.6, Theorem 6.2, Lemma A.3): By Remark 1.3 and Section 2.1, omega_L is the limit state on a four-torus of fixed physical period L in all 4 directions, including Euclidean time. On a torus of time period L, reflection positivity holds only on (0, L/2), time translation tau_t is L-periodic (so tau_t leaves (0, L/2) for t >= L/2), and two-point correlators in omega_L wrap around the temporal circle rather than decaying as e^{-m_0 t} for t -> infinity (Eq. 28).
• Quantifier conflict between the finite volume window and infinite-volume limits: Lemma 1.4 verifies L/a_k <= exp(c_w/g_k^2) at fixed L as k -> infinity, but Assumption 4.4 takes L -> infinity at fixed a, and Lemma 6.1 states its bound for all L >= L_* and all a in (0, a_0] with a_0 independent of L. At any fixed a, taking L -> infinity exits the audited volume window L/a <= exp(c_w/g(a)^2 + O(1)).
• Logical disconnection between Layer 1 and Layer 2 in Theorem 7.1 (Nontriviality): Layer 2 postulates the existence and positivity of infinite-volume Creutz-ratio limits directly (Assumptions 4.4, 4.7, 4.9) without proving that they represent expectations in the Layer 1 state omega_L (or lim_{L->infinity} omega_L). Thus, sigma_phys > 0 is not proved to constrain the state constructed in Layer 1.

Strengths

• The manuscript explicitly retracts earlier unconditional descriptions, lists the imported open hypotheses, and distinguishes its bounded-observable scope from a construction of renormalized distribution-valued fields.
• Proposition 3.8 correctly derives convergence of identified scalar expectation sequences from a summable one-step bound. Remarks 3.6 and B.1 correctly explain why a naive g_k^2 proportional to 1/k majorant does not establish summability.
• The confinement section uses fixed physical loop sizes rather than fixed lattice dimensions. The a^{-2} normalization of the Creutz ratio has the correct string-tension dimension, and the conditional lower-bound passage in Theorem 4.11 is elementary but valid.
• Appendix A correctly distinguishes a Glauber-generator gap controlled by LSI from the physical transfer-matrix gap. The conversion m_phys=Delta_a/a is also dimensionally correct.

Weaknesses

• The principal assumptions already contain most advertised outcomes: 3.5 directly gives convergence, 4.4 assumes the continuum Creutz limit, 4.9 assumes its uniform positive lower bound, and A.2 assumes the uniform physical transfer gap. No new estimate substantially reduces these open tasks; in particular, no Creutz step-scaling estimate is derived.
• The established uniqueness is only independence of subsequences within a specified dyadic trajectory. It does not establish universality across anchors, renormalization prescriptions, or arbitrary cutoff sequences. The manuscript acknowledges part of this limitation, but broader construction language should remain correspondingly restricted.
• The numerical suite and exact two-dimensional toy are described but were not available here as inspectable executable contents. Their existence is not disputed. Even successful toy checks would not establish the missing four-dimensional analytic assumptions or repair the finite-period reconstruction argument.
• The imported lattice dependencies, window conventions, and citation-based comparisons require companion results not inspected in this run. The supplied manuscript identifies many dependencies transparently, but does not provide enough theorem statements and parameter ranges to audit their application end to end.

Potential errors

• Lemma 1.4's slope comparison proves, at most, eventual window inclusion for each fixed L. Inclusion for every k also requires an initial-window condition, which is absent for arbitrary L. Its exact affine formula for g_k^{-2}, and Appendix B's O(1) remainder, omit the generally O(log k) two-loop correction consistent with equation (5).
• The window comparison needs an explicit definition of the imported L_vol. If it counts sites rather than linear extent, the required logarithmic slope is 4 ln 2, not ln 2; the representative margin becomes about 1.13 rather than 4.5. The quoted arithmetic uses Nc=3, and the Nc dependence of the imported window constant is not established in the supplied text.
• Theorems 1.2(i) and 3.9 quantify over every L>0 with a fixed dyadic sequence a_k=a_0/2^k, although the lattice requires L/a_k to be integral. Arbitrary L does not satisfy this condition. Specify compatible anchors, approximating volumes, or another boundary prescription, with the corresponding limit argument.
• Remark 3.11 states an a^{-2n} norm divergence for the observables in equation (3). Each insertion there contains tr(F^{(a)})^2, with F^{(a)}=O(a^{-2}); the generic cutoff growth is instead a^{-4n}. At each fixed cutoff these lattice observables are bounded; the issue is lack of a uniform cutoff-independent bound.
• Assumption A.2 supplies a gap only for sufficiently small a, whereas Lemmas A.3 and 6.1 assert their bounds for every a in (0,a_0]. Compactness at a fixed cutoff does not establish the same uniform m_0 throughout that interval. Restrict the conclusions to the stated range or add an assumption.
• Definition 4.5 and Theorem 4.11 establish only a positive liminf, potentially +infinity. No finite upper bound, existence of an ordinary large-loop limit, or independence from the fixed aspect ratio is proved. These distinctions matter when interpreting the quantity as a physical string tension.
• Proposition 5.3 is titled SO(4) invariance but explicitly defers that conclusion. A fixed cubic four-torus is not globally SO(4)-invariant. Moreover, engineering dimension alone does not establish uniform O(a^2) suppression without control of coefficients and mixing. Retain only the symmetry actually proved.
• Appendix B's final inequality in (35) requires exp[-c' p_0(g_k)]=O(g_k^2). Logarithmic growth of p_0 alone does not establish this: if p_0(g_k) is asymptotic to A log k, one needs c'A>=1. The necessary constants or imported bound are not supplied. This does not change the correctly identified nonsummability of the g_k^2 majorant.

Strong novelty candidates

None reported.

Unresolved material objections

• Definitions 2.4 and 3.10 do not specify a common cross-cutoff *-algebra, compatible discretization maps, or a common norm. A unit ball of lattice-dependent bounded functions is not itself a C*-algebra. Consequently (8) is not fully defined for the same O at two cutoffs, and scalar limits alone do not construct the claimed state. The asserted separability also needs an argument or a suitable countably generated algebra.
• The theory is defined on a four-torus with temporal period L, but Corollary 5.6 and Theorem 6.2 use vacuum time translations for all t>=0. Finite-period correlators are thermal traces, not ground-state matrix elements; a transfer gap alone does not justify (28) as t tends to infinity. A globally induced periodic contraction semigroup would satisfy exp(-LH)=I and hence H=0. Assumption 5.5 supplies continuity, not the missing infinite-time limit or vacuum reconstruction.
• Theorem 4.11 correctly passes an assumed lower bound to a separately assumed Creutz-ratio limit. It does not identify that limit with confinement of the state constructed at fixed L: Layer 2 takes L to infinity first and uses cutoff-dependent logarithms divided by a^2. Assumptions 4.4 and 4.7 give neither positive continuum loop expectations nor control of these singular operations. No bridge to the reconstructed state's Wilson-loop area law is proved.
• Theorems 7.1 and 1.2(v) exclude generalized free fields without defining continuum Wilson loops as functionals of constructed fields. Wick factorization alone does not imply a perimeter law: for Gaussian flux, E[exp(i Phi)]=exp[-Var(Phi)/2], so area-growing variance yields an area law. This refutes the stated general reasoning, not the full 4d SU(Nc) claim by counterexample. Additional field and covariance hypotheses, and a proof under them, are required.