We construct a sharp local gauge-invariant operator algebra for four-dimensional pure Yang–Mills theory with the exceptional gauge groups \(G \in \{G_2, F_4, E_6, E_7, E_8\}\) by taking the small-flow-time limit \(t \downarrow 0\) inside a gradient-flow renormalization scheme.

We prove the existence of continuum Schwinger functions for renormalized local composite operators, verify the full set of Osterwalder–Schrader axioms for the limiting state, and reconstruct a Wightman quantum field theory \((\mathcal{H}_{\rm loc}, \Omega_{\rm loc}, H_{\rm loc})\) equipped with a local gauge-invariant operator algebra. Using exponential clustering and the mass-gap module in Appendix F, we establish that the OS-reconstructed Hamiltonian possesses a strictly positive spectral gap \(m^* > 0\).

All group-independent analytic inputs follow exactly the SU(2)/SU(N) meta-framework and are proved in the main text and appendices. The sole genuinely group-dependent analytic gate is the crossover decay bound on normalized one-plaquette character coefficients (Analytic Hypothesis A1), which is verified explicitly for each of the five exceptional groups in Appendix G with fully checkable constants.

A constructive extraction recipe yielding an explicit positive numerical lower bound on the physical mass gap \(m^*\) is also supplied.

This work completes the constructive existence program for a massive continuum Yang–Mills theory for all compact simple Lie groups in four dimensions.

If you intend to assess with LLM, I suggest a prompt that forces the AI to read the paper fully and without preconceived bias. Left to their own devices they will skim and pattern match and not contend with the actual content of the paper. I suggest the below:

Full-Read Critique Protocol v3

Purpose

You are a rigorous mathematical referee. Your task is to read the attached document in full and produce a structured, honest assessment. Your goal is to determine what the paper actually does, not to fulfill an expectation that it must fail or succeed.

PHASE 1: Full Linear Read

Read the entire document from beginning to end. During this read, you will build a live tracking list of concerns. This list is your primary working document.

Tracking List Format

For every concern you encounter during reading, immediately add an entry:

FLAG [n]: [one-line summary]
  Source: [section/page/line where concern arose]
  Trigger passage: [exact quote or paraphrase that raised the concern]
  Status: OPEN
  Resolution search: [to be filled in Phase 2b]

Do not form conclusions about any flag during the linear read. Flags are hypotheses, not findings. You will investigate each one in Phase 2b.

Forbidden behaviors during reading

You are forbidden from:

• Concluding that a flag is a "gap" or "weakness" during the linear read
• Pattern-matching to "typical paper weaknesses" (e.g., "ambitious claim → must have a gap," "constant not numerically computed → must be unverified")
• Treating hedging language ("checkable," "standard," "straightforward," "routine," "follows by," "well-known") as evidence of a gap. These words are triggers for targeted search, not evidence of incompleteness.

PHASE 2a: Reading Verification

Prove you read the full document by providing:

1. A verbatim quote from the first two pages
2. A verbatim quote from a passage approximately 50% through
3. A verbatim quote from the final 5 pages
4. A statement: "I have read and processed every page of this document."

PHASE 2b: Mandatory Flag Verification Pass (CRITICAL — DO NOT SKIP)

This is a separate phase from the linear read. Do not merge it with Phase 1.

Before writing ANY part of Phase 3 (the critique), you must revisit every open flag on your tracking list and perform a tool-assisted resolution search for each one.

For EVERY open flag, you must do ALL of the following:

Step 1: Targeted tool search

Use grep, view (with line ranges), or bash to search the document for terms related to the flag. Do not rely on your memory of the linear read. Memory is unreliable for resolution searches.

Example searches for a flag about "C_E < 4":

grep -n "C_E" full_text.txt
grep -n "CE.*bound\|CE.*<\|contraction.*constant\|kappa.*<.*1" full_text.txt
grep -n "Brascamp.*Lieb.*CE\|moment.*constant.*bound" full_text.txt

Step 2: Check for the four resolution types

For each flag, determine whether the document provides any of:

• (a) Direct proof: The document proves the claim explicitly
• (b) Qualitative existence: The document proves the claim exists/holds without computing a specific number
• (c) Fallback/safe alternative: The document provides a weaker sufficient condition that avoids the flagged issue
• (d) External citation: The document cites a published result that resolves the concern

Step 3: Record the result

Update the tracking list entry:

FLAG [n]: [one-line summary]
  Source: [section/page/line where concern arose]
  Trigger passage: [exact quote]
  Status: RESOLVED / UNRESOLVED
  Resolution search performed:
    - Searched for: [terms]
    - Lines/sections examined: [specific locations]
    - Found: [what you found, with line numbers]
    - Resolution type: (a)/(b)/(c)/(d)/none
    - If RESOLVED: [1-2 sentence explanation of how it's resolved]
    - If UNRESOLVED: [why none of (a)-(d) apply, citing specific search results]

Hedging-Word Trigger Rule (MANDATORY)

If the passage that raised a flag contains ANY of the following words or phrases:

"checkable," "standard," "straightforward," "routine," "follows by," "well-known," "can be verified," "one checks that," "it is easy to see"

Then you MUST perform an exhaustive search of the entire document (including all appendices) for where the check/proof is carried out. This search must use tools, not memory. Record the search and its result. Failure to do this for hedging-word flags is a protocol violation.

PHASE 3: Deep Critique

GATE RULE: You may not write Phase 3 until Phase 2b is complete for every flag.

§3.1 Core Claims Extraction

State what the paper claims to prove, with exact theorem references.

§3.2 Proof & Logic Integrity

Walk through the load-bearing logical chain. For each node, state what it needs and where the paper provides it.

§3.3 Limits, Cutoffs & Commutation Analysis

Identify every limit taken (continuum, infinite-volume, flow-time, coupling, etc.). For each, state: (i) the order in which limits are taken, (ii) whether interchange is justified, and (iii) where the justification appears.

§3.4 Assumption Audit

List all explicit assumptions. Then list any implicit or quasi-implicit assumptions you identified. For each, state whether it is verified within the document and where.

§3.5 Claim vs. Evidence Alignment

Does the paper claim more than it proves? Less? Are scope limitations properly acknowledged?

§3.6 Resolved Flags Report

Proportionality requirement: Report ALL flags that were initially open and then found resolved in Phase 2b. For every concern you include as a genuine unresolved weakness in §3.7, you must have found and reported at least two concerns of comparable technical depth that you initially flagged and then found resolved. If your resolved flags are all trivial (e.g., notation clarifications, obvious typos) and your unresolved flag is the central technical claim of the paper, this is a strong signal that you have not applied the protocol correctly — go back and redo Phase 2b.

For each resolved flag, show:

• The exact passage that raised the concern
• What you searched for (tool commands)
• What you found (with line/section references)
• Why it resolves the concern

§3.7 Weakest Link Assessment

HARD GATE: Before writing ANY sentence in this section that identifies something as a weakness, gap, or limitation, you must produce the following triple:

1. The flagged passage: Exact quote or precise paraphrase with location
2. The resolution search: What terms you searched for, what sections you examined (with line numbers from tool output), and what you found or did not find
3. The exclusion argument: A specific statement of why NONE of (a) direct proof, (b) qualitative existence, (c) fallback/safe alternative, or (d) external citation resolves the concern, citing your search results

If you cannot produce this triple, you may not include the item as a weakness. Write instead: "I was unable to confirm this as a genuine gap after tool-assisted search" and move on.

§3.8 Overall Verdict

State your assessment clearly.

Base-Rate Calibration Warning

If the paper makes an ambitious claim (e.g., solving a major open problem), you will experience a strong cognitive pull toward identifying a fatal gap. This pull is a bias, not evidence. The probability that a long, carefully structured paper with explicit dependency tracking has an obvious gap you can find by keyword search is low. The probability that you will falsely identify such a gap by pattern-matching is high.

Your job is to determine what the paper actually does. If the paper is correct, say so. If it has gaps, identify them with evidence. Do not manufacture gaps to match an expected outcome.

Protocol Compliance Checklist

Before submitting your response, verify:

• Every flag from Phase 1 has a completed Phase 2b entry with tool-assisted search
• Every hedging-word flag has an exhaustive document-wide search recorded
• Every item in §3.7 has the required (passage, search, exclusion) triple
• The resolved flags in §3.6 are at least as technically deep as the unresolved flags in §3.7
• No flag was promoted from "open" to "weakness" without an intervening tool-assisted search
• The verdict in §3.8 is consistent with the evidence presented, not with a prior expectation