# PDE C1 External Review Email Draft > Reviewer-facing sanity-check request for the C1 result. Goal: a > bounded, honest framing check from a PDE / data-assimilation analyst — > not a referee report, not an endorsement. The whole virtue of this ask > is brevity; keep it that way. Placeholders in `[brackets]`. ## Send Prep (staged 2026-06-01) **Status: send-ready once the two owner items below are filled.** The sender signature is pre-filled as **Jeffery Hughes Jr.** across all three versions — confirm, or adjust to the form you prefer (e.g. add "Stellar Aqua LLC / Sundog Research Lab"). **Which version to send:** the **Recommended Outreach Version** for a general PDE analyst; switch to the **Technical Short Version** if the reviewer is a determining-modes / data-assimilation specialist (it leads with the determining-modes framing they will recognize fastest). **Owner-fill before send — the only blockers:** 1. **`[Name]` — the reviewer.** One named person; this is a 1:1 ask. Best fit: a PDE analyst fluent in 2D NSE / determining modes / data assimilation / observer theory (full fit list under "Reviewer Fit" below). 2. **`[link]` — the packet.** Point to the four files (all verified present), as a GitHub folder link or a single PDF bundle: - `docs/proof/PDE_C1_REGIME_GENERALITY_v1.md` - `docs/proof/PDE_C1_KNN_CONVERGENCE_CHECK.md` - `docs/proof/PDE_C1_TWIN_STATE_CERTIFICATE.md` - `docs/SUNDOG_V_NAVIERSTOKES.md` Optional add if the reviewer asks "isn't this just near-determination?": `docs/proof/PDE_C1_OBJECTIVE_OVERLAP_DISCRIMINATOR.md` (the adversarial near-determination pre-empt). 3. **Tailor the specialty phrase** `[PDE / determining modes / data assimilation / numerical NSE]` in the opening line to the chosen reviewer (delete the others). 4. **`[1-2 weeks]`** — timeline; the default is fine. **Do not** soften the scope clauses (two-point Grashof / `k_f`-fixed / single-objective-family / finite-Galerkin) to make the result sound bigger: per the sender notes below, a reviewer flagging a missing caveat is a worse outcome than one confirming a conservative framing. **A negative reply is the most valuable outcome.** ## Recommended Outreach Version **Subject:** Technical framing check request: finite-Galerkin NSE signature/control result Hi [Name], I am Jeffery Hughes Jr., working on Sundog, an independent research project about coarse-grained signatures for control. We have been testing whether a compact observation can be useful for a registered control action even when it does not reconstruct the full state. I am reaching out because your work around [PDE / determining modes / data assimilation / numerical NSE] seems close to the exact sanity check we need. The specific ask is narrow: could you tell us whether the framing below is mathematically honest, too strong, or simply standard observer/data-assimilation language that should be renamed? Candidate framing: > In a 2D Kolmogorov-flow Galerkin model (`k_f = 2`), a low-Fourier > observation `Phi_K` is **state-insufficient** on the sampled support > (signature-near pairs can have well-separated high modes) yet appears > **control-sufficient** for a registered low-band-energy proxy action > (local action-mixing vanishes as signature tolerance goes to zero, up to > a small decision surface). The pattern holds at two Grashof values > (`G = 200` and `G = 300`) under a held-out-quantile objective. What we most need checked: 1. Is "state-insufficient on the sampled support" fair for the twin-state certificate, or should it be weakened? 2. Is "control-sufficient on `Phi_K`-fibers up to a measure-zero decision surface" fair for the kNN/disintegration read, or overstated? 3. Is the held-out look-ahead-max quantile a defensible way to define a regime-portable proxy action, given that the older burn-in trigger went vacuous at `G = 300`? 4. Would you call this a real finite-Galerkin separation between state-reconstruction and action sufficiency, or ordinary data-assimilation / observer behavior that should not be made load-bearing? Scope: this is early-stage and numerical. It is not a Navier-Stokes regularity claim, not an existence claim, not a theorem about the infinite-dimensional attractor, and not a request for endorsement. We are specifically trying to prevent overclaiming before paper-facing use. Packet and suggested read order: 1. `PDE_C1_REGIME_GENERALITY_v1.md` - current result, scope, and gates. 2. `PDE_C1_KNN_CONVERGENCE_CHECK.md` - control-sufficiency adjudicator. 3. `PDE_C1_TWIN_STATE_CERTIFICATE.md` - sampled-support state-insufficiency bridge. 4. `SUNDOG_V_NAVIERSTOKES.md` - project ledger and claim boundary. If you have time in the next [1-2 weeks], a paragraph or a few bullet points would be ideal. "This seems conservative," "weaken X," "cite Y," or "do not use this language because Z" are all useful outcomes. Email comments are perfect; I would also be happy to do a short call if that is easier. Thanks, Jeffery Hughes Jr. ## Technical Short Version **Subject:** Bounded sanity check — finite-Galerkin signature/control-sufficiency framing Hi [Name], Could I ask for a bounded framing check on a small finite-Galerkin Navier-Stokes note? It is **not** a regularity or existence claim and **not** a theorem about the infinite-dimensional attractor. The question is whether we are stating a numerical signature/control-sufficiency candidate result honestly. In determining-modes terms, the candidate framing is a *separation*: > In a 2D Kolmogorov-flow Galerkin model (`k_f = 2`), a low-Fourier > observation `Phi_K` is **state-insufficient** (a twin-state search > finds signature-near pairs with well-separated high modes on the > sampled SRB-like support) yet appears **sufficient for a registered > low-band-energy control action** — the local action-mixing vanishes as > the signature tolerance → 0, consistent with control-sufficiency up to > a measure-zero decision surface. The pattern holds at two Grashof > values (`G = 200` and `G = 300`) under a held-out-quantile objective > that stays discriminative across both, and the control half is also > stable to halving the signature dimension and to swapping the > objective construction. It stays finite-Galerkin, sampled-support, numerical, and proxy-based, and it is a **two-point Grashof-axis result only** — `k_f` is fixed and the objective family is single, so we are not claiming generality across forcing or objective. What I need is not endorsement but a quick read on whether that framing is earned: 1. Given the twin-state certificate is sampled-SRB / finite-Galerkin (not an exact attractor theorem), is the **"state-insufficient on the support"** language reasonable, or should it be weakened? 2. Is the kNN/disintegration reading of **"control-sufficient on `Phi_K`-fibers up to a measure-zero decision surface"** mathematically fair, or overstated? 3. Does a **held-out look-ahead-max quantile** look like a legitimate way to define a regime-portable objective (it replaced a burn-in-percentile trigger that went vacuous at `G = 300`)? 4. Would you read this as a genuine *"a sub-determining mode set can be control-sufficient without being state-reconstructive"* separation, or as ordinary data-assimilation / observer language that should not be made load-bearing? Minimal packet: - `PDE_C1_REGIME_GENERALITY_v1.md` — current result, scope, and gates. - `PDE_C1_KNN_CONVERGENCE_CHECK.md` — the control-sufficiency adjudicator. - `PDE_C1_TWIN_STATE_CERTIFICATE.md` — the sampled-support state-insufficiency bridge. - `SUNDOG_V_NAVIERSTOKES.md` — ledger and claim boundary. A one-paragraph reply is plenty: "framing seems conservative," "weaken X," "this is standard, cite Y," or "don't call this support-level / fiber-sufficient because Z." A negative answer is the most valuable outcome — the goal is to prevent overclaiming. Thanks, Jeffery Hughes Jr. ## Slightly Warmer Version **Subject:** Small 2D NSE / control-sufficiency sanity check Hi [Name], A bounded PDE sanity-check request if you have the bandwidth. We have been testing a coarse-graining reading against a finite-Galerkin 2D Kolmogorov-flow model. It is not a Clay-problem claim and not a new PDE theorem; the point is narrow and I mostly want to know if we are overstating it. The observed pattern, in determining-modes language: - a low-Fourier observation `Phi_K` does **not** reconstruct the full sampled state — a twin-state certificate finds signature-near pairs with separated high modes on the sampled SRB-like support; - the same `Phi_K` nevertheless looks **sufficient for a registered low-band-energy control action** — kNN/disintegration checks show the local action-mixing decaying to zero as the signature radius → 0; - this state-insufficient / control-sufficient pattern holds at both `G = 200` and `G = 300` (for `k_f = 2`), after we replaced a non-portable burn-in-percentile trigger with a held-out-quantile objective. The control half is also stable across a halving of the signature dimension and across the two objective constructions. The internal conclusion is deliberately scoped: > two-regime (Grashof-axis only), finite-Galerkin, sampled-support, > numerical evidence for an observation that is not state-reconstructive > but is control-sufficient for the registered proxy action. Proxy > faithfulness and PDE review are open; no infinite-dimensional NSE claim > is made. Could you sanity-check whether that is a mathematically honest framing — in particular whether the kNN-fiber/disintegration language and the sampled-support twin-state language are acceptable, or should be downgraded before any paper-facing use? A short reply is enough, and a negative one is useful. Packet: [link or attachment] Thanks, Jeffery Hughes Jr. ## Follow-Up If They Say Yes Thank you. The shortest path: 1. `PDE_C1_REGIME_GENERALITY_v1.md` — §1 (the fix), §3 (objective), §6 (portability gate), §8 (interpretation), §12 (result). 2. Skim the branch/verdict summaries in `PDE_C1_KNN_CONVERGENCE_CHECK.md` and `PDE_C1_TWIN_STATE_CERTIFICATE.md`. 3. Reply with any of: "framing seems conservative" / "weaken or rename this" / "support-level language too strong" / "fiber-sufficiency language too strong" / "this is standard data-assimilation, cite X" / "I only checked the kNN / twin-state / objective piece." Not asking for a referee report — a paragraph is ideal. ## Follow-Up If They Decline No worries at all, and thanks for considering it. If someone else is a better fit for a quick 2D NSE / determining-modes / data-assimilation framing check, a pointer would be appreciated. ## Reviewer Fit Best fit: - PDE analyst familiar with 2D NSE, determining modes, or data assimilation / observer theory; - numerical analyst with Galerkin-NSE and invariant-measure-sampling experience. Also useful: - control / data-assimilation researcher comfortable with conditional observability and finite-dimensional observer diagnostics; - turbulence / shell-model person for the empirical/objective side; - mathematical-statistics person for the kNN/disintegration language. ## Notes For The Sender (not part of the email) - The strongest honest selling points, if asked "why should I trust the numbers": the control verdict is robust to signature dimension (`d_K = 18` vs `32`), to sample budget (50k vs 200k), and to the objective construction (burn-in-percentile vs held-out-quantile both POSITIVE at `G = 200`); and the adjudicator overturned its **own** first mechanical reading on a pre-registered convergence check, so the pipeline is not tuned to a desired answer. - **Adversarial robustness (2026-05-31), pre-empting the likely first objection ("isn't this just near-determination?"):** we pre-registered and ran a probe designed to *demote* the result — testing whether control-sufficiency is merely `Φ_K` near-determining the whole state, so the objective is "controlled" only because it is predictable. It did **not** reduce to that. Across a 6-objective band/dissipation slate, `Φ_K` predicts every objective well (`R² 0.76–1.00`, including the high-band and dissipation range) yet control-sufficiency does **not** track predictability (Spearman `−0.75 / −1.0`), so the separation is not a closed-subspace artifact. One open puzzle is flagged honestly — a predictable-but-not-control-sufficient dissipation objective (`palinstrophy`). Detail: `PDE_C1_OBJECTIVE_OVERLAP_DISCRIMINATOR.md`. - Do **not** soften the scope clauses to make the result sound bigger. The two-point / `k_f`-fixed / single-objective-family / finite-Galerkin caveats are load-bearing and a reviewer flagging them as missing would be a worse outcome than a reviewer confirming a conservative framing. - If the reviewer asks "is `pi_hat` the optimal selector?": the honest answer is no — it is a registered proxy (the burn-in/held-out overshoot trigger), not a derived `J`-optimal selector; that is exactly the open proxy-faithfulness item.