# PDE Determining Modes Through Postulate 1 > Proof-track sidecar for > [`COARSE_GRAINING_PROOF_ROADMAP.md`](../COARSE_GRAINING_PROOF_ROADMAP.md). > Status: drafted, unreviewed, 2026-05-28. Commissioned from > [`SUNDOG_V_NAVIERSTOKES.md`](../SUNDOG_V_NAVIERSTOKES.md) Candidate 1. > This note reads the determining-modes / synchronization literature through > Postulate 1's control-sufficiency predicate. It is not a Navier-Stokes > existence or smoothness claim, not an improved determining-mode bound, and not > public-facing theorem prose. ## Entry And Gate This sidecar enters from the Navier-Stokes ledger's Front A: > Re-read determining modes through Postulate 1 as objective-specific > Blackwell sufficiency for control, rather than sufficiency for state > reconstruction. The pre-registered negative, quoted from the ledger, is: > If every regime where state sufficiency is open also has control sufficiency > strictly equivalent to state sufficiency under the standard data-assimilation > gauge, the Postulate-1 reading is vacuous on NSE -- record, do not rescue. Draft verdict: **negative not triggered as a logical reading, not yet closed as a PDE result.** The reading produces a strict predicate difference: state-reconstruction sufficiency implies control sufficiency, but control sufficiency only requires common optimal actions on signature fibers. That is the same operator as Phase 2, now aimed at a PDE observation family. Promotion is blocked until a PDE reviewer confirms the reading is not merely ordinary determining-modes prose in different clothes. ## Claim Boundary This note does **not** claim: - a solution to the Navier-Stokes Millennium problem; - a new determining-modes theorem; - a smaller mode count for standard NSE state reconstruction; - that determining modes are a Sundog invention; - that control sufficiency has been proved for a canonical NSE control problem. It claims only a translation and a falsifiable audit surface: 1. existing determining-mode and synchronization results are naturally state-reconstruction sufficiency statements; 2. Postulate 1 asks a smaller, objective-specific question: whether the optimal control selector is measurable with respect to the signature sigma-algebra; 3. the two notions coincide for full-state objectives and can separate for objectives whose optimal actions are constant on state-distinct fibers. All literature citations and the 2026-05-28 prior-art boundary for this reading live in [`NAVIERSTOKES_LITPASS_MEMO.md`](../NAVIERSTOKES_LITPASS_MEMO.md). ## Model Work only in a regime where the NSE trajectory object is already admitted by standard theory or by a finite approximation: - 2D incompressible NSE on a bounded or periodic domain after entry to the global attractor; - a Galerkin-truncated NSE model with a fixed cutoff; - a regular 3D interval or data-assimilation setup where the cited literature already supplies the relevant well-posedness / synchronization statement. Let the decision domain `X` be the information available at a control decision. Depending on the regime, `X` may be the current velocity field on the admitted attractor, a finite Galerkin state, or an observation history. The observation/signature family is: ```text Phi_m : X -> Sigma_m ``` where `Phi_m` may be low Fourier modes, determining nodes, local observables, or the coarse observation history used by a nudging / synchronization filter. Classical determining-mode language says, schematically: ```text Phi_m(u) = Phi_m(v) over the admitted observation window => u and v agree, or their trajectories synchronize asymptotically. ``` Postulate 1 asks instead whether, for a stated objective `J`, an admitted optimal selector `pi*_J` factors through `Phi_m`: ```text pi*_J(x) = g(Phi_m(x)) mu-a.e. ``` for some measurable signature policy `g`. ## Local Symbols | Symbol | Definition | | --- | --- | | `u, v` | NSE states or trajectories in the admitted regime. | | `H` | Velocity phase space used by the underlying PDE result. | | `A_NSE` | Admitted invariant/absorbing set, Galerkin state space, or regularity class for the note. | | `P_m` | Projection onto the first `m` modes, when modes are the observation. | | `I_h` | Generic coarse interpolant / node / local-observable map. | | `Phi_m` | Signature map, either `P_m`, `I_h`, or an observation-history statistic. | | `F_m` | Sigma-algebra generated by `Phi_m`; ASCII rendering of the Postulate-1 signature sigma-algebra. | | `R_m` | Reconstruction or synchronization map/filter, when the literature supplies one. | | `A_ctrl` | Admitted control action space for the objective under study. | | `J` | Pre-registered control objective or readout objective. | | `pi*_J` | A `J`-optimal selector under the admitted information regime. | | `A*_J(x)` | Optimal-action correspondence for objective `J` at decision state `x`. | | `C_sigma` | A `Phi_m`-fiber: `{x in X : Phi_m(x)=sigma}`. | ## Reading 1 -- State Reconstruction Implies Control Sufficiency If a determining observation reconstructs state, then it is automatically control-sufficient for any state-measurable objective whose optimal selector is well-defined. ### Proposition Assume there is a measurable reconstruction map `R_m` such that ```text x = R_m(Phi_m(x)) mu-a.e. ``` on the admitted decision domain. If the full-information optimal selector `pi*_J : X -> A_ctrl` is measurable, then `pi*_J` is `F_m`-measurable. ### Proof Define: ```text g(sigma) = pi*_J(R_m(sigma)). ``` Then, for `mu`-almost every `x`, ```text g(Phi_m(x)) = pi*_J(R_m(Phi_m(x))) = pi*_J(x). ``` So `pi*_J` factors through `Phi_m`. This is exactly Postulate 1's control-sufficiency predicate. For synchronization filters, replace exact reconstruction by the limiting statement: if the observer state `R_m(Phi_m(h_t))` converges to the true state along the admitted trajectory class, and if the policy/value map is continuous enough for the objective under study, the induced signature policy converges to the full-state policy. The convergence regularity is objective-specific and must not be hidden inside the phrase "determining modes." ### Consequence This implication is useful but not novel by itself. A PDE analyst could already say: if low modes reconstruct the state, then any state-feedback decision can be made from the low modes. That is the vacuity trap. The Sundog edge, if any, must come from Reading 2. ## Reading 2 -- Control Sufficiency Is Weaker Than State Reconstruction Postulate 1 does not require reconstruction. It requires only that the admitted optimal action not change inside a signature fiber. ### Fiber Criterion For a fixed objective `J`, `Phi_m` is control-sufficient on the admitted support iff every positive-mass fiber admits a common optimal action: ```text for every sigma in Phi_m(supp mu): intersection_{x in C_sigma} A*_J(x) is nonempty. ``` Equivalently, the optimal-action correspondence admits a measurable selector that is constant on `Phi_m`-fibers. Under the Phase 0 convention where `pi*_J` is the pre-registered selector, this becomes: ```text Phi_m(x0) = Phi_m(x1) => pi*_J(x0) = pi*_J(x1) ``` on the admitted support, up to `mu`-null sets. ### What This Adds Determining modes ask whether `Phi_m` pins the state. Postulate 1 asks whether `Phi_m` pins the action required by `J`. Thus there are three logically distinct regimes: 1. **State sufficient, control sufficient.** `Phi_m` reconstructs state. Control sufficiency follows by Reading 1. 2. **State insufficient, control sufficient.** `Phi_m` collapses distinct NSE states, but the collapsed states share at least one `J`-optimal action. This is the non-vacuous Sundog target. 3. **State insufficient, control insufficient.** A `Phi_m`-fiber contains states requiring incompatible optimal actions. This is the named-negative boundary. This is the Phase 2 finite-MDP theorem with the state space replaced by an admitted PDE/Galerkin decision domain. The replacement is only legitimate after the regime pins `X`, `Phi_m`, `J`, `mu`, `pi*_J`, and the positive-mass fibers. ## Candidate Strictness Witness The clean first witness is not full 3D NSE. It is a registered finite or 2D controlled surrogate where the objective is lower-dimensional than the state. ### Low-Band Safety Trigger Let `P_K u` be a low-band projection that is lower-dimensional than a state-reconstructive reference observation `P_{m_det}` and has an independent non-injectivity witness on the admitted state domain. The witness cannot be the inequality `K < m_det` alone: an upper determining-mode bound proves sufficiency for the larger observation, not insufficiency for the smaller one. Define a binary or small-action objective: ```text J = keep low-band energy / enstrophy proxy inside a registered safe envelope ``` with actions such as: ```text {no_op, damp_low_band, trigger_observer_refinement} ``` A strict Postulate-1 separation would be witnessed if there are positive-mass states `u != v` such that: ```text P_K u = P_K v ``` and unresolved high modes differ enough to prevent state reconstruction, but: ```text A*_J(u) intersection A*_J(v) != empty. ``` Then `P_K` is control-sufficient for `J` on that fiber even though it is not state-reconstructive. A failed witness is equally useful. If some positive-mass `P_K` fiber straddles the safety-trigger boundary so that one state needs `damp_low_band` and another needs `no_op`, then the low-band signature is not control-sufficient. The note must record that as the same boundary as Phase 3, not rescue it by adding hidden state after the read. ### Why This Is Not Just Review Language The non-vacuous claim is not "determining modes are enough for control." That follows whenever they reconstruct state. The non-vacuous claim is: > For some registered NSE-class objective, the smallest control-sufficient > signature may be smaller than the smallest state-reconstructive signature, > because the objective is constant on fibers that state reconstruction would > still separate. That is an objective-specific Blackwell statement. It can be false. It has a visible failure mode. It is therefore a candidate proof-track artifact rather than a rhetorical bridge. ## Equivalence / Named-Negative Regimes The Postulate-1 reading should be considered vacuous for an objective if any of the following holds: - the objective is full-state tracking or exact trajectory reconstruction; - the action set directly actuates or penalizes all unresolved state components; - the value function or optimal selector separates every state-distinct `Phi_m`-fiber on positive measure; - the cited determining-modes machinery already supplies state reconstruction at a mode count no larger than the candidate control signature; - the only "strictness" appears after changing the objective post hoc. In these cases, control sufficiency collapses to state sufficiency for practical purposes. The correct filing is: ```text PDE-C1-NEG: no Postulate-1 edge over determining modes for this objective. ``` Candidate 2 should not inherit a stronger framing from a C1 negative. If C1 lands negative across the canonical objectives, the shell-model empirical leg can still be run as a turbulence-signature detector, but not as evidence that determining modes expose a Postulate-1 gap. ## Pre-Registered Cell Set For The Next Pass Before any numerical or public-facing NSE read, pin a cell set with: 1. **Regime.** 2D NSE attractor, finite Galerkin NSE, or another admitted regular regime. 2. **Observation.** `P_K`, `I_h`, or synchronization-history statistic. 3. **Objective.** A stated control/readout objective, preferably a low-band safety trigger or observer-refinement decision. 4. **State-reconstruction comparator.** A determining-mode / synchronization threshold from the cited literature or a fixed Galerkin comparator. 5. **Fiber classifier.** The procedure for deciding whether a positive-mass signature fiber has a common optimal action. 6. **Negative branch.** The exact incompatible-action condition that files `PDE-C1-NEG`. The first pass should be desk-auditable: one page of definitions plus a small fiber table. It does not need to solve a PDE or train a detector. **v0 instance filed.** A Kolmogorov-flow instance of this cell set is at [`PDE_C1_CELLSET_KOLMOGOROV.md`](PDE_C1_CELLSET_KOLMOGOROV.md), drafted 2026-05-28. Status: desk-auditable, unreviewed, unrun. Alternative regimes remain admissible if a reviewer prefers them. ## External Review Path Promotion requires a practicing PDE analyst familiar with determining modes, data assimilation, or synchronization for NSE to check three questions: 1. Is Reading 1 faithful to determining-modes / synchronization results? 2. Is Reading 2 a real distinction in NSE control language, or merely standard model reduction phrased in Sundog notation? 3. Is the proposed strictness witness a legitimate objective-specific control question, with no hidden state reconstruction smuggled into `Phi_m`? Until that review is complete, the artifact status is **draft reading note**. ## Cross-File Consequences If this note survives review: - update [`COARSE_GRAINING_PROOF_ROADMAP.md`](../COARSE_GRAINING_PROOF_ROADMAP.md) Phase 2 with a PDE corollary: state reconstruction implies Postulate-1 sufficiency, while objective-specific control sufficiency is the weaker fiber-selector condition; - update [`SUNDOG_V_NAVIERSTOKES.md`](../SUNDOG_V_NAVIERSTOKES.md) Promotions with Candidate 1 as a reviewed Front-A reading; - then decide whether Candidate 2 is still framed as a PDE-substrate empirical sibling or only as a shell-model regime-detector workbench. If this note fails review: - file `PDE-C1-NEG` in the Navier-Stokes ledger; - do not use determining modes as evidence for a Sundog-specific PDE coupling; - keep any shell-model signature experiment explicitly empirical and detached from the determining-modes claim. ## Exit Status Candidate 1 commission: **drafted, unreviewed.** Pre-registered negative: **not triggered by the internal logical reading, not adjudicated by external PDE review.** The reading cleanly separates state-reconstruction sufficiency from objective-specific control sufficiency, but the first strictness witness remains a proposed cell set, not a theorem or receipt. Promotion status: **blocked.** The ledger's criteria (Front-A vacuity rebuttal, runnable Phase-0 deliverable, named external sanity-check path, fixed failure boundary) are only partially satisfied here. This note supplies the Front-A reading, the failure boundary, and the review path; it does not yet supply a reviewed promotion or a runnable cell-set artifact.