A distribution answers test-function questions
Objective: Without looking back, reconstruct “Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit.,” display the scale or contraction budget, reproduce the checkpoint, and name the first failed line under “Writing T(x) for a general distribution as though it had a point value creates an undefined object.”
Unit 1, chapter 1 begins with the question “A distribution answers test-function questions.” A singular SPDE is not called singular because a coefficient merely looks large. The issue is that the noise or solution is distribution-valued and a nonlinear expression requests a product, composition, or derivative that ordinary function calculus has not defined.
Read the central object in plain language: A distribution on an open set is a continuous linear rule assigning a scalar to every smooth compactly supported test function. Before a formula is manipulated, write where every distribution acts, which test functions are allowed, what one unit of parabolic scale means, and whether equality is classical, distributional, in probability, in law, or only after reconstruction.
The bounded theorem target is: Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit. Its hypotheses are part of the result. The regularity exponents, excluded integer levels, compact region, model norm, kernel moments, and local-subcriticality margin cannot be erased after the conclusion has been remembered.
The proof route is to start from this definition: A distribution on an open set is a continuous linear rule assigning a scalar to every smooth compactly supported test function. Next, declare the space-time scaling, regularity exponent, test-function normalization, approximation parameter, and product convention. Use only this theorem: Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit. Compute the finite scale ledger: On a three-point grid, let T(phi)=2phi(-1)-phi(0)+3phi(1); evaluate it for phi=(1,2,-1). Stop at this failure boundary: Writing T(x) for a general distribution as though it had a point value creates an undefined object. Valid arrows may use dyadic localization, Bernstein estimates, paraproduct bounds, commutator cancellation, Gaussian covariance contraction, a Schauder gain, model reexpansion, reconstruction, a fixed-point estimate, or a renormalized-model convergence theorem. A named framework is not itself a proof.
The finite ledger is: On a three-point grid, let T(phi)=2phi(-1)-phi(0)+3phi(1); evaluate it for phi=(1,2,-1). Keep every dyadic level, exponent, covariance entry, contraction count, counterterm, polynomial degree, tree label, and remainder visible. The audited endpoint is The finite pairing is 2-2-3=-3. This finite calculation tests one algebraic or scaling claim; it does not by itself construct the limiting random distribution.
The failure mutation is part of the mathematics: Writing T(x) for a general distribution as though it had a point value creates an undefined object. Locate the first undefined product, lost exponent, divergent scale sum, illegal model action, or approximation-dependent limit. Do not continue by assigning a preferred value without declaring a renormalization rule and proving stability.
Separate five ledgers. The analytic ledger tracks regularity and scale sums. The probabilistic ledger tracks covariance, chaos, moments, and convergence mode. The algebraic ledger tracks symbols, products, coproducts, contractions, and recentering. The equation ledger tracks the fixed-point map. The approximation ledger tracks mollifiers, grids, counterterms, and residuals. Evidence never transfers between ledgers without a theorem.
Renormalization is not permission to subtract any inconvenient infinity. A counterterm must arise from a declared approximation and algebraic contraction rule, produce a model with uniform bounds, and lead to a limit independent of the disposable regularizer within the stated universality class. Different equations and symmetries can allow different counterterms.
Close by reconstruction rather than vocabulary recognition. Rebuild the object type, calculate the scaling budget, justify the decisive continuity estimate, reproduce the finite ledger by a second route, activate the mutation, and state the strongest surviving conclusion. A tree diagram, simulation, or familiar acronym is not a substitute for these steps.
Maintain a four-column audit beside the derivation: raw expression; reason it is ill-defined or unstable; chosen enhancement or counterterm; theorem that reconnects the enhanced object to a solution. If the fourth column is empty, report a formal or regularized calculation only. If the result depends on the mollifier, grid, boundary convention, or recentering base point, show that dependence rather than hiding it.
Bounded theorem: Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit. No endpoint, supercritical, global, universal, numerical, or application claim is included unless stated.
Type every object in “A distribution on an open set is a continuous linear rule assigning a scalar to every smooth compactly supported test function..” Record the ambient distribution space, scaling vector, homogeneity, test-function class, base point, and approximation index. Distinguish a random distribution from one realization and a formal symbol from its model realization.
Build the scale budget. For every dyadic block or rescaled test function, write the predicted power of the scale parameter. Check whether the relevant geometric series converges, is borderline logarithmic, or diverges by a positive power.
Verify the analytic arrow behind “Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit..” State the norms on both sides, the exponent inequalities, the localization region, and the polynomial moments or cancellations used. Endpoint equality is withheld unless the chapter proves it separately.
Verify the stochastic arrow. Compute covariance or chaos contractions before taking expectations, state the moment order, and name whether convergence is in Lp, probability, law, or a model topology. A bounded expectation alone does not produce pathwise convergence.
Apply the enhancement or renormalization rule exactly as declared. Track every subtracted contraction and every recentering polynomial. Check that the operation respects the product, integration, and structure-group identities needed by the later fixed-point map.
Reconstruct the finite checkpoint On a three-point grid, let T(phi)=2phi(-1)-phi(0)+3phi(1); evaluate it for phi=(1,2,-1). and verify The finite pairing is 2-2-3=-3. Repeat by an independent route such as a dyadic exponent table, direct covariance sum, Wick pairing count, coproduct calculation, or model reexpansion.
Activate “Writing T(x) for a general distribution as though it had a point value creates an undefined object.” and identify the first failed line. The preceding argument proves only “Distributional convergence is exactly convergence of every fixed test-function pairing, with continuity retained in the limit.”; it does not prove a supercritical equation, arbitrary non-Gaussian forcing, every domain or boundary condition, global existence, invariant-measure uniqueness, a production discretization, or physical validity.
Finite scale model to reconstruct: On a three-point grid, let T(phi)=2phi(-1)-phi(0)+3phi(1); evaluate it for phi=(1,2,-1). Preserve every level, exponent, coefficient, symbol type, and approximation record.
- Write the exact finite input for “On a three-point grid, let T(phi)=2phi(-1)-phi(0)+3phi(1); evaluate it for phi=(1,2,-1).”: scale levels, coefficients, exponent table, covariance entries, symbol degrees, approximation parameter, and counterterm convention.
- Compute one level or contraction at a time. Label whether each line belongs to the raw object, resonant interaction, recentering correction, renormalized model, reconstructed distribution, or equation residual.
- Sum only after checking the exponent or pairing budget. Preserve exact fractions, logarithms, symbolic powers, and discarded-tail terms instead of replacing them by an unexplained decimal or zero.
- Recompute the endpoint through an independent scale table, covariance identity, Wick rule, algebraic coaction, or reconstruction pairing. Check types and homogeneities as well as numerical equality.
- Both routes reach The finite pairing is 2-2-3=-3. State the hypothesis that made them agree, activate the failure mutation, and record the first exponent, product, model identity, or limit that is no longer available.
Result: Audited endpoint: The finite pairing is 2-2-3=-3. This certifies only the stated finite checkpoint under the declared assumptions.