Chemical observation: sufficient state and predictive information

Result and scope

For the supplied three-field chemical model below, an ideal catalyst snapshot does not close its own evolution. Its first two temporal derivatives recover the complete finite-dimensional state on a specified regular domain. Every smooth exact realization of these output histories on an open preparation family therefore needs at least three real state coordinates per retained spatial site. Singular preparations need a separate analysis; some become identifiable with one more derivative, while a restricted counterexample keeps a hidden field unobservable for the entire history.

This is conditional mathematics about an independently supplied comparison model, not a derivation of chemical dynamics from TNFR. Physical admission remains not_admitted, and empirical evaluation remains not_tested. No response array, experiment, trajectory or fitted pressure is used. The atlas admission owns source selection and measurement gaps; the research program owns scheduling. The ontology permits collective observations without identifying a sensor with primitive EPI.

1. Supplied source, state and clock

Use the modified Oregonator rows in Azhand, Totz and Engel (2014), Section 3, Eq. (1), reviewed on 2026-10-16. The paper associates u,v,w with three chemical concentrations; v is the immobilized catalyst channel. Illumination enters the bromide source phi. Its model time is dimensionless. The numerical discussion also uses spiral-derived length/time units, which supply no independent laboratory-clock calibration here. The text's ordering of epsilon_u,epsilon_w and its later unassigned epsilon,epsilon' numerical notation are not silently reconciled: retain independent positive row coefficients. No numerical parameter set is imported.

Declare n sites and a fixed known real matrix L with nonnegative off-diagonal entries and L*1=0. These conditions preserve constants and support the positivity argument; they do not certify that an arbitrary matrix realizes the article's geometry or its nineteen-point stencil. Any PDE approximation and Neumann-boundary discretization need their own consistency/error justification. Both mobile fields use this declared spatial operator, while the catalyst has no diffusion. Componentwise,

\[ \dot u=\epsilon_u^{-1}[u-u^2+(q-u)w]+Lu,\qquad \dot v=u-v,\qquad \dot w=\epsilon_w^{-1}[\phi+fv-(q+u)w]+\delta Lw. \]

Assume epsilon_u,epsilon_w,q>0, f,delta>=0, nonnegative concentrations and a prescribed continuous nonnegative input phi(t). Coefficients and support are held; no event, stochastic forcing or additional reservoir law is inferred. At each zero concentration, its derivative is nonnegative when the other concentrations are nonnegative. The orthant is therefore forward invariant on the local existence interval. This does not assert global continuation for arbitrary unbounded input or reproduce a laboratory preparation. Interior positive preparations supply the open domains below.

The ideal output is the entire labeled field y=v in this model clock. It is not identified with the experimental camera image; depth integration, optical response and calibration remain unadmitted. Derivatives below are exact model derivatives, not finite differences of data.

2. Regular reconstruction and the minimum exact state

Write y_j=d^j y/dt^j and define

\[ U=y_0+y_1,\qquad D=y_1+y_2,\qquad N=\epsilon_u(D-LU)-U+U^2,\qquad a=q-U. \]

The source rows imply U=u, D=dot u and N=a*w. On the domain a_i!=0 for every site, the inverse is exactly

\[ v=y_0,\qquad u=y_0+y_1,\qquad w_i=N_i/a_i. \]

No illumination value is needed for this instantaneous inverse. It follows from the first two supplied rows; it neither infers the third row nor defines TNFR pressure from the observed response.

In state order (v,u,w), the Jacobian of Psi_2=(y_0,y_1,y_2) is block lower triangular with diagonal blocks I,I,diag((q-u)/epsilon_u). Thus

\[ \det D\Psi_2=\epsilon_u^{-n}\prod_i(q-u_i),\qquad \operatorname{rank}D\Psi_2=3n \]

on the regular domain. This is a nonlinear local observation theorem, not an equilibrium linearization or a rank count of sampled traces.

For completeness, hold the same input history and clock fixed. Suppose a target with m real state coordinates and smooth initialization J reproduces these ideal histories for every state in an open positive preparation set. Require its order-two output-jet map K_2 to be C^1 in target state. Smooth autonomous dynamics/readout with a fixed constant input suffice; arbitrary continuous inputs are not silently differentiated. The output jet then factors as Psi_2=K_2 o J. Consequently 3n<=rank(DJ)<=m wherever the source is regular. Every such open set contains regular points: the exceptional hyperplanes u_i=q have empty interior. Hence m>=3n is necessary for this exact all-state claim. Three source fields also suffice when the supplied input is retained.

In particular, n TNFR nodes with only signed scalar form and circular phase evolving, and all capacities/support held, have local dimension 2n. They cannot give that exact realization on an open source family. This does not exclude more nodes, additional justified evolving state, an admitted restricted preparation, approximate predictions or causal memory. Counting a third coordinate alone would not establish a TNFR constitutive map.

Two elementary comparisons explain the obstruction. Holding v fixed while changing u changes y_1. Holding both v,u fixed while changing w changes y_2 by (q-u)*Delta w/epsilon_u. On regular positive states, neither a snapshot nor the pair (y_0,y_1) is a closed predictive state. This reuses the general closure criterion and the earlier circuit observation method, with the chemical model's own nonlinear coefficient and singular domain.

3. Singular observations are not automatically unobservable

At a site with a_i=0, second-jet consistency requires N_i=0 and does not determine w_i. Let D=dot u as above and

\[ \dot N=\epsilon_u(y_2+y_3-LD)+(2U-1)D. \]

Differentiating N=a*w gives dot N=-D*w+a*dot w. Thus, if D_i!=0 at that singular site,

\[ w_i=-\dot N_i/D_i. \]

This transverse recovery uses the third derivative but not illumination. For a spatially uniform u=q*1, D=q(1-q)/epsilon_u; all singular sites therefore recover this way when q!=1.

There is also a spatial route. With known instantaneous illumination, the entire third-jet consistency problem is the linear system

\[ \begin{pmatrix}A\\B\end{pmatrix}w =\begin{pmatrix}N\\\dot N-A(\phi+fv)/\epsilon_w\end{pmatrix}, \quad A=\operatorname{diag}(a), \quad B=\delta AL-\operatorname{diag}(D) -\operatorname{diag}(a(q+u))/\epsilon_w. \]

Full column rank supplies a unique solution, subject to nonnegative-state and residual admission. Regular neighbors can reveal a tangential singular site through diffusion even when its own D_i=0. If the stacked matrix has a kernel, it produces nearby distinct positive states with identical jets through order three at an interior preparation. It does not prove identical full histories. At a positivity boundary, intersection with the nonnegative cone may remove this ambiguity. Unknown illumination introduces another ambiguity in the regular rows and cannot be silently eliminated.

A genuine full-history exception exists within the symbolic model family. Take q=1, spatially uniform u=1, any uniform positive v(0) and two different uniform positive w(0), with the same nonnegative uniform input. Then u stays one and dot v=1-v, independently of w; distinct bromide histories remain invisible. This is a lower-dimensional invariant preparation at a particular parameter value, not a counterexample to the open-family dimension theorem or a claim about the paper's experimental parameters.

4. Recovering state does not recover future input

On the regular domain, substituting w=N/a into the third source row yields the exact input/output identity

\[ a\dot N+ND =\frac{a^2}{\epsilon_w}\left[\phi+fv-(q+u)\frac Na\right] +\delta a^2L\left(\frac Na\right). \]

Products and divisions are componentwise; L(N/a) is the full spatial operation, not (LN)/a. Together with u=y_0+y_1, this is a third-order output description on the regular domain. It is a rewriting of the supplied chemical law with its full state information retained, not a two-field reduction or a TNFR pressure law.

At the same state, changing the prescribed input changes the third output derivative by

\[ \Delta y_3=(q-u)\Delta\phi/(\epsilon_u\epsilon_w). \]

Thus identical y_0,y_1,y_2 do not predict y_3 without input information. More generally, agreeing past inputs need not determine their future. An externally prescribed source is not rendered autonomous by reconstructing internal state. Any controller or environment generating it needs its own law.

5. Exact information versus recoverable information

Fix the source coefficients, spatial operator and clock. Suppose the true state satisfies ||u||_infinity<=M_u, ||w||_infinity<=M_w and min_i|q-u_i|>=rho>0. Let supplied approximate jets have independent supremum error allowances eta_0,eta_1,eta_2. Set

\[ \alpha=\eta_0+\eta_1,\qquad b=\eta_1+\eta_2,\qquad B_N=\epsilon_u(b+\|L\|_\infty\alpha) +(1+2M_u+\alpha)\alpha. \]

If alpha<rho, algebraic reconstruction satisfies

\[ \|\widetilde u-u\|_\infty\le\alpha,\qquad \|\widetilde w-w\|_\infty \le\frac{B_N+M_w\alpha}{\rho-\alpha}. \]

Indeed the numerator error is at most B_N, the reconstructed denominator is at least rho-alpha, and tilde N-tilde a*w=(tilde N-N)+(tilde u-u)*w. The inverse becomes poorly conditioned as the separation vanishes. Near u=q, a fixed bromide difference can produce arbitrarily small second-jet separation. This bound assumes derivative-error allowances; it does not derive them from finitely sampled noisy images, calibrate the camera or cover parameter, spatial-discretization or clock uncertainty. The transverse inverse similarly requires control of its own nonzero denominator.

The information result is therefore about distinguishable states and prediction, not bits or a universal entropy balance. Shannon quantities would additionally require a preparation distribution and an observation/noise channel. EPI, structural capacity and coherence are not probabilities, channel capacity or entropy by name. Existing TNFR memory and nonlinear hidden-state elimination provide reusable methods, not coefficients that transfer to this chemical law.

6. Execution and verification

The detached exact control module admits rational or represented-real coefficients before calculation. It evaluates source jets and reconstructs regular or pointwise transverse states; the general stacked singular solver is a theorem here, not an implemented reconstruction mode. An unsupported singular inverse is explicit abstention. Optional y_3 is consumed only by that transverse inverse and does not certify the full source/input law. The implementation is capped at 32 sites as an algebraic resource policy, not a theorem dimension restriction.

Independent controls differentiate the supplied polynomial field symbolically, check exact rank, hidden-state/input counterexamples, diffusion-assisted recovery, singular domains and scalar admission. They do not differentiate measured traces or run the chemical reactor. Existing shared exact scalar and linear-observation owners remain unchanged. No engine solver, native law, operator, grammar rule, SDK export or physical measurement map is installed by this module.

The remaining physical obligations are the complete optical observation, retained depth/spatial state, preparation, input law, laboratory clock, uncertainty and independent acquisitions. Any exact TNFR realization must also satisfy all mapped evolution rows, including its phase/capacity/support premises. The result selects no such law and does not identify a chemical organization with an elementary constituent.