# 211. Uniqueness of the Maxwell Closure in the Stated Class This appendix proves the uniqueness claim that chapter 7 can only suggest in words. The claim is not that Maxwell closure is unique among all imaginable local relations whatsoever. The claim is narrower and precise: > Within the class of real, linear, local, homogeneous, isotropic, first-order, > purely differential, divergence-preserving two-field closures, every neutral > isotropic transporting closure is equivalent, by a real linear recombination > of the two fields, to the Maxwell pair > $$ > \partial_t\mathbf{F}_{+}=k\,\nabla\times\mathbf{F}_{-}, > \qquad > \partial_t\mathbf{F}_{-}=-k\,\nabla\times\mathbf{F}_{+}. > $$ This is the exact sense in which the closure of chapter 7 is unique. ## 211.1 The Closure Class Let $$ \mathbf{F}_1(\mathbf{r},t), \qquad \mathbf{F}_2(\mathbf{r},t) $$ be two real vector fields on $\mathbb{R}^3$, each constrained by $$ \nabla\cdot\mathbf{F}_1=0, \qquad \nabla\cdot\mathbf{F}_2=0. $$ We consider closures of the form $$ \partial_t\mathbf{F}_a=\mathcal{L}_{ab}\mathbf{F}_b, \qquad a,b\in\{1,2\}, $$ with the following assumptions. 1. Real-linear: $\mathcal{L}$ is linear over the real numbers. 2. Local and homogeneous: $\mathcal{L}$ has constant coefficients and depends only on the value of the fields and their derivatives at the same point. 3. First-order in space: $\mathcal{L}$ contains at most one spatial derivative. 4. Purely differential: there is no algebraic zero-order mixing term. This restriction is intentional. Chapter 6 separated algebraic updates from spatial reorganization; the present theorem concerns the transporting closure class. 5. Isotropic: the closure is equivariant under proper spatial rotations. 6. Divergence-preserving: if $\nabla\cdot\mathbf{F}_a=0$ initially, then $\nabla\cdot(\partial_t\mathbf{F}_a)=0$. The task is to classify all such closures and determine which of them yield neutral isotropic transport. ## 211.2 Classification of All Closures in the Class Because the closure is linear, local, homogeneous, first-order, and purely differential, there exist constants $B_{abijk}$ such that $$ (\partial_t\mathbf{F}_a)_i = B_{abijk}\,\partial_j(\mathbf{F}_b)_k. $$ Isotropy under proper rotations means the tensor $B_{abijk}$ must satisfy $$ R_{i\ell}\,B_{ab\ell mn}\,R_{jm}\,R_{kn} = B_{abijk} $$ for every rotation matrix $R\in SO(3)$. For fixed field indices $a,b$, this is the classification problem for an isotropic rank-three tensor on Euclidean space. Up to a scalar multiple, the only such tensor is the Levi-Civita symbol $\varepsilon_{ijk}$. Therefore $$ B_{abijk}=M_{ab}\,\varepsilon_{ijk} $$ for some real $2\times 2$ matrix $M=(M_{ab})$. Hence every closure in the stated class has the form $$ \partial_t\mathbf{F}_a = M_{ab}\,\nabla\times\mathbf{F}_b. $$ Equivalently, if we write $$ \mathbb{F} := \begin{pmatrix} \mathbf{F}_1\\ \mathbf{F}_2 \end{pmatrix}, $$ then $$ \partial_t\mathbb{F} = M\,(\nabla\times)\mathbb{F}, $$ where $(\nabla\times)$ acts componentwise on the two fields. This classification is exact. ## 211.3 Divergence Preservation For any real matrix $M$, $$ \nabla\cdot\bigl(M_{ab}\,\nabla\times\mathbf{F}_b\bigr) = M_{ab}\,\nabla\cdot(\nabla\times\mathbf{F}_b)=0. $$ So every closure in the classified form preserves the divergence-free condition identically. Thus the classification problem is reduced to the field-space matrix $M$. ## 211.4 Second-Order Consequence Differentiate once more in time: $$ \partial_t^2\mathbf{F}_a = M_{ab}\,\nabla\times(\partial_t\mathbf{F}_b). $$ Substitute the closure again: $$ \partial_t^2\mathbf{F}_a = M_{ab}\,\nabla\times\bigl(M_{bc}\,\nabla\times\mathbf{F}_c\bigr) = (M^2)_{ac}\,\nabla\times(\nabla\times\mathbf{F}_c). $$ Because each field is source-free, $$ \nabla\times(\nabla\times\mathbf{F}_c) = \nabla(\nabla\cdot\mathbf{F}_c)-\nabla^2\mathbf{F}_c = -\nabla^2\mathbf{F}_c. $$ Hence $$ \partial_t^2\mathbf{F}_a = -(M^2)_{ac}\,\nabla^2\mathbf{F}_c. $$ In block form, $$ \partial_t^2\mathbb{F} = -M^2\,\nabla^2\mathbb{F}. $$ So the entire second-order transport content of the closure is encoded by the matrix $-M^2$. ## 211.5 Criterion for Neutral Isotropic Transport In the present appendix, neutral isotropic transport means the following: there exists a real field-space change of variables $$ \mathbb{G}=P^{-1}\mathbb{F}, \qquad P\in GL(2,\mathbb{R}), $$ and a positive constant $k$ such that each transformed field satisfies the same wave equation $$ \partial_t^2\mathbf{G}_1-k^2\nabla^2\mathbf{G}_1=0, \qquad \partial_t^2\mathbf{G}_2-k^2\nabla^2\mathbf{G}_2=0. $$ Because $P$ is constant, it commutes with $\partial_t$ and $\nabla^2$. So the second-order equation becomes $$ \partial_t^2\mathbb{G} = -P^{-1}M^2P\,\nabla^2\mathbb{G}. $$ Therefore neutral isotropic transport holds if and only if $$ -P^{-1}M^2P=k^2I. $$ Since the identity is invariant under similarity, this is equivalent to $$ M^2=-k^2I. $$ So the transport criterion is exact: > A closure in the stated class yields neutral isotropic transport with one > common speed $k$ if and only if its field-space matrix satisfies > $M^2=-k^2I$. ## 211.6 Reduction to the Canonical Maxwell Pair We now classify all real $2\times 2$ matrices satisfying $$ M^2=-k^2I, \qquad k>0. $$ Let $\mathbf{e}\in\mathbb{R}^2$ be any nonzero vector. The vectors $\mathbf{e}$ and $M\mathbf{e}$ are linearly independent. For if they were dependent, we would have $$ M\mathbf{e}=\lambda\mathbf{e} $$ for some real $\lambda$, and then $$ M^2\mathbf{e}=\lambda^2\mathbf{e}, $$ which would imply $$ \lambda^2=-k^2, $$ impossible over the real numbers. So the vectors $$ \mathbf{e}_1:=\mathbf{e}, \qquad \mathbf{e}_2:=-\frac{1}{k}M\mathbf{e} $$ form a basis of $\mathbb{R}^2$. In this basis, $$ M\mathbf{e}_1 = k\,\mathbf{e}_2, $$ and $$ M\mathbf{e}_2 = -\frac{1}{k}M^2\mathbf{e} = -\frac{1}{k}(-k^2)\mathbf{e} = -k\,\mathbf{e}_1. $$ Therefore the matrix of $M$ in the basis $(\mathbf{e}_1,\mathbf{e}_2)$ is $$ \begin{pmatrix} 0 & -k\\ k & 0 \end{pmatrix}. $$ After exchanging the two basis vectors if desired, this becomes the canonical matrix $$ J_k := \begin{pmatrix} 0 & k\\ -k & 0 \end{pmatrix}. $$ So there exists a real invertible matrix $P$ such that $$ P^{-1}MP=J_k. $$ Now set $$ \begin{pmatrix} \mathbf{F}_{+}\\ \mathbf{F}_{-} \end{pmatrix} := P^{-1} \begin{pmatrix} \mathbf{F}_1\\ \mathbf{F}_2 \end{pmatrix}. $$ Then the closure becomes $$ \partial_t \begin{pmatrix} \mathbf{F}_{+}\\ \mathbf{F}_{-} \end{pmatrix} = \begin{pmatrix} 0 & k\\ -k & 0 \end{pmatrix} (\nabla\times) \begin{pmatrix} \mathbf{F}_{+}\\ \mathbf{F}_{-} \end{pmatrix}, $$ that is, $$ \partial_t\mathbf{F}_{+}=k\,\nabla\times\mathbf{F}_{-}, \qquad \partial_t\mathbf{F}_{-}=-k\,\nabla\times\mathbf{F}_{+}. $$ This is exactly the canonical Maxwell closure of chapter 7. We have therefore proved: > Every neutral isotropic transporting closure in the stated class is > equivalent, by a real linear recombination of the two fields, to the Maxwell > pair. That is the promised uniqueness theorem. ## 211.7 Corollary: No One-Field Real Closure in This Class Transports For a one-field closure in the same class, the matrix $M$ is just a real scalar $m$, so the condition for neutral isotropic transport would be $$ m^2=-k^2, $$ which has no real solution. So within the same class: - one real field cannot furnish neutral isotropic transport, - two real fields are necessary, - and once two are admitted, the transporting closure is unique up to real field recombination. This corollary places appendix 203 and the present appendix in one chain: - appendix 203 proved the minimality claim constructively, - appendix 211 proves the uniqueness claim in the stated class. ## 211.8 Scope of the Theorem The theorem is exact, but its scope is the class stated at the beginning. It does not address: - nonlinear closures, - anisotropic media, - higher-order spatial operators, - closures with explicit zero-order algebraic mixing, - closures with more than two independent fields, - nonlocal closures. So the theorem should be read correctly: - not as a proof that Maxwell is unique among all conceivable mathematical structures, - but as a proof that Maxwell is unique inside the exact closure class singled out by chapters 6 and 7. That is already a strong result. ## 211.9 Summary Within the real, linear, local, homogeneous, isotropic, first-order, purely differential, divergence-preserving two-field class: 1. every closure has the form $$ \partial_t\mathbb{F}=M(\nabla\times)\mathbb{F} $$ for a real $2\times 2$ matrix $M$, 2. its second-order consequence is $$ \partial_t^2\mathbb{F}=-M^2\nabla^2\mathbb{F}, $$ 3. neutral isotropic transport with speed $k$ occurs if and only if $$ M^2=-k^2I, $$ 4. every such matrix is real-similar to $$ \begin{pmatrix} 0 & k\\ -k & 0 \end{pmatrix}, $$ 5. therefore the closure is equivalent to $$ \partial_t\mathbf{F}_{+}=k\,\nabla\times\mathbf{F}_{-}, \qquad \partial_t\mathbf{F}_{-}=-k\,\nabla\times\mathbf{F}_{+}. $$ Thus Maxwell closure is unique in the stated class.