--- title: Ternary Cartesian Decomposition (TCAD) url: https://www.emergentmind.com/topics/ternary-cartesian-decomposition-tcad type: topic --- # Ternary Cartesian Decomposition (TCAD) Searching arXiv for the supplied topics and papers to ground the article. Ternary Cartesian Decomposition (TCAD) is not a single uniformly standardized construction across the arXiv literature. The term is used for at least three distinct technical frameworks that share a ternary structural motif: a Cartesian decomposition of a permutation domain into three components in permutation group theory, a fan-in-\(2\) decomposition of a ternary function \(T(x,y,z)\) through binary intermediates in decomposition complexity, and a three-part decomposition of the velocity gradient tensor \(A=\nabla u\) in Cartesian coordinates into rigid rotation, pure shear, and stretching/compression in fluid kinematics. In ternary quantum logic, by contrast, the relevant Lie-theoretic notion is explicitly Cartan decomposition rather than Cartesian decomposition [2211.00418] [1012.1283] [2106.10564] [1105.5485]. ## 1. Terminology and domain-specific meanings The phrase “Ternary Cartesian Decomposition” appears in technically different settings, and the precise meaning depends on the object being decomposed. In the group-theoretic setting, the object is a finite set \(\Omega\) carrying a ternary product structure \(\Omega \cong \Delta_1 \times \Delta_2 \times \Delta_3\). In Shen’s decomposition-complexity setting, the object is a ternary function \(T(x,y,z)\) represented via binary components \(a(x,y)\) and \(b(y,z)\). In fluid kinematics, the object is the velocity gradient tensor \(A\), decomposed in the original \(xyz\) coordinates as \(A=L+S_{\text{pure}}+D\). In ternary quantum-circuit synthesis, the cited source states that the correct term is Cartan decomposition, not Cartesian decomposition [2211.00418] [1012.1283] [2106.10564] [1105.5485]. | Domain | Object | Canonical form | |---|---|---| | Permutation groups | Finite set \(\Omega\) with preserved ternary product structure | \(\Omega \cong \Delta_1 \times \Delta_2 \times \Delta_3\), \(G \le H^3 \rtimes K\) | | Decomposition complexity | Ternary function \(T(x,y,z)\) | \(T(x,y,z)=t(a(x,y),b(y,z))\) | | Fluid kinematics | Velocity gradient tensor \(A=\nabla u\) | \(A=L+S_{\text{pure}}+D\) | | Ternary quantum logic | One-qutrit and two-qutrit gate synthesis | Cartan decomposition \(U=K_1AK_2\), not Cartesian | A plausible implication is that TCAD should be treated as a polysemous label rather than the name of a single canonical method. The commonality lies in ternary structure; the mathematical content differs sharply from one field to another. ## 2. TCAD in permutation group theory In the permutation-group setting, a Cartesian decomposition of a finite set \(\Omega\) is a tuple of partitions \(E=\{T_1,T_2,\dots,T_k\}\) such that for any choice of blocks \(Y_i \in T_i\), the intersection \(Y_1 \cap Y_2 \cap \dots \cap Y_k\) is a singleton. This induces a bijection between \(\Omega\) and the Cartesian product \(T_1 \times T_2 \times \dots \times T_k\). The ternary case is \(k=3\), with \[ \Omega \cong \Delta_1 \times \Delta_2 \times \Delta_3, \qquad \Delta_i := T_i. \] A permutation group \(G \le \mathrm{Sym}(\Omega)\) preserves the TCAD if it permutes the partitions \(T_i\) among themselves and respects the singleton-intersection property. The induced action on the set of components yields a homomorphism \[ \varphi: G \longrightarrow S_3, \] whose kernel acts componentwise on \(\Delta_1,\Delta_2,\Delta_3\) [2211.00418]. The associated wreath-product description is the structural core of the framework. For a typical component \(\Delta\), one chooses \(H \le \mathrm{Sym}(\Delta)\) realizing the induced component action and lets \(K \le S_3\) be the image of \(G\) on the set of components. The wreath product in product action is \[ W=H \wr K = H^3 \rtimes K, \qquad |W|=|H|^3|K|. \] Here \(H^3\) is the base group acting independently on the three coordinates, while \(K\) is the top group permuting coordinates. In the specialization of the Wreath Embedding Theorem to the ternary case, preservation of a homogeneous TCAD is equivalent to the existence of a permutational isomorphism under which \[ G \le H^3 \rtimes K, \qquad H \cong G^{\Delta_i}\ \text{for some } i, \qquad K \le S_3. \] The kernel \(\ker(\varphi)\) acts componentwise, and \(\mathrm{Im}(\varphi)=K\) permutes components [2211.00418]. The paper emphasizes two model cases. For regular action, if \(H\) acts regularly on \(\Delta\), then \(R:=H^3\) acts regularly on \(\Omega=\Delta^3\); if \(K\) is nontrivial, the full wreath product is not regular, but \(R \le W\) remains a regular subgroup, \(R \cap K=1\), and \(W=R \rtimes K\). For diagonal groups, the simple diagonal subgroup \[ D=\{(h,h,h): h \in H\}\le H^3 \] acts simultaneously on all three coordinates, and the broader diagonal permutation group \(D(G,3)\) embeds into \(\mathrm{Sym}(\Delta)\wr S_3\). The worked examples with \(H=C_n\) and \(H=S_m\) illustrate the regular and diagonal patterns, respectively [2211.00418]. This usage of TCAD is thus a structural mechanism for realizing permutation groups inside wreath products in product action, with the ternary decomposition furnishing both the coordinate system and the induced homomorphism into \(S_3\). ## 3. TCAD as decomposition complexity of ternary functions In Shen’s framework, TCAD denotes decomposition of a ternary function into binary components with constrained intermediate alphabets. The basic model is \[ T(x,y,z)=t(a(x,y),b(y,z)), \] where \[ a:\mathbb{B}^p \times \mathbb{B}^q \to \mathbb{B}^u,\qquad b:\mathbb{B}^q \times \mathbb{B}^r \to \mathbb{B}^v,\qquad t:\mathbb{B}^u \times \mathbb{B}^v \to M. \] The decomposition complexity of \(T\) does not exceed \(n\) if there exist \(u,v \ge 0\) with \(u+v \le n\) and such functions \(a,b,t\). The notation \[ DC^{(2)}(T)=\min\{u+v:\exists a,b,t \text{ with } T(x,y,z)=t(a(x,y),b(y,z))\} \] captures the minimal total capacity of the two upper channels in the fan-in-\(2\) network [1012.1283]. The core results are worst-case upper and lower bounds. Let \(n=p+q+r\). The paper proves that the complexity of any function does not exceed \(n\), and that the complexity of any predicate does not exceed \(2^r+r\) as well as \(2^p+p\). It also proves an existence lower bound: if \(p\) and \(r\) are at least \(\log n + O(1)\), then there exists a predicate with decomposition complexity \(n-O(1)\). The counting argument is expressed through the inequality \[ \log m + m2^{p+q} + m2^{q+r} + 2^m \ge 2^n, \] which forces either \(m \ge n-O(1)\) or logarithmic constraints on \(m\) relative to \(p\) or \(r\) [1012.1283]. The explicit hard example is indexing. For \[ T(x,y,z)=y(x,z), \] where \(y \in \mathbb{B}^{2^{2k}}\) is treated as a Boolean function \(y:\mathbb{B}^k \times \mathbb{B}^k \to \mathbb{B}\), the paper proves \[ DC^{(2)}(T)\ge 2^k. \] The proof fixes \(y\), regards \(a_y\) and \(b_y\) as functions of \(x\) and \(z\), and observes that from \(a_y\), \(b_y\), and \(t\) one must reconstruct the entire table \(y\), yielding \[ 2^{u2^k}\cdot 2^{v2^k}\ge 2^{2^{2k}}, \] hence \((u+v)2^k \ge 2^{2k}\) and \(u+v \ge 2^k\) [1012.1283]. The examples sharpen the contrast between easy and hard decompositions. For parity, one bit for each channel suffices: \[ a(x,y)=\mathrm{parity}(x),\qquad b(y,z)=\mathrm{parity}(y)\oplus \mathrm{parity}(z),\qquad t(u,v)=u\oplus v. \] Thus \(DC^{(2)}(T)\le 2\). For equality \(T(x,y,z)=[x=z]\), the paper states that one must transmit \(x\) and \(z\) completely, so \(DC^{(2)}(T)\ge |x|+|z|\) [1012.1283]. The framework is robust under approximation. For \(\varepsilon \in (0,1/2)\), the paper proves that if \(p,r \ge \log n + O(1)\), then there exists a predicate for which every \(\varepsilon\)-approximation has decomposition complexity at least \(n-O(1)\), and for the indexing predicate any \(\varepsilon\)-approximation has \(DC^{(2)}(T)\ge \Omega(2^k)\). It also develops an algorithmic-information-theoretic version: if \(3c < n-O(1)\) and \(2m+c < 3n-O(1)\), then there exist \(x,y,z,t\) with \(K(t\mid x,y,z)=O(\log n)\) but no \(a,b\) of length \(m\) satisfying \[ K(a\mid x,y)0\), the Liutex direction \(r\) is the unit real eigenvector satisfying \[ Ar=\lambda_r r,\qquad \|r\|=1. \] Let \(\omega=\nabla \times u\) be the vorticity vector. The Liutex magnitude is \[ R=\omega\cdot r-\sqrt{(\omega\cdot r)^2-4\lambda_{ci}^2}, \] and the Liutex vector is \(R_{\mathrm{vec}}=Rr\). The rigid rotation tensor is then \[ L=\frac{1}{2}\,\widehat{R_{\mathrm{vec}}}, \] where the hat operation maps a vector to the associated skew-symmetric cross-product matrix. The stretching/compression tensor is \[ D=\lambda_{cr}I+(\lambda_r-\lambda_{cr})rr^T, \] with component form \[ D_{ij}=\lambda_{cr}\delta_{ij}+(\lambda_r-\lambda_{cr})r_ir_j. \] Finally, \[ S_{\text{pure}}=A-L-D. \] Equivalently, \(\Omega=L+\Omega_{\text{shear}}\) with \(\Omega_{\text{shear}}=\Omega-L=\frac12 \widehat{(\omega-R_{\mathrm{vec}})}\), and \(S=D+S_{\text{shear}}\) with \(S_{\text{shear}}=S-D\) [2106.10564]. At irrotational points, where \(A\) has three real eigenvalues, the Liutex part vanishes: \(R=0\) and \(L=0\). If \(Q\) is a real Schur orthogonal matrix such that \(Q^TAQ\) is real lower-triangular with diagonal entries \(\lambda_1,\lambda_2,\lambda_3\), then \[ D=Q\,\mathrm{diag}(\lambda_1,\lambda_2,\lambda_3)\,Q^T,\qquad S_{\text{pure}}=A-D. \] This preserves the decomposition directly in \(xyz\) coordinates [2106.10564]. The paper characterizes the three components precisely. \(L\) is skew-symmetric and trace-free, representing pure rigid-body rotation. \(D\) is symmetric and contains only normal strain, carrying the trace \[ \mathrm{tr}(D)=2\lambda_{cr}+\lambda_r=\mathrm{tr}(A) \] at rotational points. \(S_{\text{pure}}\) is trace-free and contains all remaining shear, both symmetric and antisymmetric. The paper further states that the inputs \(\omega\), \(R_{\mathrm{vec}}\), \(\lambda_r\), \(\lambda_{cr}\), \(\lambda_{ci}\), and \(r\), and the outputs \(L\), \(D\), and \(S_{\text{pure}}\), are Galilean invariant [2106.10564]. This formulation is presented as a replacement for classical scalar vortex diagnostics that mix strain and rotation. The decomposition provides an explicit separation of rigid rotation, pure shear, and stretching/compression without coordinate rotations or pointwise optimization in a separate principal frame. ## 5. The quantum-circuit terminology collision: Cartan, not Cartesian A recurrent source of confusion is the use of “TCAD” in the vicinity of ternary quantum logic. The cited qutrit-gate paper does not define a ternary Cartesian decomposition. Instead, it studies elementary gates for ternary quantum logic circuits and derives one-qutrit synthesis from Cartan decomposition of \(U(3)\) and \(SU(3)\). The paper introduces the ternary controlled-\(X\) (TCX) gate and ternary controlled-\(Z\) (TCZ) gate as two-qutrit elementary gates and states that either TCX or TCZ, together with arbitrary one-qutrit gates, is universal for ternary quantum computation [1105.5485]. The Lie-theoretic factorization is explicitly \[ U=K_1AK_2,\qquad A=\exp\!\big(i(\alpha \lambda_3+\beta \lambda_8)\big), \] with \(K_1\) and \(K_2\) drawn from a subgroup generated by embedded \(SU(2)\) rotations and diagonal phases. The explicit synthesis given in the paper is \[ U=e^{i\alpha}\;R_{y}^{(01)}(\beta)\,R_{y}^{(02)}(\gamma)\,R_{y}^{(01)}(\delta)\,R_{z}^{(01)}(\theta)\,R_{z}^{(02)}(\varphi)\,R_{y}^{(01)}(\beta')\,R_{y}^{(02)}(\gamma')\,R_{y}^{(01)}(\delta'). \] The paper then states a terminology note: the correct mathematical term is Cartan decomposition, not Cartesian decomposition, and if “TCAD” appears in the literature in the context of ternary gate synthesis, it should be understood as “Ternary Cartan Decomposition”-based synthesis, not “Cartesian” [1105.5485]. This distinction is substantive rather than merely lexical. The synthesis relies on the Cartan subalgebra of \(su(3)\), KAK factorization, and embedded \(SU(2)\) rotations on two-level subspaces. It is therefore conceptually unrelated to the Cartesian-product structures used in permutation groups or to the ternary function decompositions of decomposition complexity. ## 6. Comparative perspective, misconceptions, and open directions The sources collectively indicate that the main misconception surrounding TCAD is the assumption of a single accepted technical meaning. The literature considered here does not support that assumption. In one setting, TCAD is a decomposition of a set into three partition systems with singleton intersections; in another, it is a constrained communication/decomposition model for ternary functions; in a third, it is a tensor decomposition in Cartesian coordinates; and in qutrit synthesis the mathematically correct term is Cartan decomposition rather than Cartesian decomposition [2211.00418] [1012.1283] [2106.10564] [1105.5485]. A plausible unifying observation is that all of these constructions resolve a ternary object into structured lower-level components. In the group-theoretic case, the components are partitions or coordinates and the residual coupling is encoded by \(K \le S_3\). In decomposition complexity, the components are binary summaries \(a(x,y)\) and \(b(y,z)\) plus an aggregator \(t\). In fluid kinematics, the components are rigid rotation \(L\), pure shear \(S_{\text{pure}}\), and stretching/compression \(D\). This suggests a family resemblance at the level of architecture rather than a common formal theory. The most explicit open problems arise in decomposition complexity. The paper states that it would be interesting to get a linear bound for an explicit function in an intermediate case when \(x\) and \(z\) are short compared to \(y\), asks what happens if \(x \oplus z\) is replaced by \(x+z \bmod 2^k\) in the definition of \(T'\), asks for lower bounds for explicit ternary functions in the algorithmic-information-theoretic setting, asks for classical Shannon-information analogues of the multi-source Kolmogorov results, and asks whether the techniques of Hansen–Lachish–Miltersen can yield bounds for explicit functions in the algorithmic-information-theory setting [1012.1283]. In the permutation-group setting, the ternary case is distinguished by the fact that the top group embeds into \(S_3\), so the possibilities are limited to the trivial group, a \(2\)-cycle subgroup \(C_2\), a \(3\)-cycle subgroup \(C_3\), or \(S_3\). In the fluid-kinematic setting, the paper’s emphasis is not on open combinatorial questions but on the claim that the decomposition in the original \(xyz\) coordinates provides a Galilean-invariant separation of rotation, shear, and stretching/compression. Taken together, these usages show that “TCAD” functions as a domain-sensitive shorthand whose meaning must be fixed by context rather than by acronym alone [2211.00418] [2106.10564]. Source: https://www.emergentmind.com/topics/ternary-cartesian-decomposition-tcad