--- title: Exclusion-Rotation Method Overview url: https://www.emergentmind.com/topics/exclusion-rotation-method type: topic --- # Exclusion-Rotation Method Overview The available literature suggests that the expression **Exclusion-Rotation Method** is not a single standardized term but an umbrella label applied to several exclusion-based transformations in distinct research areas. The most formal construction close to that label is the **exclusion dilation operator** for bilateral claims problems, where lower and upper exclusion thresholds reshape the feasible allocation domain and a standard rule is reapplied on a reduced interval [2603.15015]. In parity-oblivious prepare-and-measure protocols, exclusion likewise changes the operational task, but the relevant paper explicitly states that it uses **dilation-free** additive Bloch geometry rather than a special rotation family [2605.08745]. In open-boundary ASEP, the pertinent methodology combines exclusion, boundary coupling, Laplace-space resolvents, and signed-permutation contour formulas, though the paper does not explicitly name an “Exclusion-Rotation Method” [1304.0800]. In forensic statistics, the label is used for a beta-binomial calibration that maps traditional ACE / ACE-V outputs—“identification,” “inconclusive,” and “exclusion”—to Bayes factors [2409.00451]. ## 1. Terminological scope and cross-domain usage In the bilateral claims literature, the operative term is **exclusion dilation operator**, not rotation. The paper states that the method “does not rotate the rule; it rescales it,” and describes the transformation as a geometric reparameterization in which the feasible region is compressed horizontally and vertically according to \(s_i=u_i-\ell_i\), after which the benchmark rule is evaluated on that compressed region and shifted back by the lower thresholds [2603.15015]. In the contextuality literature, the relevant work introduces the **parity-oblivious random exclusion code (POREC)** and explicitly notes that the derivations do **not** use “a nontrivial continuous rotation protocol or a special rotation family.” Instead, they rely on additive decomposition, orthogonal Pauli axes, convexity/extremality, Cauchy–Schwarz, and Bloch-sphere constraints [2605.08745]. A common misconception is therefore that the exclusion advantage is generated by a rotation code; the cited treatment rejects that interpretation. In open-boundary ASEP, the paper develops what it calls an exclusion-plus-boundary-plus-symmetry/contour framework. The summary states that the paper “does not explicitly name an ‘Exclusion-Rotation Method,’” but that it does exploit exclusion, boundary coupling, and symmetry/contour transformations to derive transition probabilities [1304.0800]. Here “rotation” is, at most, an interpretive analogy for contour or symmetry transformations. In fingerprint forensics, the label denotes something altogether different: a Bayesian calibration procedure that leaves the examiner’s categorical workflow largely unchanged but converts the outputs to Bayes factors by means of a beta-binomial model with uninformative or informative priors [2409.00451]. This is not a geometric transformation of a feasible region or state space; it is a probabilistic calibration of categorical decisions. ## 2. Bilateral claims problems and the exclusion dilation operator The most mathematically explicit exclusion-based construction appears in bilateral claims problems with thresholds. There are two agents with claims \[ c=(c_1,c_2)\in \mathbb{R}_+^2, \qquad C=c_1+c_2, \] a divisible endowment \(E\in\mathbb{R}_+\), and a standard allocation \(x=(x_1,x_2)\) satisfying \[ 0\le x_i\le c_i,\qquad x_1+x_2=E. \] The extension introduces lower exclusions \[ \ell=(\ell_1,\ell_2)\in [0,1)^2 \] and upper exclusions \[ u=(u_1,u_2)\in (0,1]^2, \] with \(\ell_iU(c,u)\), resources are abundant enough that one agent is fully compensated in the loss regime; and between these thresholds the rule behaves like a transformed version of the original rule [2603.15015]. The key geometric quantity is \[ s=(s_1,s_2), \qquad s_i=u_i-\ell_i, \] which measures the remaining feasible “exclusion space” after thresholds are imposed. For \(a,b>0\), the dilation of a function \(f\) is \[ f(z;a,b)=b\,f(z/a). \] In the claims model, the underlying rule is reapplied on the reduced interval \[ [0,U(c,u)-L(c,\ell)]. \] This is the precise sense in which the method is a dilation-style, rather than rotation-style, transformation. For any standard rule \(R\), the exclusion dilation operator defines the extended rule by three cases: \[ R_i(c,E,\ell,u)= \begin{cases} \dfrac{\ell_i c_i}{L(c,\ell)}\,E, & \text{if } EU(c,u). \end{cases} \] The middle term is defined through \[ y_1^R+y_2^R=E-L(c,\ell), \] and \[ y_2^R = R(c,E)(y_1^R;s), \] with \(y_i^R\in[0,s_ic_i]\). Operationally, the procedure has three stages: proportional allocation with respect to lower exclusions below \(L(c,\ell)\); application of the dilated underlying rule in the interval \([L(c,\ell),U(c,u)]\); and proportional distribution of residual compensation above \(U(c,u)\). The paper works out explicit extended forms for several benchmark rules: the proportional rule, constrained equal awards (CEA), constrained equal losses (CEL), concede-and-divide (CD), reverse Talmud (RT), a nonlinear rule, and a differentiable self-dual rule. For the proportional rule, the middle-region path is especially simple: \[ P(c,E)(y;s)=\frac{s_2}{s_1}y, \] yielding \[ P_i(c,E,\ell,u)= \begin{cases} \dfrac{\ell_i c_i}{L(c,\ell)}E, & EU(c,u). \end{cases} \] ## 3. Property preservation, asymmetry, and axiomatic characterization A central feature of the exclusion dilation operator is that it preserves some standard claims axioms while deliberately violating others. The paper states in Theorem 1 that **equal treatment of equals** and **order preservation** are **not preserved**, whereas **endowment monotonicity**, **claim monotonicity**, and **homogeneity** are preserved [2603.15015]. The stated intuition is that exclusions are asymmetric by design, so equality or order in claims need not translate into equality or order in awards. Additional preservation results are conditional. **Midpoint** and **self-duality** are preserved if exclusions are symmetric: \[ \ell_i = 1-u_i \quad \forall i. \] **Restricted endowment convexity** is preserved only within the exclusion space. **Progressivity, regressivity, concavity,** and **convexity** are preserved within the exclusion space if exclusions are order-preserving, meaning that the rank order of claims is not reversed after threshold subtraction. These statements define the operator’s normative profile: it is not a symmetry-preserving extension of ordinary claims rules, but a controlled asymmetry mechanism. The axiomatic characterization is given in Theorem 2. An extended allocation rule is generated by the exclusion dilation operator if and only if it satisfies three axioms. The first is **full exclusion**: \[ \text{if } \ell_i c_i \ge E,\quad \delta_j(c,E,\ell,u)=0,\qquad i\ne j. \] The second is **null exclusion**: \[ \text{if } u_i c_i \le E-c_j,\quad \delta_j(c,E,\ell,u)=c_j,\qquad i\ne j. \] The third is **proportional exclusion invariance**: \[ \delta_i(c,E+M(c,m),\ell,u) = s_i\,\delta_i(c,E,(0,0),(1,1)) + m_i c_i, \] with \[ M(c,m)=m_1c_1+m_2c_2. \] The paper also shows that these three axioms are independent by constructing counterexamples when any one of them is dropped. A plausible implication is that the operator is not merely one convenient implementation among many; within the bilateral framework, the three axioms uniquely pin down the transformation. The applications suggested in the paper include bankruptcy with privileged public claims, tax or exemption systems, divorce settlements, education or welfare budgets, and other policy problems where one party must be protected differently from another. These applications are presented as contexts in which “symmetry is not appropriate due to legal and policy considerations.” ## 4. Exclusion in parity-oblivious prepare-and-measure protocols In quantum contextuality, exclusion enters through the parity-oblivious random exclusion code. Alice receives a uniformly random string \[ x=x_1x_2\cdots x_n\in\mathbb{Z}_m^n, \] with prime \(m\ge 3\), and Bob receives a uniformly random index \[ y\in\{1,\dots,n\}. \] In parity-oblivious retrieval, Bob must output \(b=x_y\); in POREC, he must output any value satisfying \(b\neq x_y\). Parity-obliviousness requires that for masks \(r\in\mathbb{Z}_m^n\) with Hamming weight \(\mathrm{wt}(r)\ge 2\), the average encoding over each parity class be independent of the parity value. In the quantum case, \[ \rho_k^{(r)}=\rho_{k'}^{(r)}\quad \forall k,k'\in\mathbb{Z}_m. \] The figure of merit is \[ P_{\mathrm{POREC}}=\frac{1}{nm^n}\sum_{x\in\mathbb{Z}_m^n}\sum_{y=1}^n P(b\neq x_y\mid x,y). \] For prime \(m\), the optimal classical and preparation-noncontextual value is \[ P_{\mathrm{POREC}^{\mathrm{NC}}}=1-\frac{n-1}{mn}. \] The proof uses a Fourier-collapse argument: parity-obliviousness kills all Fourier components with \(\mathrm{wt}(r)\ge 2\), and for prime \(m\) this forces the encoding to depend on at most one digit [2605.08745]. For the first nontrivial case \((n,m)=(2,3)\), the paper gives the exact qubit optimum \[ P_{\mathrm{POREC}^{Q}(2)}=\frac{2}{3}+\frac{1}{3\sqrt{2}}, \] numerically \(0.902369\), exceeding the noncontextual bound \[ P_{\mathrm{POREC}^{\mathrm{NC}}}=\frac{5}{6}\approx 0.833333. \] The optimal qubit realization is in the Bloch \(x\)-\(z\) plane, with states arranged as a \(3\times 3\) grid, one digit encoded along \(\sigma_z\), the other along \(\sigma_x\), and measurements \[ M_{b|y}=\frac{1}{2}\bigl(I+(-1)^b\,\hat m_y\cdot\vec\sigma\bigr), \qquad \hat m_1=\hat z,\quad \hat m_2=\hat x, \] with \(M_{2|y}=0\). The paper states that the advantage comes from **anti-aligned measurements** for exclusion. The paper’s comparison with parity-oblivious retrieval is especially important. It states that retrieval shows no qubit advantage under the same parity-oblivious constraints, whereas exclusion does. The difference is operational: exclusion only requires Bob to rule out the correct value, while retrieval requires full identification. The paper further states that there is **no rotation-like encoding beyond** the additive Bloch structure \[ \rho_{x_1x_2}=\frac12\bigl(I+(\vec a_{x_1}+\vec b_{x_2})\cdot\vec\sigma\bigr), \] with the zero-sum gauge \[ \sum_{x_1}\vec a_{x_1}=0,\qquad \sum_{x_2}\vec b_{x_2}=0. \] A common misunderstanding is therefore that exclusion-based contextuality protocols derive their advantage from a continuous rotation code. The cited analysis states the opposite: the mechanism is additive decomposition plus orthogonal, complementary measurements. The exact qubit optimum also yields a semi-device-independent dimension witness: \[ P_{\mathrm{POREC}} > 0.902369 \quad\Rightarrow\quad d\ge 3. \] The paper reports nonzero critical white-noise thresholds \(\omega_c^{(d)}\) in all tested cases for \((2,m)\), with \(\omega_c^{(d)}\) increasing with dimension. ## 5. Exclusion, boundary coupling, and symmetry in open-boundary ASEP A separate lineage arises in the asymmetric simple exclusion process on \(\mathbb Z_{\ge 0}\) with an open boundary at \(0\). Particles obey the exclusion rule, move right with probability \(p\), move left with probability \(q=1-p\), are injected at site \(0\) from a reservoir at rate \(\alpha\) if site \(0\) is empty, and are ejected from site \(0\) into the reservoir at rate \(\beta\). Because of the reservoir, particle number is not conserved: a system that starts in \(\mathcal X_m\) may later be in \(\mathcal X_n\) [1304.0800]. The paper studies transition probabilities \(\mathfrak p(\mathbf x,\mathbf y;t)\) and their Laplace transforms \[ \widehat{\mathfrak p}(\mathbf x,\mathbf y;s)=\int_0^\infty \mathfrak p(\mathbf x,\mathbf y;t)e^{-st}\,dt. \] The central open-boundary formula is the Laplace-space resolvent \[ P(s)=\Big(I-L(s)\,((\alpha-\beta)+\alpha A+\beta B)\Big)^{-1}\,L(s)\,P(0). \] Here \(A_n\) and \(B_n\) are boundary operators coupling neighboring particle-number sectors. When one boundary rate vanishes, the matrix becomes triangular and the paper gives explicit recursions. For \(\beta=0\), \[ P_k(s)=M_k(s)L_k(s)\,\delta_{\mathbf y}, \qquad P_n(s)=\alpha\,M_n(s)L_n(s)A_nP_{n-1}(s)\quad (n>k), \] and for \(\alpha=0\), \[ P_k(s)=M_k(s)L_k(s)\,\delta_{\mathbf y}, \qquad P_n(s)=\beta\,M_n(s)L_n(s)B_nP_{n+1}(s)\quad (n0\), the open-boundary problem is rewritten in terms of lower-dimensional operators involving kernels such as \(\widehat{\mathfrak p}((0,\mathbf x),(0,\mathbf y);s)\). This is the main boundary-reduction method in the paper. The symmetry/contour aspect appears in the appendix through the Weyl group \(\mathbb B_n\) of signed permutations. For \(q\neq 0\), the transition probability is \[ \mathfrak p(\mathbf x,\mathbf y;t)=\frac{1}{n!}\sum_{\sigma\in\mathbb B_n}\frac{1}{(2\pi i)^n}\int\cdots\int A_\sigma(\xi)\,\prod_i\Big(\xi_i^{x_i}\,\xi_i^{-y_i-1}e^{\varepsilon(\xi_i)t}\Big)\,d\xi_1\cdots d\xi_n. \] For TASEP \((p=1)\), this simplifies to the determinant formula \[ \mathfrak p(\mathbf x,\mathbf y;t)=\det\left(\int_{\mathcal C_r}(1-\xi)^{j-i}\xi^{x_i-y_j-1}e^{t\varepsilon(\xi)}\,d\xi\right). \] The paper therefore supplies an exclusion-plus-boundary-plus-symmetry/contour method, but not a literal rotation construction. A plausible implication is that the phrase “rotation” can only be used metaphorically here, to denote contour or group-action transformations rather than geometric rotations of states or paths. ## 6. Bayesian calibration of “identification,” “inconclusive,” and “exclusion” In forensic statistics, the method labeled **Exclusion-Rotation Method** is a beta-binomial calibration of traditional fingerprint ACE / ACE-V outputs into Bayes factors. The conceptual move is to keep the familiar categorical workflow—“identification,” “inconclusive,” or “exclusion”—but to map each response to a Bayes factor using empirical examiner behavior under known same-source and different-source conditions [2409.00451]. For each response category \(RS\in\{\mathrm{ID},\mathrm{IN},\mathrm{EX}\}\) and truth state \(t\in\{s,d\}\), the model uses counts \(c(RS\mid t)\) out of \(n_t\) opportunities: \[ c(RS\mid t)\sim \mathrm{Bin}\bigl(n_t,\theta(RS\mid t)\bigr), \qquad \theta(RS\mid t)\sim \mathrm{Beta}(a_t,b_t). \] By conjugacy, \[ \theta(RS\mid t)\mid \text{data}\sim \mathrm{Beta}\bigl(c(RS\mid t)+a_t,\ c(-RS\mid t)+b_t\bigr), \] where \(c(-RS\mid t)=n_t-c(RS\mid t)\). The posterior mean is \[ \mathbb E[\theta(RS\mid t)\mid \text{data}] = \frac{c(RS\mid t)+a_t}{n_t+a_t+b_t}. \] The Bayes factor assigned to response \(RS\) is then \[ B_{RS} = \frac{\mathbb E[\theta(RS\mid s)\mid \text{data}]} {\mathbb E[\theta(RS\mid d)\mid \text{data}]}. \] For uninformative priors, the paper uses weighted Jeffreys-type choices \[ a_s=b_s=\frac{n_s}{n_s+n_d}, \qquad a_d=b_d=\frac{n_d}{n_s+n_d}. \] For informative priors, it uses a leave-one-out cross-validated procedure in which posterior hyperparameters from the remaining examiners are averaged and used as prior hyperparameters for the held-out examiner. The proof-of-concept uses Langenburg et al. data with \(n_s=7\) same-source pairs and \(n_d=5\) different-source pairs. For one illustrative examiner, under uninformative priors, the counts for “identification” are \(c(\mathrm{ID}\mid s)=5\) and \(c(\mathrm{ID}\mid d)=0\), giving posterior means \(0.684\) and \(0.0714\), and therefore \[ B_{\mathrm{ID}}=9.57. \] Under informative priors \(a_s=4.93\), \(b_s=3.24\), \(a_d=0.591\), \(b_d=5.24\), the posterior means become \(0.655\) and \(0.0545\), yielding \[ B_{\mathrm{ID}}=12.0. \] The paper reports analogous examples for \(B_{\mathrm{IN}}\) and \(B_{\mathrm{EX}}\), and emphasizes that “inconclusive” often does **not** correspond to neutrality \(B\approx 1\). In the example dataset, “inconclusive” often produced values above 1, indicating that it was somewhat more associated with same-source than different-source evidence. This use of the term differs fundamentally from the allocation, contextuality, and ASEP usages. There is no rotated geometry or reparameterized path. The method is a calibrated evidential mapping from categorical outputs to Bayes factors. Its stated advantage is that the examiner’s workflow changes minimally, while the evidential interpretation becomes probabilistically explicit. ## 7. Comparative assessment Across these literatures, the common element is **exclusion**, but the operative mechanism varies sharply. In bilateral claims problems, exclusion thresholds carve out a reduced feasible interval and the underlying rule is **dilated** into that interval. In POREC, exclusion is a task requirement that reveals preparation contextuality under parity-obliviousness; the paper explicitly denies the use of a special rotation family. In open-boundary ASEP, exclusion is the particle interaction itself, and the key tools are boundary operators, resolvent inversion, and signed-permutation contour formulas. In fingerprint calibration, exclusion is one categorical response among three, and the method is a beta-binomial conversion to Bayes factors. The strongest unifying interpretation is therefore methodological rather than formal. These constructions use exclusion to reshape either a feasible domain, an operational success criterion, a boundary-coupled stochastic evolution, or an evidential response scale. What varies is the transformation that follows: dilation in claims, additive Bloch encoding in contextuality, contour/symmetry reduction in ASEP, and Bayesian calibration in forensics. The literature thus supports no single canonical “Exclusion-Rotation Method.” It instead supports a family of exclusion-centered procedures whose mathematical content depends entirely on the domain in which the term is invoked. Source: https://www.emergentmind.com/topics/exclusion-rotation-method