PROSE AID ONLY — use PDF/page images and exact source for mathematics. PDF PAGE 1 Contextuality and Exact Information Resources over Finite Chain Rings Brian Theory B-Theory 26 September 2026 Abstract We study nonzero bipartite matrices over a finite commutative chain ring using all primitive projective local bases, setting-indexed outcomes and nonzero-amplitude support. Two Smith invariants determine the support type: the numberr of nonzero factors and the multiplicity h of the least exponent. The model is local whenr = 1, logically but not strongly contextual when r≥ 2 and h = 1, and strongly contextual whenh≥ 2. A residue-hyperplane criterion and a valuation-uniform Hardy construction give the classification. For a squared-dimensional resource, invertible local orbit encoding and one arbitrary complete joint basis readout admit exactly dh messages. Universal exact raw-vector teleportation, with a complete joint covector basis and unrestricted invertible corrections, instead requires an invertible resource. We prove the existence of complete invertible-matrix analyzers over finite local rings and exhibit nonprimitive resources separating maximal orbit coding from teleportation. For primitive resources, maximal coding and universal exact raw teleportation coincide. Tensor choice and preparation closure delimit the interpretation. Finite checks accompany the proofs; no physical probability model or priority claim follows from them. 1 Introduction Replacing complex amplitudes by a finite scalar system makes it possible to separate algebraic support from probability. In modal quantum theory, an effect is possible when its contraction with a state is nonzero. Field models already support entanglement, nonlocal support and communication protocols [9, 10]. Passing from a field to a ring introduces a different question: which aspects of a resource survive when nonzero amplitudes can annihilate one another? This paper answers that question for static bipartite resources over finite commutative chain rings, with complete local basis families. The classification uses two elementary pieces of Smith data. At least two nonzero Smith factors are necessary and sufficient for a Hardy obstruction. Leading multiplicity detects whether any global outcome assignment survives. Under a separately stated exact communication task, the same leading multiplicity measures the number of linearly distinguishable orbit messages. Universal exact teleportation distinguishes yet another property, invertibility. The contribution is the complete chain-ring classification and the comparison of contextuality, exact orbit coding and raw-vector teleportation through related Smith invariants under explicit contracts. Smith theory, the language of support contextuality, Hardy implications and modal protocols are established ingredients. The results do not propose a physical theory, and do not infer a Born rule. In particular, allowing all nonzero ring vectors is useful for static resource questions but does not define a preparation theory closed under arbitrary balanced tensor products. 1 PDF PAGE 2 The proofs are self-contained. The source comparison below identifies established ingredients and the hypotheses that distinguish the present tasks. It makes no claim of historical firstness. A standalone computational supplement supplies finite checks with their exact scope; an optional Boolean-interface note is separate from the article and is not a dependency of its results. 1.1 Relation to established work Schumacher and Westmoreland’s two-mobit model already gives a nonlocal support table and a four-message orbit code [9]; their later treatment develops generalized field effects and modal probability-completion questions [10]. These supply background, not new claims here. The support hierarchy follows the global-section formulation of Abramsky and Brandenburger [1]. Our settings are local bases. We do not additionally require the same effect ray to carry a context-independent truth value across distinct bases. That distinction matters for comparisons with modal noncontextuality and free-will arguments [8] and the uncolorability of integer-matrix idempotents [2]. The latter concerns compatible idempotent decompositions and transports to arbitrary scalar rings; it is not a Smith classification of a fixed bipartite state with basis-indexed outcomes. Finite-chain-ring module structure and projective ring geometry are likewise classical [7, 3]. The use of primitive rows is a standard local-ring construction. Our assertions concern their complete bilinear support tables and two specified information tasks. State admissibility is a substantive difference from other ring models. In de Beaudrap’s generic state space, the coordinates must generate the unit ideal [4]. Over a local ring this means that at least one coordinate is a unit. Our nonprimitive nilpotent separator is therefore excluded from that state space; its conclusions belong to the static all-nonzero-vector contract used here. Categorical finite-field protocols can also use a different measurement notion: Varughese’s Example 8.3 uses three Bell outcomes on a nine-dimensional pair [11]. It does not provide the complete nine-row covector analyzer required by our teleportation contract. The matrix-spanning ingredient is classical: nonsingular-summand results go back to Wolfson and Zelinsky, and Henriksen proves the stronger statement that every square matrix of size greater than one over a unital ring is a sum of three units [12, 6]. Lemma 4.1 supplies a short proof sufficient for our use, followed by residue-basis lifting. Neither this spanning fact nor the existence of modal protocols is asserted as a new general result. Finite-chain-ring coding also uses module invariants, but its optimization problem must be distinguished. Feng et al. study a stochastic channelY =AX +BE with random transfer and error matrices, deriving capacity results under specified distributions [5]. Here a sender controlsU inUM for a fixed resourceM, and a single complete joint basis readout must identify the message at every supported outcome. A direct consequence of the former channel definition makes the distinction concrete: ifA is uniform inGLd(K) and noise is absent, thenAUM has the same distribution for everyU∈ GLd(K). This observation is our comparison of the two tasks, not a claim made in that source about the present theorem. 2 PDF PAGE 3 Source and domain Admissible objects Operations, readout and conclusion Modal theory [9, 10]; fields Nonzero state vectors; composite vector spaces Reversible linear maps and basis effects; nonzero contraction gives possibility. Includes finite-field coding and teleportation examples. Ring computation [4]; commutative rings Generic vectors whose coordinates generate the unit ideal Model-dependent algebraic transformations and outcome rules; this admissibility excludes our nonprimitive separator. Chain-ring codes [7, 3] Submodules and their module types Code equivalence, geometry and enumeration; no fixed-resource joint support readout or raw teleportation task. Matrix channels [5]; chain rings Input matrices in a prescribed ambient module Random A,B,E in Y =AX +BE; stochastic decoding and capacity, rather than a controlled orbit with disjoint nonzero supports. This article; chain rings Every nonzero static matrix, including nonprimitive ones All local bases for support; fullGLd(K) encoders and one complete joint decoder for exact coding; complete covector analysis and invertible corrections for exact raw recovery. Conclusions: Smith trichotomy,dh messages, and invertibility criterion. Table 1: Comparison by scalar domain, admissibility and task. The support hierarchy itself is the established global-section hierarchy [1]. Theorem 4.4 also applies to finite commutative local rings that are not chain rings. 2 Scalar, state and measurement contracts LetK be a finite commutative chain ring with identity, maximal idealJ = (π), residue fieldk =K/J of sizeq, and nilpotency indexs, soπs = 0 and πs−1̸= 0. The field case hass = 1. Bars denote reduction moduloJ. An element is a unit exactly when its residue is nonzero. A square matrix is invertible exactly when its residue is invertible, by the determinant criterion. Every nonzero element is a unit times a unique powerπa, 0≤a < s. Repeatedly moving an entry of minimal valuation to the first position and using elementary row and column operations gives Smith form PMQ T = diag(πa1,...,π ar, 0,... ), 0≤a1≤···≤ ar 0; its rank is then not the leading multiplicity. Definition 2.1 (Complete-basis support). A primitive row inKm has a unit coordinate. Its projective ray is its orbit under unit scaling. A local setting is an unordered basis of primitive projective rows, represented by an invertible matrix. For a nonzero resourceM∈Km⊗K Kn∼= Matm×n(K), the outcome pair represented by rowsx,y is supported exactly when xMy T̸= 0. 3 PDF PAGE 4 All such local settings are admitted, and local reversible changes are the full groupsGLm(K) and GLn(K). Unit rescaling does not change support. Each context has a supported event because invertible changes cannot send nonzeroM to zero. The support marginal forx is independent of Bob’s basis: some coordinate ofxMB T is nonzero if and only ifxM̸= 0. Hence these support tables satisfy possibilistic no-signalling, without any assigned real probabilities. A global assignmentchooses one outcome for each local setting, with every chosen cross pair supported. A model islocal here if every supported event extends to such an assignment. It is logically contextualif some event does not extend, andstrongly contextualif no global assignment exists. This is a support-level definition; “local” is not a statement about a particular probability distribution. Two further tasks use the same scalars but different measurements. Their distinctions are essential. Task Contract Support Nonzero staticM; balanced tensor overK; all primitive projective local bases; nonzero contraction means possible. Orbit coding SquareM,d≥ 2; encodingsUM withU∈ GLd(K); one invertibleK-linear joint decoder ond2 coordinates; every possible outcome identifies the message. No discard and reprepare. Teleportation Every nonzero raw input, including correlations with an untouched reference; complete basis of primitive joint covectors; all possible branches corrected exactly by allowed invertible maps. No selected successful branch or postselection. For support and coding, multiplying a whole resource by a unit is immaterial. Teleportation below demands actual vector recovery and does not silently replace this by an unspecified projective task. 2.1 Resource map and a worked comparison Table 2 previews the consequences of Theorems 3.3, 4.2 and 4.4 for square resources. Primitivity of M, viewed as a vector of entries, is equivalent toa = 0. Smith region Orbit messages Support and raw teleportation r = 1 d Local; singular ford≥ 2 r≥ 2, h= 1 d Logical, not strong; singular 2≤h 0 d2 Strong; nonprimitive, singular h =d, a= 0 d2 Strong; invertible, universally teleporting Table 2: Square resources under the stated contracts,d≥ 2. Zero is excluded. ForA = F2[u]/(u4), compare the four resources in Table 3. InM = diag(u,u ) =uI2, a choice of leading preimage isN =I2, soa = 1 and h = 2 even though ¯M = 0. All four orbit messages remain distinguishable, but the nonzero input(u3, 0)T is annihilated byM and cannot be universally recovered. By contrast,diag(1,u 2) is primitive and has two nonzero factors, but only one leading factor; it has a Hardy obstruction and also a global assignment. These examples distinguishr, h and invertibility without changing the scalar ring. 4 PDF PAGE 5 M overA r a h Support Messages Teleports diag(1, 0) 1 0 1 Local 2 No diag(1,u 2) 2 0 1 Logical, not strong 2 No diag(u,u ) 2 1 2 Strong 4 No I2 2 0 2 Strong 4 Yes Table 3: A worked comparison over one chain ring. Teleportation means universal exact raw-vector recovery. 3 Complete-basis contextuality The first lemma requires only thatK be finite, commutative and local. Lemma 3.1(Residue-hyperplane criterion). A complete-basis support model has a global assignment if and only if there are hyperplanesHA⊂km and HB⊂kn such thatxMy T̸= 0 for every pair of primitive lifts whose residues lie outsideHA and HB. Proof. Let UA be the union of the rays selected by a global assignment. Its complement contains no basis, since the corresponding setting would have to select a ray in that complement. If the reductions of the complement spannedkm, one could select a residue basis from them; its lifts would form an invertible basis overK. Thus those reductions lie in a proper subspace, which is contained in a hyperplaneHA. Every primitive lift of a residue vector outsideHA belongs toUA. Construct HB in the same way. Every pair fromUA×UB is selected in some pair of settings, and is therefore supported. Conversely, any basis has a row reducing outside a given hyperplane. Select one in every setting on each side. The stated condition makes every chosen cross pair supported, giving a global assignment. Lemma 3.2(Uniform Hardy witness). ForD = diag(p,q 0) with p =πa, q0 =πb̸= 0 and a≤b, put c =πb−a. The local bases F = ( 0 1 1 0 ) , X = ( 1 0 1 1 ) , C = ( 1 0 −c 1 ) give a supported event with no global extension, usingF,C for Alice andF,X for Bob. Proof. All three determinants are units, andpc =q0. Direct multiplication gives FDF T = ( q0 0 0 p ) , FDX T = ( 0 q0 p p ) , CDX T = ( p p −q0 0 ) , CDF T = ( 0 p q0 −q0 ) . Index outcomes by0, 1. The forcing chain from the supported seed is (FA = 0,F B = 0) =⇒ XB = 1 =⇒ CA = 0 =⇒ FB = 1, a contradiction. Each implication excludes an outcome using, successively, the zero entries inFDX T, CDX T and CDF T. The support strings, in row order, are1001, 0111, 1110, 0111. The minus sign cannot be dropped outside characteristic two. 5 PDF PAGE 6 Theorem 3.3(Smith trichotomy). For nonzeroM∈ Matm×n(K), the complete-basis support type is r = 1 ⇐⇒ local, r≥ 2, h = 1 ⇐⇒ logical but not strong, h≥ 2 ⇐⇒ strong. Proof. Invertible left and right changes merely relabel the complete basis families, so take Smith form. Supposer = 1, with sole nonzero entryp in position(1, 1). Retain the outcomes of any supported seed event(x,y ). In every other setting choose a row with unit first coordinate. A new–new pair has contraction equal top times a unit. The seed conditionpx1y1̸= 0 implies px1̸= 0 and py1̸= 0, so a seed–new pair is also supported. The seed therefore extends. If h = 1, choose a row with unit first coordinate in every setting. In the divided matrix, each selected contraction is a product of units plus terms in the radical, hence is a unit. Multiplication by the nonzeroπa gives a supported cross pair. A global assignment exists. If h≥ 2, writeM =πaM0, so ¯M0 has rankh. Suppose the hyperplanes of Lemma 3.1 existed, and writeHB = kerβ. The vectors outsideHA span km. Since the row space of¯M0 has dimension at least two, somex /∈HA has w =x ¯M0 outside the line spanned byβ. There isy∈ kerw with β(y)̸= 0. Liftx,y to rows ˜x, ˜y. The row ˜xM0 has a unit coordinate, whereas its contraction with˜y lies inJ. Subtracting that contraction divided by the unit from the corresponding coordinate of ˜y cancels it exactly and changes no residue. This yields lifts outside both hyperplanes with zero contraction, a contradiction. Finally, ifr≥ 2, embed Lemma 3.2 in the first two nonzero Smith coordinates and extend its bases with standard rows. The embedded matrices are block diagonal. At each forcing step, the already selected row is supported only in the first two coordinates and therefore has zero contraction with every added coordinate outcome on the other side. No additional outcome can escape the forcing chain. Thus the same seed cannot extend, even in higher dimensions. These observations exhaust all nonzero cases. There areqsm−q(s−1)m primitive rows inKm. Unit scaling acts freely, and|K×| =qs−1(q− 1). Consequently the numbers of rays and unordered projective bases are q(s−1)(m−1)qm− 1 q− 1 , qm2(s−1)| GLm(k)| |K×|mm! . ForK = F2[u]/(u4) andm = 2, these are24 rays and192 bases. These counting facts are background; enumeration is unnecessary for the classification proof. 4 Exact orbit coding and teleportation The following classical spanning fact is included with a constructive proof; stronger units-generation results are available in [12, 6]. Lemma 4.1 (Invertible matrices span). For a fieldk and d≥ 2, GLd(k) spans Matd(k). Over a finite commutative local ringK, there is aK-basis of Matd(K) consisting entirely of invertible matrices. Proof. Fori̸=j, Eij = (I +Eij)−I. For a diagonal unitEii, choose a permutation matrixP with Pe j =ei, j̸=i. ThenP (I +Eji)−P =Eii, again a difference of invertibles. This works over every field, including F2. Selectd2 independent invertibles from this spanning family overk and lift their 6 PDF PAGE 7 entries toK. Each lifted matrix is invertible. The square matrix of their flattened vectors is also invertible, because its residue is a basis matrix. The lifts therefore form the asserted basis. Theorem 4.2 (Leading-layer orbit codebook). For M∈ Matd(K)\{ 0}, d≥ 2, the maximum number of messages under the orbit-coding contract of Section 2 isNmax =dh. Proof. WriteM =πaN. The leading residue of each divided codewordUN lies in V ={X ¯N :X∈ Matd(k)}, dimkV =dh. Each row ofX ¯N belongs to theh-dimensional row space of ¯N, and all such rows can be chosen independently, proving dimkV = dh. Put vi = vec(UiN) and letD∈ GLd2(K) be a proposed decoder. The vectors ¯vi and ¯D¯vi are nonzero. If a coordinate ofDvi has nonzero residue, it is a unit, and multiplication byπa̸= 0 cannot kill it. Thus supp( ¯D¯vi)⊆ supp(πaDvi). Exact decoding requires the supports on the right to be pairwise disjoint. Consequently the nonzero field vectors on the left have disjoint supports and are linearly independent. They lie in the dh-dimensional space ¯D vec(V ), giving the upper bound. The reverse support inclusion is neither asserted nor needed. By Lemma 4.1, images of invertible residue matrices spanV. Choosedh of these images that are independent and lift the encoders toU1,...,U dh∈ GLd(K). The columnsvec(UiN) are independent modulo J. Extend the residue columns to a basis and lift the added columns arbitrarily. The resulting matrixB overK is invertible. Its inverse sends the actual codewords toπaei, each with a unique possible coordinate outcome. This attains the bound. Corollary 4.3. With the complete-basis support and orbit-coding contracts matched,Nmax >d if and only ifM is strongly contextual. Proof. Both conditions are equivalent toh≥ 2. Thed-message comparison is the dimension baseline for a freshd-coordinate carrier with complete basis readout. It is not an entropy or an asymptotic channel capacity. If encoders may discard the resource, or decoders are restricted, the stated optimization changes. Over a fieldh =r, giving the familiar linear-algebra orbit countdr; the two-mobit four-message protocol is already known [9]. A joint effect is a primitive covector e :Kd⊗KKd−→K, E ij =e(ei⊗ej), where the first factor holds the input and the second Alice’s share of the resource. A complete analysis is a basis ofd2 such covectors. Its analysis matrix has rowsvec(E)T and must be invertible. This is different from requiring an individual reshapeE to have matrix rank one:E may be invertible and need not represent a decomposable local effect. ForM =∑ j,kMjkej⊗ek, contraction gives (Tex)k = ∑ i,j EijxiMjk, T e =M TET. The branch label is communicated to Bob, who applies a fixed correction for that label. Theorem 4.4(Universal raw teleportation). LetK be a finite commutative local ring. For nonzero M∈ Matd(K), d≥ 2, universal exact raw-vector teleportation with a complete joint covector basis and unrestricted invertible corrections is possible if and only ifM is invertible. 7 PDF PAGE 8 Proof. Use the contraction convention above, so a branch with reshapeE has transferT =M TET. If a nonzero branch map had a nonzero kernel vectorz, choosex withTx̸= 0. Bothx andx +z are nonzero and give the same possible branch output. A fixed correction cannot recover both. Thus each nonzero universally correct branch is injective. On the finite free moduleKd, it is bijective and its inverse is linear; its invertibility forcesM to be invertible. A complete effect basis and nonzero M exclude the possibility that every branch map is zero: the reshapes span all matrices, so this would forceM TEij = 0 for alli,j and henceM = 0. Conversely, choose the invertible-matrix basis of Lemma 4.1 as the reshapes of the joint effects. It is a complete analysis basis, and every transferM TET is invertible. Correct with its inverse. The identityT −1T =I remains an identity after tensoring with the identity on an untouched reference, proving the reference-preserving statement as well. The analyzer existence step is a residue-lifting corollary of the classical spanning fact. It requires both invertibility of each reshape and invertibility of the full analysis matrix. Its inclusion removes a dimension restriction from sufficiency when all the required bases and corrections are admitted; it does not assert that a physically restricted gate set can implement them. Corollary 4.5(A nonprimitive separator). For a chain ring of lengths> 1 and 0 2. 5 Boundaries of the resource comparison 5.1 Balanced tensors and regular binary expansion In A = F2[u]/(u4), balanced scalar multiplication identifiesA⊗AA with A. Therefore u3⊗Au = 0, u 3⊗F2u̸= 0. The second assertion follows by applying field-linear functionals that are nonzero on each factor. Nonzero preparations thus need not remain nonzero under the balanced tensor. Primitive vectors do remain primitive, since a product of unit coordinates is a unit. Yet that restriction is not closed under the proposed conditional support update: applying the second standard effect todiag(1,u ) gives the nonprimitive nonzero vector(0,u ). 8 PDF PAGE 9 The regular representation ofA is a faithful algebra map to4× 4 binary matrices. It is not a monoidal equivalence between the balanced and literal tensor spaces. Expanding a Smith factorua gives binary rank4−a, whence rankF2λ(M) = r∑ i=1 (4−ai). Unrestricted independent binary row and column operations classify a matrix only by that total rank. They can consequently erase distinctions detected by restrictedA-linear operations. For instance,diag(1,u 2) and diag(u,u ) both have binary rank six but leading multiplicities one and two. To avoid relying on a nonprimitive comparison, the primitive matrices diag(1, 1,u 3), diag(1,u,u 2) both have binary rank nine, while their leading multiplicities are two and one. TheirA-linear orbit codebooks have sizes six and three; both are singular, and hence neither universally teleports in the stated task. 5.2 A boundary on the number of settings Theorem 5.1 (Absence of strong contextuality with two settings). A nonzero tensor with any finite number of parties overF2, two-dimensional local carriers, and at most two complete binary settings per party has a global assignment. The same conclusion holds over a finite commutative local ring with residue fieldF2. Proof. Two distinct binary bases share a rowa and have other rowsb and a +b. If they coincide, use b as the other row in both settings. Expand the tensor in the coordinates dual to(a,b ) at each site. Choose an inclusion-minimal setS of sites using theb coordinate whose coefficient is nonzero. Outside S select a. InsideS select the nonshared row of the chosen setting. Every joint contraction is the coefficient forS plus coefficients of proper subsets, all of which vanish. Thus every selected joint event is supported. For a local ring with radicalJ, choose the last nonzero radical layer containing all tensor coefficients: the tensor lies in(J a)2n but not in(J a+1)2n . Choose anF2-linear functional onJ a/J a+1 for which the coefficient tensor is nonzero. Reduce each local basis moduloJ and apply the field argument. Every selected original contraction has nonzero image under this functional and is therefore nonzero. This does not exclude Hardy logical contextuality, which concerns failure to extend a particular event. Nor does it cover three settings or a different residue field. For two parties, the exclusion of a strongly contextual binary two-setting table is close to the established PR-box and modal-table boundaries [1, 10]; the proof above states explicitly what extends to arbitrary party number and radical layers. 6 Finite verification and reproducibility The proofs establish the general statements. The standalone supplement contains a separate checker developed for the September 2026 audit and rerun for this revision. It uses only the Python standard library. It is not the earlier historical runner, whose dependencies are absent from the original article package. The current evidence claims refer only to the files and commands in this supplement. 9 PDF PAGE 10 The support checker enumerates primitive rays, complete bases and exact contraction tables. Forbidden outcome pairs become binary clauses; implication-graph reachability decides global assignments and whether each supported event extends. Smith data are computed afterward for comparison, not used to decide support type. Identical support tables are cached. For the binary field these decisions are also checked by direct assignment enumeration. Check Coverage and limitation Direct support All 22,170 nonzero 2× 2 resources over F2, F3, F4, F2[u]/(u2), F2[u]/(u3), F3[u]/(u2), and Z/4, Z/8, Z/9; zero disagreements. Length-four support Fourteen nonzero Smith representatives and 24 deterministic sam- pled resources overA = F2[u]/(u4), each using all 192 local bases; not all 65,535 nonzero resources. Hardy and analyzers 36 ring/exponent Hardy cases; 28 analyzer configurations, checking each reshape and the complete analysis matrix, plus correction identities for one invertible resource per configuration. Coding 52 explicit encoder/decoder certificates across seven rings, includ- ing all fourteen nonzeroA Smith representatives. These certify achievability; the general upper bound is analytic. Binary coding optimum All 15 nonzero2× 2 binary resources, all20,160 invertible joint decoders per resource, and all disjoint-support subsets of each decoded orbit. Two-setting boundary All 255 nonzero three-party binary tensors for a representative pair of distinct settings per party; the arbitrary-party and ring statements are analytic. (a,b ) Count r h Binary rank Messages Support Teleport (0, 0) 24,576 2 2 8 4 Strong Yes (0, 1) 18,432 2 1 7 2 Logical No (0, 2) 9,216 2 1 6 2 Logical No (0, 3) 4,608 2 1 5 2 Logical No (0, 4) 4,608 1 1 4 2 Local No (1, 1) 1,536 2 2 6 4 Strong No (1, 2) 1,152 2 1 5 2 Logical No (1, 3) 576 2 1 4 2 Logical No (1, 4) 576 1 1 3 2 Local No (2, 2) 96 2 2 4 4 Strong No (2, 3) 72 2 1 3 2 Logical No (2, 4) 72 1 1 2 2 Local No (3, 3) 6 2 2 2 4 Strong No (3, 4) 9 1 1 1 2 Local No (4, 4) 1 0 0 0 n/a Excluded n/a Table 4: All fifteen Smith classes overA = F2[u]/(u4) for 2× 2 matrices. Exponent 4 denotes zero. Counts are exhaustively enumerated; the remaining columns apply the analytic formulas. Logical means logical but not strong. Zero is excluded from the resource tasks. Table 4 uses a separate exhaustive Smith inventory of all65,536 matrices overA, including zero. Its support labels are assigned by Theorem 3.3; they are not direct support decisions on the entire population. The nonzero counts are5,265 local, 34,056 logical but not strong, and26,214 strong. The direct dual-number check yields81, 72 and 102 respectively for its 255 nonzero matrices. 10 PDF PAGE 11 A separately supplied process-semantics implementation is included unchanged for an additional, scoped cross-check: its Smith types and independently computed regular-representation binary ranks are checked on all65,536 matrices overA. Its inventory agrees with the audit checker. This comparison does not constitute another full support classification or a rerun of every historical process-semantics suite. From the extracted revision directory, runpython supplement/run_reproduction.py . The wrapper refuses optimized Python execution, runs the full stated population, compares substantive outputs with the supplied reference results, regenerates Table 4, and writes commands, inter- preter/platform, exit statuses, runtimes and SHA-256 hashes. Runtime and platform fields are excluded from substantive-output comparisons. The Boolean coefficient-support identity is analytic; no historical LM-product or missing canonical-runner result is claimed as reproduced here. These finite checks corroborate the proofs and do not certify historical priority. 7 Discussion The resource hierarchy is controlled by different questions about the same Smith form. At least two nonzero Smith factors permit a logical obstruction; leading multiplicity controls the absence of every global assignment and the size of an exact orbit code; invertibility controls universal raw-vector transfer. None of these statements identifies a physical capability without the relevant operation and readout assumptions. There are two particularly useful limits. First, the nilpotent maximal-code example is a static nonprimitive resource, so it cannot by itself justify a closed process theory. For primitive resources, maximal orbit coding and universal raw teleportation coincide by Corollary 4.6. Second, the codebook/contextuality equivalence compares matched complete-basis contracts. Restricting gates, changing the scalar tensor, adding discard-and-reprepare, or weakening exactness requires a new theorem. The next mathematical questions are a classification beyond chain rings, a process semantics closed under both preparation and conditioning, and resource comparisons for restricted measurement families. These remain open here. The contribution is the classification and the explicit comparison of resource criteria. The primary-source comparison is bounded; it does not certify historical firstness for the specific classification or orbit formula. References [1] Samson Abramsky and Adam Brandenburger. The sheaf-theoretic structure of non-locality and contextuality. New Journal of Physics, 13:113036, 2011. doi:10.1088/1367-2630/13/11/113036; arXiv:1102.0264. [2] Michael Ben-Zvi, Alexander Ma, and Manuel Reyes. A Kochen–Specker theorem for integer matrices and noncommutative spectrum functors.Journal of Algebra, 491:280–313, 2017. With an appendix by Alexandru Chirvasitu. doi:10.1016/j.jalgebra.2017.08.008; arXiv:1509.03618. [3] Eimear Byrne, Anna-Lena Horlemann, Karan Khathuria, and Violetta Weger. Density of free modules over finite chain rings. Linear Algebra and its Applications, 651:1–25, 2022. doi:10.1016/j.laa.2022.06.013; arXiv:2106.09403. [4] Niel de Beaudrap. On computation with ‘probabilities’ modulo k. arXiv:1405.7381, 2014. 11 PDF PAGE 12 [5] Chen Feng, Roberto W. Nóbrega, Frank R. Kschischang, and Danilo Silva. Communication over finite-chain-ring matrix channels.IEEE Transactions on Information Theory, 60:5899–5917, 2014. doi:10.1109/TIT.2014.2346079; arXiv:1304.2523. [6] Melvin Henriksen. Two classes of rings generated by their units.Journal of Algebra, 31(1):182– 193, 1974. doi:10.1016/0021-8693(74)90013-1. [7] Thomas Honold and Ivan Landjev. Linear codes over finite chain rings.Electronic Journal of Combinatorics, 7(1):R11, 2000. doi:10.37236/1489. Volume year 2000; published online December 1999. [8] Benjamin Schumacher and Michael D. Westmoreland. Non-contextuality and free will in modal quantum theory. arXiv:1010.5452, 2010. [9] Benjamin Schumacher and Michael D. Westmoreland. Modal quantum theory.Foundations of Physics, 42(7):918–925, 2012. doi:10.1007/s10701-012-9650-z; arXiv:1010.2929. [10] Benjamin Schumacher and Michael D. Westmoreland. Almost quantum theory. In Giulio Chiribella and Robert W. Spekkens, editors,Quantum Theory: Informational Foundations and Foils, volume 181 ofFundamental Theories of Physics, pages 45–81. Springer, Dordrecht, 2016. doi:10.1007/978-94-017-7303-4_3; arXiv:1204.0701. [11] Matthew Varughese. Categorical quantum computing with finite fields. Master’s thesis, University of Oxford, 2009. Examples 8.3–8.4. https://www.cs.ox.ac.uk/people/aleks. kissinger/theses/bob/Varughese.pdf. [12] Daniel Zelinsky. Every linear transformation is a sum of nonsingular ones.Proceedings of the American Mathematical Society, 5(4):627–630, 1954. doi:10.1090/S0002-9939-1954-0062728-7. 12