> Full-text translation from Chinese. The original remains authoritative for its research claims; this translation is not independent mathematical verification. # NIOT-05 — Numerical Index Geometry: Addressing, Reindexing, Local Neighborhoods, and Identity Preservation ## Numerical Index Geometry: Addressing, Reindexing, Local Neighborhoods, and Identity Preservation **Series:** Numerical Information Object Theory (NIOT) / Numerical Information Object Theory **Paper:** Paper 05 **Author:** Neo.K **Institution:** EveMissLab / 一言諾科技有限公司 **Version:** v0.1 **Date:** 2026-09-22 **Document type:** Index geometry / address geometry / local information spaces paper **Prerequisites:** NIOT-00, NIOT-01, NIOT-02, NIOT-03, NIOT-04 **Related theories:** Identity Calculus, Active Index Calculus (AIC), ISQL, UEFM, BCODR, CAIR **Status:** DRAFT / INDEX-GEOMETRY FOUNDATION / RUNTIME-READY **Canonical Source:** UTF-8 Markdown **Mathematical source conventions:** Inline math uses only `$...$`, and display math uses only `$$...$$` --- # Abstract NIOT-01 establishes numerical information objects and identity units; NIOT-02 elevates information probes to composable and verifiable observation operators; NIOT-03 establishes Unboundedly Expandable Finite Matrices, keeping information dimensions finite at every epoch while retaining unbounded extensibility overall; NIOT-04 further introduces spectral discrimination, multiple scales, and multi-probe coupling. At this point, a numerical information object is no longer a scalar, but a dynamic local world with probes, matrices, spectra, and coupling fields. Yet a crucial layer is still missing: **How is this information actually located, navigated, sliced, reindexed, and reorganized?** This paper proposes **Numerical Index Geometry** (NIG), unifying probe indices, spectral coordinates, matrix coordinates, ISQL semantic addresses, AIC active indices, and UEFM dynamic coordinates in an operable address–index space. The central idea is: $$ \boxed{ \text{address} \neq \text{value} } $$ and: $$ \boxed{ \text{reindexing} \neq \text{identity change}. } $$ For a numerical information object $\mathfrak N$, define its index space: $$ \boxed{ \mathcal I(\mathfrak N) } $$ Each index $i\in\mathcal I(\mathfrak N)$ corresponds to an observable, operable, and traceable position. These positions may be: - probe coordinates; - matrix coordinates; - spectral coordinates; - provenance nodes; - derivation nodes; - certificate locations; - representation components; - bridge endpoints; - local neighborhoods; - semantic anchors. This paper further defines address graphs, local neighborhoods, index paths, index distances, index transport, reindexing, index fibers, address overlaps, local charts, atlases, gauge-like reparameterizations, index holonomy, and reindex invariants. Numerical Index Geometry does not force information spaces to be Euclidean; it permits different numerical information objects to use graph, metric, topological, spectral, or hybrid geometries. This paper specifically builds on the core of Identity Calculus and Active Index Calculus: $$ d_I(\mathfrak N) = \{ (\mathfrak N,i):i\in I \}, $$ and: $$ \widehat{\mathcal E}_{i\to j} : (\mathfrak N,i) \mapsto (\mathfrak N,j). $$ In NIOT, these operations acquire more general semantics: an index is not merely a slice label, but a navigable coordinate in the microscopic numerical world. Whenever an operation is identity-preserving reindexing: $$ \boxed{ \iota(\mathfrak N,i) = \iota(\mathfrak N,j) = \iota_{\mathfrak N}. } $$ Thus this paper establishes more than a numerical coordinate system: it establishes an **information-navigation geometry that preserves canonical identity**. **Keywords:** Numerical Index Geometry, Address Spaces, Reindexing, Active Index Calculus, ISQL, Index Fibers, Local Neighborhoods, Atlases, Charts, Holonomy, Numerical Information Objects --- # 1. The Problem: Information Already Exists, but Cannot Become a Computable Structure Unless It Can Be Located The first four NIOT papers have established: $$ \mathfrak N, $$ $$ \mathcal P, $$ $$ M_t, $$ $$ \mathcal S_{\mathfrak N}, $$ and: $$ C_{\mathfrak N}. $$ Yet these symbols still lack an answer to a common question: $$ \boxed{ \text{where is a piece of information?} } $$ If an information state has no address: - it cannot be queried locally; - it cannot be reindexed; - it cannot be expanded locally; - its dependencies cannot be traced; - it cannot be updated precisely; - it cannot be given provenance; - agents cannot navigate it. Therefore: $$ \boxed{ \text{distinguishable} \not\Rightarrow \text{addressable}. } $$ --- # 2. Separation of Address and Value ## 2.1 Value A value is: $$ v\in V. $$ For example: $$ 0.5. $$ --- ## 2.2 Address An address is: $$ a\in\mathcal A. $$ For example: ```text niot://object/123/probe/precision ``` or: $$ a=(i,j,\lambda,t). $$ --- ## 2.3 Fundamental Separation $$ \boxed{ a \neq v. } $$ Two different addresses may have the same value: $$ v(a_1)=v(a_2). $$ The same address may also have different values across epochs: $$ v_t(a) \neq v_{t+1}(a). $$ --- # 3. Index Space For a numerical information object: $$ \mathfrak N, $$ Define: $$ \boxed{ \mathcal I(\mathfrak N) } $$ as its index space. Each: $$ i\in\mathcal I(\mathfrak N) $$ identifies an: - observable position; - semantic location; - dynamic coordinate; - structural role. --- # 4. An Index Need Not Be an Integer A traditional array index: $$ i\in\mathbb N. $$ A NIOT index may be a: - tuple; - string path; - graph node; - semantic key; - spectral coordinate; - structured address; - symbolic coordinate. Therefore: $$ \boxed{ \mathcal I \not\subseteq \mathbb N } $$ does not hold in general. --- # 5. Typed Index Define: $$ \boxed{ i = (\tau_i,\alpha_i), } $$ where: - $\tau_i$: index type; - $\alpha_i$: index payload. For example: $$ \tau_i = \text{SPECTRAL}, $$ $$ \alpha_i = \lambda=3.2. $$ --- # 6. Index Classes The first version may distinguish the following index types: ```text VALUE REPRESENTATION PROBE MATRIX SPECTRAL PROVENANCE HISTORY RELATION CERTIFICATE BRIDGE TASK LOCAL_REGION ``` --- # 7. Indexed View For any: $$ i\in\mathcal I(\mathfrak N), $$ Define: $$ \boxed{ (\mathfrak N,i) } $$ as an indexed view. It is not a new canonical numerical object. --- # 8. Identity Preservation If: $$ i,j $$ are merely different admissible addresses of the same numerical object, then: $$ \boxed{ \iota(\mathfrak N,i) = \iota(\mathfrak N,j) = \iota_{\mathfrak N}. } $$ --- # 9. Interface with Identity Calculus Identity Calculus gives: $$ d_I(\mathfrak N) = \{ (\mathfrak N,i) : i\in I \}. $$ NIOT interprets this as: $$ \boxed{ \text{index expansion without object destruction}. } $$ --- # 10. Forgetful Integration $$ \boxed{ \int_{\mathrm{id}} (\mathfrak N,i) = \mathfrak N. } $$ This ensures that an index is an observation structure rather than canonical fragmentation. --- # 11. Active Index A passive index only marks a position. An active index allows: $$ \boxed{ \widehat{\mathcal E}_{i\to j} : (\mathfrak N,i) \mapsto (\mathfrak N,j). } $$ This is the minimal navigation operator. --- # 12. Address Graph Define: $$ \boxed{ G_A(\mathfrak N) = (V_A,E_A). } $$ where: $$ V_A = \mathcal I(\mathfrak N). $$ An edge denotes admissible navigation / a relation. --- # 13. Edge Types These may include: ```text contains depends_on derived_from bridges_to projects_to refines coarsens adjacent_to same_provenance same_probe_family same_spectrum ``` --- # 14. Address Path Define: $$ \boxed{ \gamma = (i_0,i_1,\dots,i_k). } $$ If every: $$ (i_j,i_{j+1}) \in E_A, $$ then $\gamma$ is an admissible address path. --- # 15. Path Length In the simplest case: $$ \boxed{ L(\gamma)=k. } $$ One may also have a weighted length: $$ \boxed{ L_w(\gamma) = \sum_j w(i_j,i_{j+1}). } $$ --- # 16. Index Distance Define: $$ \boxed{ d_I(i,j) = \inf_\gamma L(\gamma). } $$ If the graph is disconnected: $$ d_I(i,j)=\infty. $$ --- # 17. Distance Is Not Necessarily a Geometric Distance $d_I$ may denote: - semantic hops; - derivation hops; - dependency hops; - bridge costs; - probe transition costs. --- # 18. Local Neighborhood For: $$ i\in\mathcal I, $$ Define: $$ \boxed{ B_r(i) = \{ j: d_I(i,j)\le r \}. } $$ This is an index ball. --- # 19. Local Numerical World For: $$ i, $$ Define: $$ \boxed{ \mathcal W_r(\mathfrak N,i) = \mathfrak N[B_r(i)]. } $$ This is the local numerical information world centered at $i$. --- # 20. Locality Principle If a probe acts only on: $$ B_r(i), $$ one should not automatically materialize: $$ \mathcal I(\mathfrak N) $$ in its entirety. --- # 21. Dynamic Neighborhood As the epoch: $$ t, $$ changes, the index graph may change. Therefore: $$ \boxed{ B_r^{(t)}(i) } $$ is also dynamic. --- # 22. Address Topology If a family of open sets is defined: $$ \mathcal O_A, $$ then: $$ \boxed{ (\mathcal I,\mathcal O_A) } $$ forms an address topology. --- # 23. Topology Need Not Come from a Metric One can use: - dependencies; - semantic containment; - probe reachability; - spectral adjacency; to induce a topology directly. --- # 24. Basis of Open Sets Local neighborhoods: $$ \{B_r(i)\} $$ may be candidates for a topological basis. --- # 25. Index Continuity If a map: $$ f: \mathcal I_1 \rightarrow \mathcal I_2 $$ preserves the open structure, it is called a continuous reindex map. --- # 26. Reindexing Define: $$ \boxed{ \rho: \mathcal I \rightarrow \mathcal I'. } $$ and: $$ \boxed{ \rho_\ast : M \rightarrow M'. } $$ --- # 27. Pure Reindexing If: $$ \rho $$ changes only addresses, not semantic information, it is called: $$ \boxed{ \text{pure reindexing}. } $$ --- # 28. Identity-Preserving Reindexing Require: $$ \boxed{ \iota ( \rho_\ast(\mathfrak N) ) = \iota(\mathfrak N). } $$ --- # 29. Semantic Reindexing If: $$ \rho $$ changes semantic grouping as well as addresses, an additional certificate is needed. --- # 30. Reindex Certificate It includes at least: - the source index space; - the target index space; - the mapping; - preserved invariants; - lost distinctions; - reversibility. --- # 31. Bijective Reindexing If: $$ \rho $$ is a bijection and the payload is unchanged, it can usually be regarded as lossless. --- # 32. Non-Bijective Reindexing If: $$ \rho $$ is not a bijection, it may involve: - merging; - splitting; - quotients; - expansion. --- # 33. Split as Reindexing $$ \boxed{ i \mapsto \{i_1,\dots,i_k\}. } $$ This is one-to-many reindexing. --- # 34. Merge as Reindexing $$ \boxed{ \{i_1,\dots,i_k\} \mapsto \hat i. } $$ This is many-to-one reindexing. --- # 35. Quotient Index If: $$ i\sim j, $$ then: $$ \boxed{ [i] } $$ may serve as a quotient index. --- # 36. Quotient Address Space $$ \boxed{ \mathcal I/\sim. } $$ --- # 37. Fiber of Reindexing For: $$ j\in\mathcal I', $$ Define: $$ \boxed{ F^\rho_j = \rho^{-1}(j). } $$ --- # 38. Nontrivial Fiber If: $$ |F^\rho_j|>1, $$ multiple original addresses have been compressed into one target address. --- # 39. Reindex Ambiguity If one cannot, from: $$ j $$ uniquely recover the original: $$ i, $$ then: $$ \boxed{ \rho^{-1} } $$ is not a function. --- # 40. Address Loss Thus merging / compression must declare: $$ \boxed{ \text{address loss}. } $$ --- # 41. Chart For a local region: $$ U\subseteq\mathcal I, $$ define a chart: $$ \boxed{ \phi_U: U \rightarrow \mathbb K^n } $$ Or, more generally: $$ \phi_U: U \rightarrow C. $$ --- # 42. A Chart Is a Convenient Representation, Not the Ontology $$ \boxed{ \phi_U(U) \neq U } $$ holds in general. --- # 43. Local Coordinate In a chart: $$ \boxed{ x_U = \phi_U(i). } $$ --- # 44. Multiple Charts An index space may require: $$ \boxed{ \{(U_\alpha,\phi_\alpha)\} } $$ multiple charts. --- # 45. Atlas Define: $$ \boxed{ \mathcal A = \{ (U_\alpha,\phi_\alpha) \}. } $$ --- # 46. Transition Map If: $$ U_\alpha\cap U_\beta\neq\varnothing, $$ Define: $$ \boxed{ \phi_{\beta\alpha} = \phi_\beta \circ \phi_\alpha^{-1}. } $$ --- # 47. Transition Consistency If transition maps are inconsistent, this may indicate: - a chart mismatch; - a schema mismatch; - an identity mismatch; - an invalid bridge. --- # 48. Atlas Incompleteness Some index regions may have no chart. They should be marked: ```text ATLAS_MISSING ``` They cannot be treated as nonexistent. --- # 49. BCODR's Atlas Lesson BCODR-LUC has shown that: $$ \boxed{ \text{global symmetry} \not\Rightarrow \text{same local chart}. } $$ In NIOT, this generalizes to: $$ \boxed{ \text{global information equivalence} \not\Rightarrow \text{same local index coordinate}. } $$ --- # 50. Gauge-Like Reparameterization If: $$ g: \mathcal I \rightarrow \mathcal I' $$ changes only the coordinate convention, not the canonical structure, it can be regarded as an: $$ \boxed{ \text{index gauge transform}. } $$ --- # 51. Gauge Orbit Multiple indexings of the same information structure: $$ \boxed{ \mathcal O_g(\mathfrak N) = \{ g\cdot\mathfrak N \}. } $$ --- # 52. Gauge-Invariant Observable If: $$ O(g\cdot\mathfrak N) = O(\mathfrak N), $$ then: $$ \boxed{ O } $$ is gauge-invariant. --- # 53. Gauge Fixing The Runtime may choose: $$ \boxed{ g^\star } $$ as the canonical index convention. --- # 54. Gauge Fixing Does Not Mean Ontological Uniqueness It merely chooses a more convenient representation. --- # 55. Spectral Coordinate as Index NIOT-04's: $$ \lambda $$ can directly become an index: $$ \boxed{ i_\lambda = (\text{SPECTRAL},\lambda). } $$ --- # 56. Probe Coordinate as Index $$ \boxed{ i_P = (\text{PROBE},P). } $$ --- # 57. Matrix Coordinate as Index $$ \boxed{ i_{ab} = (\text{MATRIX},a,b). } $$ --- # 58. Composite Index One can use: $$ \boxed{ i = (P,\lambda,t,a,b). } $$ to represent a highly precise information position. --- # 59. Hierarchical Address One can use: ```text /object /probe /spectral /lambda=3.2 /epoch=17 ``` --- # 60. Path Address This can connect to ISQL / filesystem / graph paths. --- # 61. Address Normalization Different expressions: ```text /probe/a/../b ``` and: ```text /probe/b ``` may point to the same address. One needs: $$ \boxed{ N_A(a) } $$ for canonical normalization. --- # 62. Address Equality Define: $$ \boxed{ a_1\equiv_A a_2 } $$ if they are the same after normalization. --- # 63. Address Equality Is Not Object Equality Equality of two addresses indicates only the same location. --- # 64. Object Identity vs Address Identity $$ \boxed{ \iota_O \neq \iota_A. } $$ --- # 65. Moving Address, Stable Object The same object may move from: $$ a_t \rightarrow a_{t+1} $$ However: $$ \iota_O $$ remains unchanged. --- # 66. Stable Address, Changing Content The content at the same address may also change over time. --- # 67. Temporal Index Define: $$ \boxed{ i_t = (\text{TIME},t). } $$ --- # 68. Spatiotemporal Index One can combine: $$ \boxed{ i=(a,t). } $$ --- # 69. Version Index $$ \boxed{ i_v = (\text{VERSION},v). } $$ --- # 70. Historical Navigation One can use: $$ \widehat{\mathcal E}_{t_1\to t_2}. $$ --- # 71. Historical Identity If this is merely version evolution of the same object: $$ \boxed{ \iota_{\mathfrak N} } $$ may persist. --- # 72. Fork Identity If a canonical fork produces: $$ \mathfrak N^{(A)}, \mathfrak N^{(B)}, $$ new branch identities are needed. --- # 73. Branch Address $$ \boxed{ i = (\text{BRANCH},b). } $$ --- # 74. Branch Merge Merging two branches requires: - reconciliation; - conflict resolution; - provenance preservation. --- # 75. Index Dependency If: $$ i_j $$ depends on: $$ i_i, $$ Write: $$ \boxed{ i_i \prec i_j. } $$ --- # 76. Partial Order If dependencies have no cycles, they may form a poset: $$ \boxed{ (\mathcal I,\preceq). } $$ --- # 77. Cycle If there are cycles, the structure cannot be forced into a DAG. A general directed graph is needed. --- # 78. Strongly Connected Component A cyclic region can be compressed into an SCC: $$ \boxed{ \operatorname{SCC}(\mathcal I). } $$ --- # 79. Condensation Graph Taking the SCC quotient gives a DAG. --- # 80. Index Depth For a DAG: $$ \boxed{ d(i) } $$ can denote derivation depth. --- # 81. Index Height One may also define the distance to a root / sink. --- # 82. Index Boundary For a region: $$ U, $$ Define: $$ \boxed{ \partial_I U } $$ as the indices with edge interactions with the exterior. --- # 83. Interior $$ \boxed{ \operatorname{Int}_I(U) = U\setminus\partial_I U. } $$ --- # 84. Boundary Probe A high-value probe may act only on: $$ \partial_I U. $$ --- # 85. Index Frontier If: $$ U $$ is the expanded region, the frontier can be defined as: $$ \boxed{ F_I = \partial_I U. } $$ --- # 86. Frontier Expansion In the next round, expand only: $$ F_I. $$ --- # 87. BCODR Responsibility Frontier BCODR can use: - outward reachability; - inward reachability; to turn their intersection into: $$ \boxed{ F_I^{\mathrm{resp}}. } $$ --- # 88. Local Responsibility Chart In: $$ F_I^{\mathrm{resp}} $$ construct a finer chart on it. --- # 89. Address Overlap Two index regions: $$ U,V $$ may satisfy: $$ U\cap V\neq\varnothing. $$ --- # 90. Overlap Is Not Identity Even if: $$ i\in U\cap V, $$ this does not mean: $$ U=V. $$ --- # 91. Overlap Transition A transition rule is needed: $$ \boxed{ \phi_{VU}. } $$ --- # 92. Multi-Chart Consistency In a triple overlap: $$ U\cap V\cap W, $$ ideally: $$ \boxed{ \phi_{WU} = \phi_{WV}\circ\phi_{VU}. } $$ --- # 93. Transition Defect If: $$ \phi_{WU} \neq \phi_{WV}\circ\phi_{VU}, $$ Define: $$ \boxed{ \Delta_{\mathrm{trans}}. } $$ --- # 94. Index Holonomy Along a closed path: $$ \gamma, $$ apply a sequence of reindexings / transitions: $$ \boxed{ \mathcal H_\gamma. } $$ --- # 95. Trivial Holonomy If: $$ \mathcal H_\gamma = \operatorname{id}, $$ the path closes without a defect. --- # 96. Nontrivial Holonomy If: $$ \mathcal H_\gamma \neq \operatorname{id}, $$ this may indicate: - path-dependent indexing; - an active state change; - incompatible charts; - genuine information curvature. --- # 97. Index Curvature This paper provisionally regards nontrivial holonomy as an: $$ \boxed{ \text{index curvature candidate}. } $$ --- # 98. Not Traditional Riemann Curvature They cannot be equated directly unless a smooth manifold and connection are additionally established. --- # 99. Index Connection A local transport rule can be defined: $$ \boxed{ \nabla_I. } $$ It describes how indices are transported along paths. --- # 100. Parallel Transport Analogy $$ \boxed{ T_\gamma(i) } $$ denotes transport of an index role along a path. --- # 101. Role Preservation If, after transport: $$ \operatorname{Role}(T_\gamma(i)) = \operatorname{Role}(i), $$ the transport is called role-preserving. --- # 102. Role Drift If the role changes, mark: $$ \boxed{ \text{role drift}. } $$ --- # 103. Index Metric Not every index space needs a metric. If one is needed, define: $$ \boxed{ d: \mathcal I\times\mathcal I \rightarrow [0,\infty]. } $$ --- # 104. Weighted Metric Candidate For example: $$ d(i,j) = \alpha d_{\mathrm{semantic}} + \beta d_{\mathrm{provenance}} + \gamma d_{\mathrm{spectral}}. $$ --- # 105. Task-Relative Metric $$ \boxed{ d_{\mathcal T} } $$ can have its weights adjusted to the task. --- # 106. A Metric Is Not Necessarily Unique Different tasks may use different distances. --- # 107. Index Geodesic If a metric is defined, the shortest navigation path is: $$ \boxed{ \gamma^\star = \arg\min_\gamma L(\gamma). } $$ --- # 108. Geodesic Query For example: > What is the shortest admissible path from a scalar-value address to its nearest provenance certificate? --- # 109. Probe Navigation AI can follow an index graph: $$ i_0 \rightarrow i_1 \rightarrow i_2 $$ to decide the next probe step by step. --- # 110. Navigation Policy Define: $$ \boxed{ \pi_{\mathrm{nav}} ( i_t, M_t, \mathcal T ) \rightarrow i_{t+1}. } $$ --- # 111. Local Search Search only within: $$ B_r(i) $$ for high-utility probes. --- # 112. Global Search If local closure fails, expand the radius. --- # 113. Radius Schedule $$ r_0c_{ij}, $$ index congestion occurs. --- # 158. Routing Optimization AI can use: - distance; - cost; - trust; - congestion; to select a path. --- # 159. Trust-Weighted Distance $$ \boxed{ d_T(i,j) = \min_\gamma \sum_{e\in\gamma} w_T(e). } $$ --- # 160. Certificate-Weighted Distance A low-trust edge may have a higher cost. --- # 161. Secure Navigation Some addresses are accessible only with specific permissions. --- # 162. Permissioned Index Space Define: $$ \boxed{ \mathcal I_u \subseteq \mathcal I } $$ as the subspace visible to the user / agent. --- # 163. Same Object, Different Visible Geometry Different agents: $$ u,v $$ may see: $$ \mathcal I_u \neq \mathcal I_v. $$ --- # 164. Visibility Projection $$ \boxed{ \Pi_u(\mathcal I) = \mathcal I_u. } $$ --- # 165. Security Projection Non-Identity The visible index space is not the complete canonical index space. --- # 166. Hidden Does Not Mean Absent If an index is invisible to an agent: $$ \boxed{ \text{WITHHELD} } $$ does not mean: $$ \boxed{ \text{NONEXISTENT}. } $$ --- # 167. Access Certificate Navigating to a sensitive index requires a permission certificate. --- # 168. Index Mutation If an active probe changes the index graph: $$ \boxed{ \mathcal I_t \rightarrow \mathcal I_{t+1}. } $$ --- # 169. Mutation Log One must record: - added indices; - removed indices; - remapped indices; - the reason; - the certificate. --- # 170. Index Diff Define: $$ \boxed{ \Delta\mathcal I_t = \mathcal I_{t+1} \ominus \mathcal I_t. } $$ --- # 171. Structural Diff Consider not only additions and removals, but also: - edge changes; - role changes; - chart changes; - gauge changes. --- # 172. Reindex Diff $$ \boxed{ \Delta_\rho } $$ specifically records coordinate remapping. --- # 173. Canonical Index Snapshot At each important epoch, one can save: $$ \boxed{ S_I^{(t)}. } $$ --- # 174. Index Hash If serialization is deterministic: $$ \boxed{ h_I^{(t)} = H(S_I^{(t)}). } $$ --- # 175. Cross-Runtime Index Sync Two runtimes can compare: $$ h_I^{(A)} $$ and: $$ h_I^{(B)}. $$ --- # 176. Same Hash, Same Serialized Index State This does not mean that all external world states are identical, only that the canonical serialized index states are identical. --- # 177. Index Replay From the mutation log: $$ \boxed{ \mathcal I_0 \rightarrow \mathcal I_1 \rightarrow \cdots } $$ it should be possible to replay. --- # 178. Index Rollback For erroneous reindexing: $$ \mathcal I_{t+1} \rightarrow \mathcal I_t. $$ --- # 179. Rollback May Have Side Effects Compensation is needed if an address in an external system has already been updated. --- # 180. Index Migration If the schema is revised: $$ \boxed{ \mu: \mathcal I^{(v1)} \rightarrow \mathcal I^{(v2)}. } $$ --- # 181. Migration Certificate One needs: - the mapping; - lost addresses; - new addresses; - ambiguity; - compatibility. --- # 182. Index Compatibility If an old query can execute equivalently in the new index space, this is called backward compatibility. --- # 183. Index Deprecation An old index can be marked: ```text DEPRECATED ``` rather than simply disappearing. --- # 184. Alias One can retain: $$ \boxed{ i_{\mathrm{old}} \mapsto i_{\mathrm{new}}. } $$ --- # 185. Alias Chain Long chains should be normalized. --- # 186. Alias Loop If: $$ i_1\to i_2\to i_1, $$ this must be detected. --- # 187. Canonical Address An identity may have multiple aliases, but only one canonical address: $$ \boxed{ a^\star. } $$ --- # 188. A Canonical Address Is Not Object Identity It is still only the current canonical locator. --- # 189. Index Ambiguity If a surface address resolves to multiple indices: $$ \boxed{ a \mapsto \{i_1,\dots,i_k\}. } $$ --- # 190. Ambiguity Resolution One needs: - a schema; - context; - a namespace; - a version. --- # 191. ICNS Analogy ICNS resolves numeral ambiguity. NIG resolves address / index ambiguity. --- # 192. Namespace Define: $$ \boxed{ \mathcal N_S. } $$ Different index types should have namespaces. --- # 193. Fully Qualified Address For example: ```text niot://object/123/spectral/scale/3.2 ``` --- # 194. Relative Address A shorter path can be used in a local chart. --- # 195. Relative-to-Absolute Resolution One needs: $$ \boxed{ R_{\mathrm{addr}}. } $$ --- # 196. Root Address Each numerical information object may have: $$ \boxed{ a_{\mathrm{root}}. } $$ --- # 197. Rooted Index Geometry The entire index space may be regarded as: $$ \boxed{ (\mathcal I,a_{\mathrm{root}}). } $$ --- # 198. Multiple Roots If the object itself has multiple canonical subdomains, multiple roots may be allowed. --- # 199. Rooted vs Unrooted Geometry Not every analysis needs a root. --- # 200. Local Index Basis In a region: $$ U, $$ one may choose: $$ \boxed{ \mathcal B_I(U). } $$ as a local coordinate basis. --- # 201. Basis Change $$ \boxed{ B: \mathcal B_I(U) \rightarrow \mathcal B_I'(U). } $$ --- # 202. Basis Invariant Task-critical structure should be preserved under admissible basis changes. --- # 203. Index Tensor If there are multiple axes: $$ i=(i_1,\dots,i_k), $$ they can form a tensor-product-like address space. --- # 204. Factorizable Index Space If: $$ \mathcal I = \mathcal I_1\times\cdots\times\mathcal I_k, $$ navigation can proceed axis by axis. --- # 205. Non-Factorizable Index Space If relations prevent the axes from being separated, joint indexing is needed. --- # 206. Coupled Index Geometry NIOT-04's coupling: $$ C_{ij} $$ may change local distances / navigation. --- # 207. Coupling-Weighted Metric $$ \boxed{ d_C(i,j) = f(C_{ij}). } $$ --- # 208. Strong Coupling, Short Distance A common choice is: $$ d_C(i,j) = \frac{1}{\epsilon+|C_{ij}|}. $$ This is only an example. --- # 209. Coupling-Induced Neighborhood $$ \boxed{ B_r^C(i) = \{j:d_C(i,j)\le r\}. } $$ --- # 210. Spectral-Coupling Geometry A distance may depend simultaneously on: $$ \lambda $$ and: $$ C. $$ --- # 211. Dynamic Metric $$ \boxed{ d_t(i,j). } $$ which changes with the epoch. --- # 212. Geometry Drift If: $$ d_t \neq d_{t+1}, $$ this indicates index-geometry drift. --- # 213. Drift Attribution One must determine whether it results from: - a source change; - a probe change; - a coupling change; - a reindexing change. --- # 214. Index Geometry Closure For a task: $$ \mathcal T, $$ If: 1. critical addresses are resolved; 2. required paths are valid; 3. no unresolved aliases / ambiguities remain; 4. the local frontier is closed; 5. needed charts are available; then: $$ \boxed{ \operatorname{Close}^{I}_{\mathcal T}=1. } $$ --- # 215. Closure Certificate It includes: - the index version; - the root; - charts; - unresolved indices; - the alias map; - the frontier; - critical paths. --- # 216. Geometry Reopen A new probe / new task may reopen the index geometry. --- # 217. Finite Materialization At any epoch: $$ \boxed{ |\mathcal I_t|<\infty. } $$ --- # 218. Unbounded Address Expansion Allow: $$ \boxed{ \sup_t|\mathcal I_t|=\infty. } $$ --- # 219. Consistency of NIG and UEFM In NIG, UEFM's unbounded dimension corresponds to: $$ \boxed{ \text{unbounded index birth}. } $$ --- # 220. Consistency of NIG and Probe Calculus Probe discovery: $$ P_{\mathrm{new}} $$ may generate: $$ i_{\mathrm{new}}. $$ --- # 221. Consistency of NIG and Spectra Adding a spectral sample: $$ \lambda^\star $$ generates a spectral index: $$ i_{\lambda^\star}. $$ --- # 222. Consistency of NIG and BCODR A new responsibility-overlap region may generate local responsibility indices. --- # 223. Consistency of NIG and CAIR Reindexing / mutation should first be a candidate diff, then be committed. --- # 224. Consistency of NIG and ISQL ISQL can serve as the canonical semantic locator layer. --- # 225. A First Minimal Example Suppose: $$ \mathfrak N_x $$ has: ```text /value /provenance /precision ``` as its three indices. --- # 226. Adding a Probe A new provenance-depth probe produces: ```text /provenance/source /provenance/transform ``` Therefore: $$ \boxed{ \mathcal I_0 \rightarrow \mathcal I_1. } $$ --- # 227. Split The original: ```text /provenance ``` is split into two finer addresses. --- # 228. Identity Remains Unchanged $$ \iota_{\mathfrak N} $$ remains unchanged. --- # 229. A Second Example: Spectral Indices Adding: $$ \lambda\in\{1,2,4,8\}. $$ produces: ```text /spectral/1 /spectral/2 /spectral/4 /spectral/8 ``` --- # 230. Adaptive Refinement If: $$ 2<\lambda<4 $$ contains a transition, add: ```text /spectral/3 ``` --- # 231. A Third Example: Reindexing The old: ```text /quality/uncertainty ``` is revised to: ```text /uncertainty/model ``` If this is only a schema migration, the canonical object identity remains unchanged. --- # 232. A Fourth Example: Aliases Retain: ```text /quality/uncertainty ``` as an alias pointing to the new canonical address. --- # 233. A Fifth Example: The BCODR Frontier An outward probe and an inward certificate, near: ```text /branch/131 ``` form a responsibility overlap. Only in: ```text /branch/131/local ``` construct finer indices. --- # 234. Localized Expansion Thus global index explosion is unnecessary. --- # 235. Minimal Runtime Operations The first NIG Runtime version supports at least: ```text resolve navigate neighbors reindex split merge alias normalize snapshot rollback ``` --- # 236. Resolve $$ \boxed{ \operatorname{Resolve}(a) \rightarrow i. } $$ --- # 237. Navigate $$ \boxed{ \operatorname{Navigate}(i,j) \rightarrow \gamma. } $$ --- # 238. Neighbors $$ \boxed{ \operatorname{Neighbors}(i,r) \rightarrow B_r(i). } $$ --- # 239. Reindex $$ \boxed{ \operatorname{Reindex}(\rho). } $$ --- # 240. Split $$ \boxed{ \operatorname{Split}(i). } $$ --- # 241. Merge $$ \boxed{ \operatorname{Merge}(S). } $$ --- # 242. Alias $$ \boxed{ \operatorname{Alias}(a_{\mathrm{old}},a_{\mathrm{new}}). } $$ --- # 243. Normalize $$ \boxed{ \operatorname{NormalizeAddress}(a). } $$ --- # 244. Snapshot $$ \boxed{ \operatorname{SnapshotIndexSpace}(). } $$ --- # 245. Rollback $$ \boxed{ \operatorname{RollbackIndex}(t). } $$ --- # 246. Query Example ```text START: /value FIND: nearest certified provenance node CONSTRAINT: trust >= T2 ``` --- # 247. Navigation Result The Runtime should return: - the path; - total cost; - trust; - unresolved edges; - aliases used. --- # 248. Index Query Language A future interface with ISQL may use: ```text FROM /value FOLLOW derived_from UNTIL certificate >= T2 ``` --- # 249. Local Geometry Query ```text NEIGHBORHOOD /spectral/3.2 RADIUS 2 ``` --- # 250. Reindex Query ```text REINDEX /quality/uncertainty TO /uncertainty/model PRESERVE identity ``` --- # 251. Index Security Query ```text VISIBLE_TO agent:A ``` --- # 252. MVP Data Model ```json { "object_id": "niot:number:...", "index_version": 1, "root": "/", "nodes": [], "edges": [], "aliases": {}, "charts": [], "frontier": [], "history": [] } ``` --- # 253. Index Node Schema ```json { "index_id": "idx:...", "type": "SPECTRAL", "address": "/spectral/3.2", "role": "probe-coordinate", "state": "ACTIVE" } ``` --- # 254. Edge Schema ```json { "source": "idx:a", "target": "idx:b", "type": "derived_from", "weight": 1.0 } ``` --- # 255. Chart Schema ```json { "chart_id": "chart:local:1", "domain": ["idx:a", "idx:b"], "coordinate_system": "local-v1" } ``` --- # 256. Index Invariant A $$ \boxed{ \text{reindexing} \not\Rightarrow \text{identity change}. } $$ --- # 257. Index Invariant B $$ \boxed{ \text{missing address} \neq \text{nonexistent structure}. } $$ --- # 258. Index Invariant C $$ \boxed{ \text{alias equality} \neq \text{canonical object equality}. } $$ --- # 259. Index Invariant D $$ \boxed{ \text{local chart} \neq \text{global index space}. } $$ --- # 260. Index Invariant E $$ \boxed{ \text{coordinate overlap} \neq \text{mechanism identity}. } $$ --- # 261. Index Invariant F $$ \boxed{ \text{index transform} \text{ must be traceable}. } $$ --- # 262. Theorem 1 — Bijective Pure Reindexing Preserves Address Cardinality If: $$ \rho: \mathcal I \rightarrow \mathcal I' $$ is a bijection, then: $$ \boxed{ |\mathcal I| = |\mathcal I'|. } $$ --- # 263. Theorem 2 — Pure Reindexing Does Not Change Canonical Identity By definition: $$ \boxed{ \iota ( \rho_\ast(\mathfrak N) ) = \iota(\mathfrak N). } $$ --- # 264. Theorem 3 — Noninjective Reindexing Produces Nontrivial Fibers If: $$ \rho $$ is noninjective, there is: $$ j $$ such that: $$ \boxed{ |F^\rho_j|>1. } $$ --- # 265. Theorem 4 — Local Address Equality Is Weaker Than Global Structural Equality If: $$ \phi_U(i)=\phi_U(j), $$ this still does not imply: $$ i=j $$ unless the chart map is injective. --- # 266. Theorem 5 — Finite Epoch Geometry Can Remain Unboundedly Extensible If, at every epoch: $$ |\mathcal I_t|<\infty, $$ and index birth is allowed without an upper bound, then: $$ \boxed{ \sup_t|\mathcal I_t|=\infty } $$ is compatible with finite execution. --- # 267. Formalization Targets The first Lean / Coq stage: 1. typed indices; 2. indexed views; 3. pure reindexing; 4. identity preservation; 5. alias normalization; 6. quotient indices; 7. finite epochs / unbounded expansion; 8. address-graph reachability. --- # 268. Graph Library Integration Many properties can directly use: - finite graphs; - reachability; - paths; - quotients; - SCCs; - DAGs. --- # 269. Topology Formalization This can be deferred to the second stage. --- # 270. Holonomy Formalization This can be deferred further, initially retained as a theoretical direction. --- # 271. Open Problem: The Canonical Index Core Which indices must always be retained? For example: - value; - schema; - the provenance root; - the identity anchor? This remains undecided. --- # 272. Open Problem: A Minimal Address Basis Is there a: $$ \boxed{ \mathcal B_A } $$ from which all task-relevant addresses can be generated? --- # 273. Open Problem: Index Curvature What kind of path dependence suffices for a genuine definition of curvature? --- # 274. Open Problem: Address Complexity How does navigation complexity change as: - node count; - edge density; - hierarchy depth; vary? --- # 275. Open Problem: A Semantic Metric Can a task-stable semantic distance be established? --- # 276. Open Problem: An Index Gauge Group Which reindex transforms should be regarded as pure gauge? --- # 277. Open Problem: Cross-Object Index Alignment How can the index spaces of two different numerical objects: $$ \mathfrak N_1, \mathfrak N_2 $$ be aligned? --- # 278. Alignment Map One needs: $$ \boxed{ A: \mathcal I(\mathfrak N_1) \rightarrow \mathcal I(\mathfrak N_2). } $$ --- # 279. Partial Alignment A may be a partial map. --- # 280. Structural Alignment Comparing only address strings is insufficient. One must compare: - roles; - topology; - provenance; - probe families; - coupling profiles. --- # 281. Cross-Object Atlas Multiple numerical objects may share a higher-level atlas. --- # 282. Global Numerical Atlas In the future, one might construct: $$ \boxed{ \mathfrak A_{\mathrm{NIOT}}. } $$ to connect the local index spaces of different objects. --- # 283. But This Paper Does Not Claim That a Universal Atlas Already Exists This is only a direction for future work. --- # 284. Interface with the Next Paper NIOT-06 will enter: $$ \boxed{ \textbf{Floating-Point Projection, Completion, and Cross-Domain Bridging.} } $$ The central interface supplied by NIG is: $$ \boxed{ \text{Bridge endpoints between different representation domains should also have addresses.} } $$ --- # 285. Bridge Endpoint Index For: $$ B: X\rightsquigarrow Y, $$ Define: $$ \boxed{ i_X, \qquad i_Y. } $$ --- # 286. Bridge Path $$ \boxed{ i_X \xrightarrow{B} i_Y. } $$ --- # 287. Bridge Provenance A bridge itself may become an edge object. --- # 288. Cross-Domain Reindex Representation conversion may simultaneously involve: $$ \boxed{ \text{value bridge} + \text{index bridge}. } $$ --- # 289. Float Address A float representation is more than a value; it also has: - bit-field addresses; - exponent addresses; - significand addresses; - exception addresses; - rounding-context addresses. --- # 290. Exact Real Address An exact real may have: - a symbolic source; - a generator; - a completion certificate; - an approximation stream. --- # 291. Cross-Domain Atlas NIOT-06 will study how these representation charts can be bridged. --- # 292. The Most Important Central Statements of This Paper First: $$ \boxed{ \text{value} \neq \text{address}. } $$ Second: $$ \boxed{ \text{index} \neq \text{identity}. } $$ Third: $$ \boxed{ \text{reindex} \neq \text{replace object}. } $$ Fourth: $$ \boxed{ \text{local chart} \neq \text{global ontology}. } $$ Fifth: $$ \boxed{ \text{addressable information} = \text{distinguishable information} + \text{stable locator}. } $$ --- # 293. Conclusion The first four NIOT papers have completed the progression from: $$ x $$ to: $$ \mathfrak N $$ then to: $$ P_i(\mathfrak N), $$ $$ M_t, $$ $$ \mathcal S_{\mathfrak N}, $$ $$ C_{\mathfrak N}. $$ This paper adds the next indispensable layer: $$ \boxed{ \mathcal I(\mathfrak N). } $$ That is: $$ \boxed{ \text{the address–index geometry of numerical information objects.} } $$ Only once information can be addressed precisely can it truly support: - local probing; - local updates; - reindexing; - splitting; - merging; - navigation; - replay; - rollback; - bridging; - governance. Thus the microscopic numerical world is no longer merely rich in information. It begins to have: $$ \boxed{ \text{positions} + \text{neighborhoods} + \text{paths} + \text{boundaries} + \text{chart} + \text{atlas} + \text{transport}. } $$ Here, "Every number is one" acquires a stronger version: $$ \boxed{ \text{The same numerical information object may have many local addresses and views, but provided the operations are identity-preserving, these slices and navigation still belong to the same canonical identity.} } $$ Therefore: $$ \boxed{ \text{the microscopic numerical world} } $$ is no longer a metaphor. It now has a local geometric skeleton that can be engineered: $$ \boxed{ \mathfrak N \rightarrow \mathcal I(\mathfrak N) \rightarrow G_A \rightarrow B_r(i) \rightarrow \text{chart} \rightarrow \text{reindex} \rightarrow \text{navigation}. } $$ The next paper: $$ \boxed{ \textbf{NIOT-06 — Floating-Point Projection, Completion, and Cross-Domain Bridging.} } $$ will formally return to the question that originally prompted this series: $$ \boxed{ \text{If float is merely a projection, how can exact values, completions, rounding, intervals, representations, and cross-domain operators be interchanged admissibly within the same framework?} } $$ --- # Canonical Source Declaration This file is the UTF-8 Markdown canonical source for `NIOT-05-v0.1`. The only mathematical delimiters are `$...$` and `$$...$$`. This paper formally introduces: - numerical index spaces; - typed indices; - address / value separation; - indexed views; - active index transport; - address graphs; - index distances; - local neighborhoods; - address topology; - reindexing; - splitting / merging; - quotient indices; - reindex fibers; - charts; - atlases; - gauge-like reparameterizations; - index holonomy; - dynamic index geometry; - permissioned index spaces; - index mutation / migration / aliases. This paper does not claim that: - every index space is a smooth manifold; - every address graph has a unique metric; - index holonomy has already been identified with curvature in traditional geometry; - every reindexing is lossless; - all numerical objects share a universal atlas; - all address equality is canonical object equality. Subsequent NIOT-06 should use this paper's bridge endpoints, cross-domain addresses, and representation charts as the indexing foundation for cross-domain numerical conversion. ===END===