GRM Residual Geometry

Detailed Table of Contents

Agreed. The TOC needs an opening section that explains why Residual Geometry is worth constructing before it dives into the machinery.

Introduction — Why Residual Geometry Matters

Residual Geometry begins from a simple but consequential claim: failure is not merely the absence of success. A failed construction can contain structure.

In ordinary mathematical and computational practice, residuals are usually treated as secondary objects. They are measured, minimized, thresholded, averaged, or discarded once an algorithm converges. Their role is largely diagnostic: they tell us how far a candidate is from satisfying a pre-existing model.

GRM reverses that relationship.

A residual is treated as structured output describing how an attempted construction fails, differs, or remains incomplete under an operator. The GRM discovery program explicitly defines Residual Geometry as the study of residuals as geometric objects capable of carrying support, interaction, symmetry, locality, transport, and higher structure.

The value of this shift is that it allows mathematics to learn from the structure of its own failures.

If a residual persists only on a particular carrier, that carrier may be mathematically significant.

If a residual vanishes on every proper subsystem but remains nonzero jointly, the failure contains irreducible higher-order interaction.

If a residual changes under transport between equivalent representations, the transport itself carries structure.

If two local residuals vanish while their composition produces a new defect, composition contains information that the local pieces do not.

If a residual cannot be represented in the current mathematical language, that may indicate not impossibility, but insufficiency of the current grammar.

Residual Geometry is therefore valuable because it converts what would normally be thrown away as error into a source of mathematical distinctions.

Its central question is not:

How small is the error?

It is:

What structure is the residual revealing that the current mathematical construction cannot yet express?

That difference changes the research program.

Instead of beginning with a fixed ontology and asking whether observations fit it, Residual Geometry allows persistent residual structure to pressure the ontology itself. The larger GRM program already establishes this direction through constraint-generated ontology: constraints generate distinctions, distinctions generate required operators, and operators can force new mathematical objects. Residual Geometry supplies a concrete mechanism for that process by making unresolved structure executable.

This gives the field five major sources of value.

First, it preserves information normally lost by simplification. A scalar error norm can collapse location, interaction structure, ancestry, symmetry, and carrier dependence into one number. Residual Geometry retains those distinctions before any compression occurs.

Second, it exposes higher-order structure. The program explicitly treats irreducible interaction as structure that cannot be reconstructed from proper lower-order subsystems. Residual analysis therefore provides a systematic way to detect mathematics hidden by unary or pairwise decomposition.

Third, it makes local-to-global failure mathematically productive. When locally valid constructions fail to reconstruct a global object, that failure is not automatically treated as noise. It can generate support, boundary, gluing, or interaction structure.

Fourth, it makes representation dependence visible. GRM separates source, representation, readout, and validation, and its Residual Geometry program treats transport and representation friction as mathematical questions rather than implementation details. A defect created by changing representation may itself be an invariant of the reconstruction process.

Fifth, it provides a route from failure to discovery. The GRM discovery spine is explicitly recursive:

failure → structured information → constructor pressure → new mathematics

and the common constructor sequence moves from source distinctions through requirements, operators, residuals, new requirements, synthesis, and language change.

This is the strongest reason to develop Residual Geometry.

A residual is not only evidence that a construction failed.

It may be evidence that:

the wrong carrier was chosen;
an interaction of higher arity has been suppressed;
a local model has exceeded its valid region;
an assumed factorization is not invariant;
a transport law is incomplete;
a composition rule is missing information;
an equivalence relation is too coarse;
a singular transition has been crossed;
an operator required by the source does not yet exist;
or the current ontology lacks an object type required to represent the distinction.

Residual Geometry turns these possibilities into explicit mathematical objects and executable tests.

The field therefore does not propose “geometry of errors” in the conventional sense. It proposes a geometry generated by structured mismatch itself.

The direction of construction is:

operator → execution → residual → carrier/support/interaction/transport → geometry

and, when the residual cannot be resolved inside the existing language:

residual → requirement → new operator → possible new object type → successor mathematics

This creates a fundamentally different discovery strategy.

Traditional approaches usually choose the mathematical language first and then measure how well reality or a source structure fits inside it.

Residual Geometry allows the persistent failures of that language to participate in constructing its successor.

The purpose of the theory developed in the following chapters is therefore to determine, rigorously and executably, when a residual is merely incidental and when it carries enough invariant structure to justify new mathematics.

GRM Residual Geometry

Updated Detailed Table of Contents

Introduction — Why Residual Geometry Matters

0.1 Failure as structured mathematical information
0.2 Why residuals should not be collapsed into error magnitudes
0.3 From residual diagnostics to residual mathematics
0.4 The failure of fixed-ontology modeling
0.5 Residuals as pressure on mathematical language
0.6 The streetlight problem in mathematical representation
0.7 Local validity versus global failure
0.8 Pairwise visibility versus higher-order structure
0.9 Representation equivalence versus transport triviality
0.10 Residuals as generators of operators
0.11 Residuals as generators of object types
0.12 Why geometry should be derived rather than assumed
0.13 The value of branch-honest mathematics
0.14 Open discovery with closed replay
0.15 Residual Geometry as a theory of mathematical self-extension

Part I — Foundations

1. Residual Geometry as Constructed Mathematics

1.1 Residual Geometry within GRM
1.2 Source mathematics versus constructor machinery
1.3 Source sovereignty
1.4 Domain neutrality
1.5 Residual-first discovery
1.6 Operator-relative semantics
1.7 Carrier-relative semantics
1.8 Local versus global authority
1.9 Pairwise versus higher-order authority
1.10 Information flow versus authority flow
1.11 Semantic canonicality versus serialization canonicality
1.12 Failure versus impossibility
1.13 Resource exhaustion versus mathematical falsehood
1.14 Open language frontiers
1.15 Global closure prohibition

2. The Minimal Residual Geometry Basis

2.1 Why the primitive basis must be minimized
2.2 Residual generation R
2.3 Restriction ρ
2.4 Residual comparison CMP
2.5 Residual transport τ
2.6 Residual composition μ
2.7 Typed domains and codomains
2.8 Operator-owned zero semantics
2.9 Partiality and UNKNOWN
2.10 Branch-valued operations
2.11 Ancestry preservation
2.12 Scope preservation
2.13 Residual extensions
2.14 Primitive independence
2.15 Deriving the remainder of Residual Geometry

3. Residual Objects and Events

3.1 Operator execution
3.2 Residual event records
3.3 Input, output, operator, residual, and carrier seed
3.4 Raw residuals
3.5 Normalized residuals
3.6 Residual extensions
3.7 Residual owner
3.8 Residual ancestry
3.9 Residual family
3.10 Live residuals
3.11 Persistent versus derived residual data
3.12 Residual identity
3.13 Residual typing
3.14 Residual scope
3.15 Residual replay identity

Part II — Residual Comparison and Equivalence

4. Residual Comparison

4.1 Comparison as a primitive operation
4.2 Comparison domains
4.3 Comparison codomains
4.4 ZERO, NONZERO, and UNKNOWN comparison states
4.5 Operator-relative zero
4.6 Comparison of raw residuals
4.7 Comparison of normalized residuals
4.8 Comparison of residual sets
4.9 Comparison under restriction
4.10 Comparison under transport
4.11 Comparison under composition
4.12 Comparison witnesses
4.13 Comparison incompleteness
4.14 Comparison-induced distinctions
4.15 Comparison as the basis of derived geometry

5. Epoch-Relative Residual Equivalence

5.1 Why timeless equivalence fails
5.2 Language-indexed equivalence
5.3 Scope-indexed equivalence
5.4 Epoch-indexed equivalence
5.5 EQE[r,r′,e,Γ,S]
5.6 Exact equality
5.7 Normalized equality
5.8 Behavioral equivalence
5.9 Transport equivalence
5.10 Generative equivalence
5.11 Ancestry equivalence
5.12 Disequivalence witnesses
5.13 Unknown equivalence
5.14 Branch preservation
5.15 Equivalence refinement after language growth
5.16 Equivalence-class splitting
5.17 Equivalence-class merging
5.18 Residual moduli by epoch

6. Semantic Branching

6.1 Residual branch spaces
6.2 Equivalence-generated merging
6.3 Disequivalence-generated separation
6.4 Unknown-generated separation
6.5 Branch persistence
6.6 Branch ancestry
6.7 Branch transport
6.8 Branch comparison
6.9 Branch composition
6.10 Branch refinement
6.11 Serialization representatives without semantic authority
6.12 Branch consequences
6.13 Branch-specific residuals
6.14 Branch collapse conditions
6.15 Branch honesty as a geometric constraint

Part III — Carrier Geometry

7. Provisional Residual Carriers

7.1 The carrier circularity problem
7.2 Carrier seeds
7.3 Carrier proposers
7.4 Carrier candidates
7.5 Carrier admissibility
7.6 Carrier obligations
7.7 Carrier type checks
7.8 Carrier regions
7.9 Restriction compatibility
7.10 Transport compatibility
7.11 Carrier complexity
7.12 Carrier branches
7.13 Carrier debt
7.14 Carrier mutation
7.15 Carrier candidates as provisional structures

8. Carrier Minimization

8.1 Why the final carrier must come after restriction and transport
8.2 Restriction-stable carriers
8.3 Transport-stable carriers
8.4 Consequence-preserving carrier comparison
8.5 Minimal carrier classes
8.6 Nonunique minima
8.7 Carrier antichains
8.8 Carrier equivalence
8.9 Carrier embeddings
8.10 Carrier retractions
8.11 Partial carrier relations
8.12 Carrier disjointness
8.13 Final residual carrier C★
8.14 Carrier singularities
8.15 Carrier-generated mathematical structure

Part IV — Restriction, Support, and Locality

9. Residual Restriction

9.1 Restriction as a primitive operation
9.2 Regions
9.3 Admissible regions
9.4 Region ancestry
9.5 Restriction bodies
9.6 Restriction validation
9.7 Nested restriction
9.8 Restriction composition
9.9 Restriction equivalence
9.10 Restriction under transport
9.11 Restriction under carrier mutation
9.12 Restriction failure
9.13 Restriction-generated residuals
9.14 Restriction coherence
9.15 Restriction as the basis of localization

10. Residual Support

10.1 Support as derived structure
10.2 Support from restriction and zero classification
10.3 Epoch-indexed support
10.4 Minimal support
10.5 Support antichains
10.6 Distributed support
10.7 Disconnected support
10.8 Boundary support
10.9 Interaction support
10.10 Joint-only support
10.11 Support transport
10.12 Support under carrier change
10.13 Support persistence
10.14 Support collapse
10.15 Support bifurcation
10.16 Support singularities

11. Residual Locality

11.1 Locality versus support
11.2 Local visibility
11.3 Local reconstructibility
11.4 Covers
11.5 Cover admissibility
11.6 Local residual families
11.7 Source-owned gluing
11.8 Reconstructed residuals
11.9 Locality comparison residual
11.10 Locality closure
11.11 Locality obstruction
11.12 Local-to-global failure
11.13 Higher-order overlaps
11.14 Coherence gaps
11.15 Locality-generated boundaries
11.16 Locality singularities

Part V — Residual Factorization

12. Residual Factorization

12.1 Factor candidates
12.2 Factorization constructors
12.3 Reconstruction through μ
12.4 Factorization comparison through CMP
12.5 Valid factorization
12.6 Irreducible residual factors
12.7 Nonunique factorization
12.8 Factorization branches
12.9 Factorization completeness
12.10 Incomplete factorization search
12.11 Factorization debt
12.12 Factorization transport
12.13 Factorization under restriction
12.14 Factorization under carrier mutation
12.15 Higher-order factorization failure
12.16 Factorization singularities

13. Factorization-Invariant Structure

13.1 Consequences of a factorization
13.2 Common consequences across factorizations
13.3 Factor core FCORE
13.4 Factorization-invariant support
13.5 Factorization-invariant interaction
13.6 Factorization-invariant carrier features
13.7 Factorization-dependent artifacts
13.8 Factorization-equivalence witnesses
13.9 Cross-factorization transport
13.10 Structural promotion criteria
13.11 Factorization ambiguity as information
13.12 Factorization-generated requirements

Part VI — Residual Transport Geometry

14. Residual Transport

14.1 Morphisms
14.2 Transport domains
14.3 Transport codomains
14.4 Operator lifting
14.5 Residual transport
14.6 Branch-valued transport
14.7 Transport ledgers
14.8 Preserved structure
14.9 Lost structure
14.10 Created structure
14.11 Exact transport
14.12 Lossy transport
14.13 Structure-generating transport
14.14 Transport ancestry
14.15 Transport obstruction
14.16 Transport incompleteness

15. Transport Coherence

15.1 Why holonomy must not be primitive
15.2 Morphism composition
15.3 Direct composite transport
15.4 Sequential transport
15.5 Transport-composition comparison
15.6 Transport coherence defect TC
15.7 Exact path coherence
15.8 Path-dependent transport
15.9 Transport coherence classes
15.10 Coherence failure
15.11 Transport associativity
15.12 Higher transport coherence
15.13 Coherence-generated residuals
15.14 Coherence singularities
15.15 Transport coherence as a primary invariant

16. Residual Paths and Holonomy

16.1 Residual paths
16.2 Path execution
16.3 Path composition
16.4 Closed paths
16.5 Return comparison
16.6 Residual holonomy
16.7 Trivial holonomy
16.8 Nontrivial holonomy
16.9 Branch-valued holonomy
16.10 Holonomy under path deformation
16.11 Holonomy classes
16.12 Local coherence versus global holonomy
16.13 Holonomy accumulation
16.14 Holonomy singularities
16.15 From holonomy toward curvature

17. Residual Symmetry

17.1 Symmetry as derived transport equivalence
17.2 Carrier automorphisms
17.3 Exact residual symmetry
17.4 Behavioral residual symmetry
17.5 Epoch-relative symmetry
17.6 Symmetry groups when closure exists
17.7 Symmetry groupoids when closure branches
17.8 Stabilizers
17.9 Orbits
17.10 Symmetry transport
17.11 Broken symmetry
17.12 Emergent symmetry
17.13 Symmetry transitions
17.14 Symmetry bifurcations
17.15 Symmetry singularities

Part VII — Residual Composition Geometry

18. Residual Composition

18.1 Composition as a primitive operation
18.2 Operator composition versus residual composition
18.3 Residual compositors
18.4 Composite operators
18.5 Direct composite residual
18.6 Reconstructed composite residual
18.7 Composition comparison
18.8 Nonadditivity
18.9 Order sensitivity
18.10 Composition closure
18.11 Composition failure
18.12 Composition under transport
18.13 Composition under restriction
18.14 Composition under factorization
18.15 Composition-generated structure

19. Composition Defects

19.1 Composition defect κ
19.2 Vanishing composition defect
19.3 Nonvanishing composition defect
19.4 Unknown composition defect
19.5 Composition defect support
19.6 Composition defect carrier
19.7 Composition defect transport
19.8 Composition defect persistence
19.9 Composition defect classes
19.10 Composition-generated requirements
19.11 Composition-generated operators
19.12 Composition singularities

20. Associator Residuals and Higher Coherence

20.1 Triple residual composition
20.2 Left-associated reconstruction
20.3 Right-associated reconstruction
20.4 Associator residual α
20.5 Associator zero
20.6 Associator nonzero
20.7 Associator unknown
20.8 Associator transport
20.9 Associator locality
20.10 Associator interaction
20.11 Higher associators
20.12 Coherence towers
20.13 Composition coherence classes
20.14 Higher-order composition singularities
20.15 Associator residuals as primary RG invariants

Part VIII — Higher-Order Interaction Geometry

21. Residual Interaction

21.1 Residual families
21.2 Joint residual construction
21.3 Proper-subfamily residuals
21.4 Lower-order residual closure
21.5 Explicit closure completeness
21.6 Irreducible interaction residual
21.7 ZERO, NONZERO, and UNKNOWN interaction
21.8 Interaction arity
21.9 Pairwise shadows
21.10 Joint-only structure
21.11 Higher-order support
21.12 Interaction carriers
21.13 Interaction locality
21.14 Interaction transport
21.15 Interaction composition
21.16 Interaction singularities

22. Arity and Higher-Order Closure

22.1 Minimal interaction arity
22.2 Arity escalation
22.3 Lower-arity impossibility witnesses
22.4 Higher-arity constructor availability
22.5 Arity completeness
22.6 Arity uncertainty
22.7 Arity under restriction
22.8 Arity under transport
22.9 Arity under factorization
22.10 Arity under language growth
22.11 Stable interaction arity
22.12 Arity transitions
22.13 Arity bifurcations
22.14 Arity singularities
22.15 Higher-order residual invariants

Part IX — Residual Complexity and Multiplicity

23. Residual Complexity Profiles

23.1 Why scalar rank was removed
23.2 Generator complexity
23.3 Carrier complexity
23.4 Interaction arity
23.5 Transport depth
23.6 Ancestry complexity
23.7 Complexity profiles K(r,e)
23.8 Unknown profile coordinates
23.9 Profile partial orders
23.10 Incomparable profiles
23.11 Profile transport
23.12 Profile refinement
23.13 Profile contraction
23.14 Profile transitions
23.15 Complexity as a discovery-state invariant

24. Residual Multiplicity

24.1 Behavioral residual classes
24.2 Generative distinction
24.3 Ancestry equivalence
24.4 Ancestry disequivalence
24.5 Generative equivalence GEQ
24.6 Generative graphs
24.7 Connected realization components
24.8 Multiplicity structure
24.9 Multiplicity versus cardinality
24.10 Multiplicity under transport
24.11 Multiplicity under composition
24.12 Multiplicity under language growth
24.13 Multiplicity transitions
24.14 Multiplicity bifurcations
24.15 Multiplicity singularities

Part X — Residual Signatures and Geometric Organization

25. Residual Signatures

25.1 Signature as derived structure
25.2 Zero-class coordinate
25.3 Carrier coordinate
25.4 Support coordinate
25.5 Factor coordinate
25.6 Locality coordinate
25.7 Symmetry coordinate
25.8 Complexity coordinate
25.9 Multiplicity coordinate
25.10 Interaction coordinate
25.11 Transport coordinate
25.12 Ancestry coordinate
25.13 Epoch-indexed signatures
25.14 Signature equivalence
25.15 Signature refinement

26. Residual Adjacency

26.1 Adjacency as minimal admissible mutation
26.2 Restriction adjacency
26.3 Transport adjacency
26.4 Composition adjacency
26.5 Repair adjacency
26.6 Behavioral nonidentity requirement
26.7 Directed adjacency
26.8 Undirected reachability
26.9 Residual components
26.10 Adjacency under language growth
26.11 Adjacency persistence
26.12 Adjacency collapse
26.13 Adjacency-generated topology
26.14 Residual neighborhoods
26.15 Discovery topology from residual adjacency

27. Residual Stratification

27.1 Stable-signature connectivity
27.2 Strata as connected residual components
27.3 Epoch-indexed strata
27.4 Stratum boundaries
27.5 Stratum adjacency
27.6 Stratum transport
27.7 Stratum refinement
27.8 Stratum splitting
27.9 Stratum merging
27.10 Stratum persistence
27.11 Stratum history SHIST
27.12 Witnessed stratum history
27.13 Stratification under successor-language growth
27.14 Stratification instability
27.15 Stratification-generated geometry

Part XI — Transition, Bifurcation, and Singularity

28. Residual Transitions

28.1 Signature-preserving continuation
28.2 Signature-changing continuation
28.3 Transition criterion
28.4 Carrier transitions
28.5 Support transitions
28.6 Factor transitions
28.7 Locality transitions
28.8 Symmetry transitions
28.9 Interaction transitions
28.10 Complexity transitions
28.11 Transport transitions
28.12 Composition transitions
28.13 Language transitions
28.14 Ontology transitions
28.15 Transition histories

29. Residual Bifurcations

29.1 Why bifurcation differs from transition
29.2 Continuation branches
29.3 Behavioral branch counting
29.4 Two-way bifurcation
29.5 Higher branching
29.6 Carrier bifurcation
29.7 Factorization bifurcation
29.8 Transport bifurcation
29.9 Interaction bifurcation
29.10 Repair bifurcation
29.11 Language bifurcation
29.12 Ontology bifurcation
29.13 Bifurcation persistence
29.14 Bifurcation resolution
29.15 Bifurcation invariants

30. Residual Singularities

30.1 Why signature change alone is not singularity
30.2 Continuation completeness
30.3 Singularity criterion
30.4 Carrier singularities
30.5 Support singularities
30.6 Locality singularities
30.7 Factorization singularities
30.8 Transport singularities
30.9 Composition singularities
30.10 Interaction singularities
30.11 Complexity singularities
30.12 Language singularities
30.13 Ontology singularities
30.14 Singularity transport
30.15 Singularity persistence
30.16 Singularities as constructor pressure

Part XII — Residual-Generated Mathematics

31. Residual Requirement Compilation

31.1 Residual signatures as requirement input
31.2 Requirement extraction
31.3 Minimal capability requirements
31.4 Requirement ownership
31.5 Requirement scope
31.6 Requirement ancestry
31.7 Requirement checks
31.8 Requirement equivalence
31.9 Requirement branches
31.10 Persistent residual pressure
31.11 Failed compilation
31.12 Requirement language gaps
31.13 Requirement refinement
31.14 Requirement closure
31.15 Requirement persistence across epochs

32. The Reduction Court

32.1 Why ontology generation needs an adversarial court
32.2 Existing type search
32.3 Existing operator search
32.4 Carrier substitution search
32.5 Factorization search
32.6 Representation search
32.7 Higher-arity search
32.8 Reduction basis
32.9 Resolved court state
32.10 Exhausted court state
32.11 Unknown court state
32.12 Exhaustiveness witnesses
32.13 Unknown forbids ontology promotion
32.14 Reduction court under language growth
32.15 Reduction court as anti-self-sealing machinery

33. Residual-Generated Operators

33.1 Candidate generation
33.2 Operator validation
33.3 Requirement satisfaction
33.4 Residual reevaluation
33.5 Complexity-profile comparison
33.6 Contraction
33.7 Refinement
33.8 Explicit new debt
33.9 No-relabel progress
33.10 Repair operators
33.11 Transport-generating operators
33.12 Interaction-resolving operators
33.13 Composition-repair operators
33.14 Singularity-resolving operators
33.15 Failed operator synthesis as residual evidence

34. Residual-Generated Ontology

34.1 Ontology pressure
34.2 Court exhaustion as prerequisite
34.3 Ontology synthesis
34.4 Candidate object types
34.5 Candidate relations
34.6 Candidate operator families
34.7 Candidate laws
34.8 Candidate carriers
34.9 Ontology validation
34.10 Executable consequence change
34.11 Existing-type equivalence rejection
34.12 Minimal ontology extension
34.13 Ontology branch spaces
34.14 Ontology deltas
34.15 Successor-language construction
34.16 Ontology non-self-sealing criterion

Part XIII — Residual Discovery Dynamics

35. The Residual Discovery State

35.1 Language state
35.2 Operator state
35.3 Live residual state
35.4 Ancestry state
35.5 Discovery epoch
35.6 Scheduler layer
35.7 Frontier
35.8 Debt
35.9 Resource state
35.10 Terminal state
35.11 Full discovery state
35.12 State identity
35.13 State equivalence
35.14 State evolution
35.15 Discovery orbit

36. Residual Repair Dynamics

36.1 Residual analysis
36.2 Signature generation
36.3 Requirement generation
36.4 Reduction-court execution
36.5 Existing repair
36.6 New-operator repair
36.7 Ontology repair
36.8 Frontier persistence
36.9 Epoch transition
36.10 Repair ancestry
36.11 Repair debt
36.12 Repair contraction
36.13 Repair refinement
36.14 Repair cycles
36.15 Stable residual recurrence

37. Discovery Orbits

37.1 Theory states as dynamical objects
37.2 Epoch sequences
37.3 Residual-language coevolution
37.4 Operator-language coevolution
37.5 Ontology-language coevolution
37.6 Orbit equivalence
37.7 Orbit branching
37.8 Orbit recurrence
37.9 Orbit stabilization
37.10 Stable nonclosure
37.11 Discovery attractors
37.12 Discovery cycles
37.13 Discovery bifurcations
37.14 Discovery singularities
37.15 Discovery-orbit invariants

Part XIV — Conservation, Loss, and Creation

38. Residual Transport Ledgers

38.1 Preserved residual structure
38.2 Lost residual structure
38.3 Created residual structure
38.4 Ledger ancestry
38.5 Ledger composition
38.6 Ledger comparison
38.7 Exact conservation
38.8 Controlled loss
38.9 Controlled creation
38.10 Unexplained loss
38.11 Unexplained creation
38.12 Conservation under path transport
38.13 Conservation under composition
38.14 Conservation under restriction
38.15 Conservation defects as invariants

39. Residual Conservation Laws

39.1 Conditions for residual conservation
39.2 Carrier-relative conservation
39.3 Support conservation
39.4 Interaction conservation
39.5 Factorization conservation
39.6 Symmetry conservation
39.7 Ancestry conservation
39.8 Conservation under language change
39.9 Conservation under ontology change
39.10 Split residual conservation
39.11 Fusion residual conservation
39.12 Residual transfer
39.13 Residual annihilation
39.14 Residual creation
39.15 Conservation-law failure as discovery signal

Part XV — Semantic Compression and Rehydration

40. Semantic Compression

40.1 Why textual compression is insufficient
40.2 Executable compression
40.3 Codecs
40.4 Encoding
40.5 Decoding
40.6 Exact semantic roundtrip
40.7 Semantic hashes
40.8 Compression ancestry
40.9 Compression failure
40.10 Compression residue
40.11 Unique expansion
40.12 Duplicate consequence elimination
40.13 No unnamed executable abstraction
40.14 Compression under language growth
40.15 Semantic compression as storage discipline

41. Replay

41.1 Deterministic replay
41.2 Schema verification
41.3 Ancestry verification
41.4 Closure-body verification
41.5 Expansion of compressed state
41.6 Language reconstruction
41.7 Residual reconstruction
41.8 Carrier reconstruction
41.9 Support and locality reconstruction
41.10 Equivalence reconstruction
41.11 Transport/coherence reconstruction
41.12 Interaction reconstruction
41.13 Composition reconstruction
41.14 Strata and history reconstruction
41.15 Reduction-court replay
41.16 Discovery-step replay
41.17 Replay invariants

42. Rehydration

42.1 Canonical serialization
42.2 Payload verification
42.3 State completeness
42.4 NO_GHOST
42.5 ZERO_DUP
42.6 Unique expansion
42.7 Replay before acceptance
42.8 Semantic reconstruction
42.9 Rehydration failure
42.10 Rehydration versus truth
42.11 Rehydration versus closure
42.12 Source authority after rehydration
42.13 Successor-state rehydration
42.14 Cold reconstruction
42.15 Rehydratable mathematics

Part XVI — Terminal Semantics and Adversarial Validation

43. Terminal Semantics

43.1 CONTINUE_T
43.2 SUCCESSOR_T
43.3 RESOURCE_T
43.4 CLOSED_T
43.5 KERNEL_T
43.6 Frontier persistence
43.7 Epoch debt discipline
43.8 Unresolved singularity states
43.9 No-early-stop discipline
43.10 Resource nonauthority
43.11 Local terminal versus global closure
43.12 Successor-language terminals
43.13 Replay-gated continuation
43.14 Terminal ancestry
43.15 Terminal misuse

44. Residual Geometry Adversarial Court

44.1 Premature residual projection
44.2 Universal-zero collapse
44.3 Premature carrier commitment
44.4 Representation-driven carrier selection
44.5 Local-to-global collapse
44.6 Pairwise-to-higher collapse
44.7 Unknown-equivalence collapse
44.8 Serialization authority leak
44.9 Transport branch flattening
44.10 Missing transport ledger
44.11 Holonomy without coherence
44.12 Assumed additive composition
44.13 Ignored associator defect
44.14 Scalarized complexity
44.15 Transition/singularity confusion
44.16 Stratum-history erasure
44.17 Resource-as-impossibility
44.18 Ontology before court exhaustion
44.19 New type without new executable consequence
44.20 Relabeling as progress
44.21 Hidden language frontier
44.22 Lossy semantic compression
44.23 Ghost executables
44.24 Duplicate semantic consequences
44.25 Replay-as-truth error
44.26 Global-closure leakage

Part XVII — Core Theorem Program

45. Foundational Theorem Targets

45.1 Residual object existence
45.2 Provisional-carrier existence
45.3 Carrier-minimization conditions
45.4 Carrier antichain theorem
45.5 Support existence
45.6 Support invariance under exact transport
45.7 Locality obstruction theorem
45.8 Factor-core invariance theorem
45.9 Epoch-equivalence refinement theorem
45.10 Transport-coherence theorem
45.11 Holonomy accumulation theorem
45.12 Composition-defect theorem
45.13 Associator-obstruction theorem
45.14 Interaction irreducibility theorem
45.15 Stratum refinement theorem
45.16 Singularity necessity theorem
45.17 Reduction-court soundness theorem
45.18 Ontology-extension necessity theorem
45.19 Discovery-orbit invariance theorem
45.20 Rehydration equivalence theorem

46. Nonclosure Theorem Program

46.1 Stable residual nonclosure
46.2 Stable language nonclosure
46.3 Stable ontology nonclosure
46.4 Nonclosure depth
46.5 Nonclosure recurrence
46.6 Nonclosure rank alternatives
46.7 Persistent residual classes
46.8 Successor-language chains
46.9 Finite versus unbounded extension
46.10 Nonclosure under transport
46.11 Nonclosure under composition
46.12 Nonclosure under interaction
46.13 Nonclosure under ontology repair
46.14 Discovery-orbit nonclosure
46.15 Global-closure impossibility as a forbidden inference

Part XVIII — Worked Research Programs

47. Projection and Reconstruction Residuals

47.1 Projection operator
47.2 Reconstruction operator
47.3 Composite residual
47.4 Support
47.5 Locality
47.6 Factorization
47.7 Composition defect
47.8 Transport
47.9 Singularity
47.10 Generated repair

48. Compression Residual Geometry

48.1 Compression carrier
48.2 Expansion carrier
48.3 Lost distinctions
48.4 Lost interactions
48.5 Lost ancestry
48.6 Compression support
48.7 Compression locality
48.8 Compression transport
48.9 Compression singularities
48.10 Residual-generated representation

49. Higher-Order Interaction Residuals

49.1 Joint residuals
49.2 Proper-subfamily closure
49.3 Pairwise shadows
49.4 Higher-arity irreducibility
49.5 Interaction carriers
49.6 Interaction support
49.7 Arity transitions
49.8 Interaction singularities
49.9 Operator generation
49.10 Ontology pressure

50. Local-to-Global Residual Geometry

50.1 Local residuals
50.2 Overlap residuals
50.3 Gluing residuals
50.4 Coherence residuals
50.5 Global reconstruction
50.6 Joint-only global defects
50.7 Boundary generation
50.8 Locality singularities
50.9 Repair construction
50.10 Successor mathematics

Part XIX — Cross-Theory Interfaces

51. Residual Geometry and Fracture Mathematics

51.1 Failed reconstruction
51.2 Fracture-induced residuals
51.3 Shared support
51.4 Fracture factorization
51.5 Interaction loss
51.6 Fracture singularities
51.7 Repair correspondence

52. Residual Geometry and Interaction Cohomology

52.1 Lower-order closure
52.2 Irreducible joint classes
52.3 Interaction obstruction
52.4 Arity structure
52.5 Local-to-global interaction
52.6 Residual classes versus cohomology classes
52.7 Constructor discovery

53. Residual Geometry and Semantic Curvature

53.1 Transport coherence
53.2 Path transport
53.3 Holonomy
53.4 Curvature candidates
53.5 Flat residual regions
53.6 Curvature singularities
53.7 Representation transitions

54. Residual Geometry and Grammar Cohomology

54.1 Unrepresentable residuals
54.2 Language gaps
54.3 Residual-induced primitives
54.4 Residual-induced laws
54.5 Successor grammar
54.6 Grammar obstruction classes
54.7 Stable language nonclosure

55. Residual Geometry and Constraint-Generated Ontology

55.1 Residual distinctions
55.2 Requirements
55.3 Reduction court
55.4 Operator necessity
55.5 Carrier necessity
55.6 Type necessity
55.7 Minimal ontology extension

Part XX — Open Problems

56. Foundational Open Problems

56.1 Minimality of the five primitive operators
56.2 Existence of residual carriers
56.3 Uniqueness of carrier minima
56.4 Completeness of factorization
56.5 Decidability of epoch-relative equivalence
56.6 Completeness of lower-order interaction closure
56.7 Transport-coherence classification
56.8 Classification of composition defects
56.9 Classification of associator residuals
56.10 Conditions for residual conservation
56.11 Stratum-history invariants
56.12 Singularity classification
56.13 Reduction-court completeness
56.14 Minimal ontology extensions
56.15 Discovery-orbit equivalence
56.16 Stable nonclosure
56.17 Residual geometry beyond finite closure bounds
56.18 Higher coherence of residual composition
56.19 Residual geometry under transfinite language growth
56.20 Criteria for genuinely new mathematics

57. Central Research Questions

57.1 When does failure define structure rather than noise?
57.2 Which residual relations are representation-independent?
57.3 When is a carrier mathematically forced?
57.4 When is locality recoverable?
57.5 When does transport generate curvature-like structure?
57.6 When does composition create information?
57.7 When is an interaction genuinely higher-order?
57.8 When must a stratum split after language growth?
57.9 When is a transition a singularity?
57.10 When does a residual force a new operator?
57.11 When does it force a new object type?
57.12 When is ontology growth unavoidable?
57.13 What is conserved through mathematical repair?
57.14 What defines equivalence between discovery orbits?
57.15 Can a mathematics remain permanently and structurally nonclosed?

GRM Residual Geometry

Architecture Document

RGΩ v0.1

0. Architectural Purpose

Residual Geometry is the GRM-constructed mathematics of structured failure.

Its primitive event is not “an error occurred.” It is:

operator F executes on x → output y is produced → a typed residual R_F(x) is generated

The residual records how the execution fails, differs, remains incomplete, loses structure, creates structure, or exposes an unresolved distinction.

The source program defines Residual Geometry as mathematics in which residuals themselves carry support, interaction, symmetry, locality, transport, and higher structure. It also places Residual Geometry explicitly before Zero-Defect Mathematics and Nonclosure, making residual structure the substrate from which those later theories are constructed.

The architecture therefore uses this spine:

SOURCE
→ OPERATOR
→ EXECUTION
→ RESIDUAL
→ RESIDUAL STRUCTURE
→ GEOMETRIC STRUCTURE
→ REQUIREMENT
→ NEW OPERATOR
→ POSSIBLE ONTOLOGY CHANGE
→ REPLAY

Residual Geometry does not begin with a manifold, metric, topology, vector space, coordinate system, dimension, or smooth structure.

Those are possible outputs, not admissible priors.

1. GRM Boundary

Residual Geometry is constructed mathematics. It is not part of the GRM kernel.

GRM v43 supplies a domain-neutral operator/residual interface:

REQ
OP
RUN
RCORE
REXT
RES
ZCLASS
RCLS
KREQ
SYNTH
CHECK

A residual has a minimal kernel core:

RCORE = <raw, nf, owner, anc, ext>

while source mathematics may extend the residual through validated RESIDUAL_EXTENSION schemas. GRM explicitly preserves the complete residual before projection, assigns zero semantics to the generating operator, and allows the residual vocabulary to expand without modifying the kernel.

This is important because GRM v41 previously hard-coded fields such as factor, support, symmetry, rank, locality, interaction, and transport into its residual schema.

Residual Geometry should recover those structures as source-owned extensions rather than regress to a topic-contaminated kernel.

The architecture is therefore:

GRM kernel
    RCORE
    REXT
    OP
    MOR
    LIFT
    BEQ
    REQ
    SYNTH

        ↓ constructs

Residual Geometry source layer
    Carrier
    Factorization
    Support
    Locality
    Symmetry
    Rank
    Multiplicity
    Interaction
    Transport
    Composition
    Equivalence
    Stratification
    Singularity
    Generated Operator
    Generated Ontology

2. Root Residual Object

For an executable operator

F : X → Y

and admissible input x ∈ X:

y = RUN(F,x)

r = RES(F,x)

Residual Geometry promotes this execution into a residual event:

RGEvent = <F, x, y, r, scope, owner, ancestry>

The mathematical residual is therefore not merely r.raw.

Its identity depends on the generating operation, source context, scope, carrier, ancestry, and executable consequences.

The global residual population is heterogeneous:

R = disjoint union of R[F, carrier, scope, ancestry]

Residuals belonging to different fibers cannot be compared merely because their encoded values resemble one another.

This prevents numerical or representational coincidence from becoming semantic identity.

4.1 Residuals as Mathematical Objects

The first architectural decision is that a residual persists as an object.

A failure label is not the residual.

A debt token is not the residual.

A scalar magnitude is not the residual.

A classification is not the residual.

The complete residual is retained first; projections are derived afterward. This follows GRM's residual-first rule and the program's definition of a residual as structured output describing how a construction fails, differs, or remains incomplete.

The object interface is:

RG.RESIDUAL(F,x)
    → RGResidual

RGResidual
    core
    generating_operator
    source_input
    produced_output
    scope
    owner
    ancestry
    extensions

A valid residual object must remain replayable back to the execution that generated it.

Architecture invariant:

same raw payload ≠ same residual object

unless equivalence is independently witnessed.

4.2 Residual Carriers

Every residual requires a structure on which its distinctions remain meaningful.

Define the carrier candidate set:

CarrierCandidates(r)

as all structures on which the residual can be represented, restricted, tested, and transported without destroying source-required consequences.

The residual carrier is the minimal admissible carrier or antichain of minimal carriers:

Carrier(r) = Minimal(CarrierCandidates(r))

Uniqueness is not assumed.

Possible carrier forms include a source object, region, interaction domain, boundary, dependency hypergraph, family of components, construction path, or newly generated carrier.

The key invariant is:

same residual formula on different carriers ≠ same mathematics

unless carrier transport proves equivalence.

Carrier inadequacy is itself detectable. If the current carrier cannot express a source-required distinction but another carrier can, Residual Geometry emits a carrier mutation requirement.

residual
→ carrier inadequacy
→ carrier requirement
→ candidate carrier
→ transport/reconstruction test
→ accept or reject

A carrier is generated because the residual requires it, not because an available mathematical formalism makes it convenient.

4.3 Residual Factorization

Factorization asks whether a residual can be generated from smaller residual structures.

A candidate factorization has the form:

r ≃ ComposeResidual(r1,...,rn)

It is accepted only with a witness that the reconstruction preserves the residual consequences required by the source.

Define:

Factorizations(r)

as the branch of all admissible witnessed factorizations.

Residual Geometry does not assume unique factorization.

Different decompositions may expose different interaction structures. Therefore the architecture retains all unresolved factorization branches.

The factorization-invariant core is the structure shared across all source-equivalent admissible factorizations.

FactorCore(r)
    = consequences surviving every admissible factorization

If factorization branches disagree on a supposedly structural feature, that feature cannot yet be promoted to an invariant.

This prevents a decomposition chosen by representation convenience from becoming mathematical ontology.

4.4 Residual Support

Support answers:

Where is the residual actually nontrivial?

Given a carrier region or subobject U, construct a restriction:

r|U

through a licensed restriction/transport operation.

Support is then generated from the minimal regions on which the residual remains nonzero under its proper operator-owned zero semantics.

Support(r)
    = minimal U such that ZClass(r|U) ≠ ZERO

Support may be localized, distributed, disconnected, boundary-supported, interaction-supported, or joint-only.

The important adversarial case is joint-only support:

for every proper region U:
    ZClass(r|U) = ZERO

but:

    ZClass(r) = NONZERO

A residual can therefore exist globally while disappearing under every lower-order restriction.

This prevents the architecture from reducing support to visible local defects.

4.5 Residual Locality

Support and locality are different.

Support asks where the residual exists.

Locality asks whether the residual can be reconstructed from local pieces.

For a source-generated family of regions:

U = {U1,...,Un}

construct:

Localize(r,U) = {r|U1,...,r|Un}

If a licensed gluing constructor exists:

Glue_U(Localize(r,U)) → r*

then define the locality residual:

LocalityResidual(r,U) = CompareResidual(r,r*)

Classification:

ZERO
    locally reconstructible

NONZERO
    irreducible local-to-global defect

UNKNOWN
    current grammar cannot decide locality

Therefore:

locally visible ≠ locally reconstructible

and:

all local residuals zero ≠ global residual zero

This distinction is structurally necessary. Otherwise Residual Geometry would silently erase exactly the higher-order information it is intended to detect.

4.6 Residual Symmetry

A residual symmetry is a source-admissible transformation under which residual behavior is preserved.

For a carrier transformation:

g : C → C

transport the residual:

Transport(g,r) → r'

Then g is a residual symmetry when:

r' ≃R r

Residual Geometry distinguishes at least three levels:

exact symmetry
— residual object preserved;

behavioral symmetry
— all required residual consequences preserved;

class symmetry
— only the residual classification preserved.

These are not interchangeable.

The symmetry object is therefore not automatically a classical symmetry group. Closure, inversion, and composition must be demonstrated by the source transformations.

A change in residual symmetry across a family is a structural event:

Sym(r_before) ≠ Sym(r_after)

Such changes later become one source of residual stratification and singularity.

4.7 Residual Rank and Multiplicity

A single scalar “rank” is too destructive.

Residual Geometry therefore uses a rank profile.

ResidualRank(r) =
    factor rank
    carrier rank
    interaction rank
    transport rank
    ancestry rank

Factor rank measures the minimal irreducible residual factors required to generate r.

Carrier rank measures the minimal carrier complexity required to sustain r.

Interaction rank measures the smallest arity at which the residual becomes irreducible.

Transport rank measures irreducible transport depth.

Ancestry rank measures how much construction history is necessary to preserve residual identity.

These ranks remain distinct unless the source proves a relation among them.

Multiplicity measures something different:

Multiplicity([r])

is the number or structure of generatively distinct realizations belonging to the same residual equivalence class.

Thus a residual may have low structural rank but high multiplicity.

This separates complexity of structure from frequency of realization.

4.8 Residual Interactions

Residuals can themselves interact.

Given residual objects:

r1,...,rn

construct the directly evaluated joint residual:

r[1...n]

Then construct everything obtainable from proper subfamilies.

The irreducible interaction residual is:

InteractionResidual(r1,...,rn)
    =
    consequences of r[1...n]
    not generated by the closure
    of all proper-subset residuals

No numerical subtraction is implied.

The interaction arity is the smallest n for which the joint residual contains irreducible consequences.

arity(r) = minimal n with irreducible n-way residual structure

This architecture explicitly forbids the pairwise streetlight:

pairwise analysis cannot establish absence of higher-order residual structure

GRM's earlier residual machinery already used an interaction_signature and required witness-backed failure of lower-arity constructor families before arity escalation. That discipline remains useful at the Residual Geometry layer.

4.9 Residual Transport

Transport moves residual structure across an admissible transformation.

GRM provides generic morphism and lift machinery rather than assigning mathematical meaning to the morphism itself. A morphism has source, destination, kind, map, data, checking semantics, owner, and ancestry; valid operator lifts are constructed through source-owned morphism specifications.

Residual Geometry builds on this:

z : C → C'

F on C

Lift(z,F) → {F'1,...,F'n}

Transport(z,R_F(x))
    → {R_F'1(zx),...,R_F'n(zx)}

Transport is therefore generally branch-valued.

A transport record contains:

source residual
target residual
preserved structure
lost structure
created structure
branch identity
ancestry
transport witness

Exact transport requires both zero unexplained loss and zero unexplained creation.

For a path:

γ = z_n ∘ ... ∘ z_1

Residual Geometry composes transports.

For a closed path, compare the returned residual with the original:

ResidualHolonomy(γ,r) = CompareResidual(r,Transport(γ,r))

A nonzero return defect is generated geometric structure.

No curvature terminology is needed yet.

4.10 Residual Composition

Operator composition and residual composition are not the same operation.

For:

X --F→ Y --G→ Z

construct separately:

rF  = R_F(x)
rG  = R_G(F(x))
rGF = R_(G∘F)(x)

A residual compositor proposes:

ResidualCompose(rG,rF) → r*

The composition defect is:

CompositionResidual(G,F,x) = CompareResidual(rGF,r*)

This distinguishes:

ZERO
    component residual structure reconstructs the composite residual

NONZERO
    composition creates irreducible residual structure

UNKNOWN
    current language cannot decide

The architecture therefore forbids:

R_(G∘F) = R_G + R_F

as a default law.

For three operators, two residual assembly routes can be compared:

F → G → H

If the two residual reconstructions disagree, the resulting associator residual records structure invisible to pairwise composition.

This is one of the strongest candidate invariants produced by the architecture because it can detect higher-order failure even when the underlying ordinary operator composition remains associative.

4.11 Residual Equivalence

Residual Geometry requires multiple equivalence layers.

Serialization equality is the weakest.

Normalized equality removes representational variation.

Class equivalence preserves a residual label.

Behavioral equivalence preserves required executable consequences.

Transport equivalence requires witnessed reversible transport.

Generative equivalence preserves relevant factorization and ancestry.

Full residual equivalence requires every source-required residual consequence to survive a witnessed bidirectional transformation.

GRM already enforces the critical branching rule:

BEQ = ✓   → merge permitted
BEQ = ⊥   → separate
BEQ = ?   → separate

Unknown equivalence is not silently treated as identity.

Residual Geometry therefore constructs a residual moduli structure from equivalence classes and unresolved branches rather than selecting one serialized representative as canonical mathematics.

4.12 Residual Stratification

Stratification is derived from stable residual structure.

It is not imposed in advance.

Define a residual signature:

ResidualSignature(r) =
    zero class
    carrier class
    support class
    factor class
    locality class
    symmetry class
    rank profile
    multiplicity structure
    interaction arity
    transport class
    composition class

Residuals belong to the same provisional stratum when they have equivalent signatures and are connected through admissible residual transports that do not cross a structural transition.

ResidualSpace
    → Stratum A
    → Stratum B
    → ...

The important architectural inversion is:

strata come from residual invariants

rather than:

residuals are forced into predetermined strata

This follows the earlier GRM principle that filtration, ordering, covering, and stratification should be derived special cases rather than privileged prerequisites.

4.13 Residual Singularities

A residual singularity is a point or regime where the current stable residual structure cannot be continued without changing type.

No derivative divergence is required.

A singularity exists when arbitrarily nearby admissible residual states exhibit incompatible residual signatures.

Primary singularity classes are:

carrier singularity
support singularity
factorization singularity
locality singularity
symmetry singularity
rank singularity
interaction singularity
transport singularity
composition singularity
ontology singularity

An ontology singularity is the strongest case.

It occurs when a live residual contains a distinction that the current residual vocabulary cannot type or express.

That condition is not interpreted as impossibility.

It becomes language pressure:

unrepresentable residual distinction
→ language gap
→ new requirement

4.14 Residual-Generated Operators

Residual Geometry reverses the usual operator-to-residual direction.

Normal execution:

Operator → Residual

Discovery execution:

Residual → Requirement → New Operator

For persistent residual r, compile:

ResidualRequirement(r)

This requirement describes the minimal capability necessary to remove, preserve, distinguish, transport, factor, or otherwise resolve the residual.

GRM then supplies:

KREQ → SYNTH → CHECK

which allows candidate operators to be generated and tested.

A generated operator is accepted only if it causes structural residual progress.

Acceptable progress includes reduction of an irreducible residual, resolution of an unknown branch, successful transport previously unavailable, lower required interaction arity, closure of a locality defect, or construction of a capability that previously did not exist.

Renaming the residual is not progress.

Reclassifying the residual is not progress.

Compressing the residual into a debt token is not progress.

The new operator must alter executable consequences.

4.15 Residual-Generated Ontology

Residual-generated ontology occurs only when the current mathematical object vocabulary cannot satisfy a source-required residual requirement.

Let the current language be Γ.

Suppose a persistent residual remains unresolved under every available construction in Closure(Γ).

The architecture extracts the missing distinction:

r → distinction d

Then:

d → requirement q

Suppose satisfying q requires an object type τ' that cannot be constructed or behaviorally simulated inside Γ.

Then Residual Geometry may generate:

new type τ'
new operator family Op(τ')
new laws required for τ'
new language Γ'

with:

Γ' = Γ + ontology delta

A new object type is legitimate only when it changes what can actually be constructed or distinguished.

If an existing object type can satisfy the same requirement under a witnessed equivalence, the proposed type is rejected as relabeling.

The architecture therefore requires:

persistent residual
→ irreducible missing distinction
→ unmet requirement
→ absence of equivalent existing capability
→ new type
→ new executable capability
→ residual consequence changes

Only then is ontology growth earned.

5. Actual Execution Dependency Graph

The chapter order is pedagogical. The runtime dependency order is different.

In particular, symmetry requires transport and equivalence; stratification requires almost every preceding analysis; singularity requires stratification; generated operators require persistent residual structure.

The executable order is therefore:

R_F(x)

→ validate complete residual
→ determine candidate carrier
→ build restriction operations
→ test support
→ test factorization
→ test locality

→ construct morphisms and lifts
→ construct residual transport
→ construct residual equivalence

→ derive symmetry
→ derive rank and multiplicity
→ test higher-order interactions
→ test residual composition

→ compute residual signature
→ derive strata
→ detect singularities

→ compile persistent residual requirements
→ synthesize new operators

→ test whether operator synthesis requires new object types
→ generate ontology delta if forced

→ replay everything

This ordering removes a hidden circularity in the chapter sequence.

For example, “symmetry” cannot be rigorously defined before the architecture knows what legitimate residual transport and equivalence mean.

6. Residual Geometry Source Package

Residual Geometry should be implemented as source-owned residual extensions.

RG.CARRIER
RG.FACTORIZATION
RG.SUPPORT
RG.LOCALITY
RG.TRANSPORT
RG.EQUIVALENCE
RG.SYMMETRY
RG.RANK
RG.MULTIPLICITY
RG.INTERACTION
RG.COMPOSITION
RG.STRATUM
RG.SINGULARITY
RG.OPERATOR_GENERATION
RG.ONTOLOGY_GENERATION

Each extension has:

schema
validator
constructor
dependencies
ancestry
scope
owner
replay rule

The kernel sees these only as validated residual extensions.

It does not acquire their subject-matter semantics.

7. Persistent Versus Derived State

Not every calculated feature belongs in persistent storage.

Persist the things whose loss changes replay:

raw residual
operator identity
source execution
ancestry
source-owned extension data
restriction witnesses
factorization witnesses
transport witnesses
equivalence/disequivalence witnesses
interaction witnesses
composition witnesses
generated requirements
accepted generated operators
ontology deltas

Prefer to derive:

support when uniquely reconstructible
rank profile
multiplicity
symmetry class
residual signature
stratum membership
singularity classification

If one of those cannot be reconstructed uniquely, it stops being merely derived and its missing witness becomes explicit state.

The architecture therefore follows:

persist causes and irreducible witnesses; derive consequences

rather than caching interpretive labels as authority.

8. Replay Contract

Residual Geometry is valid only if cold replay reconstructs the same residual mathematics.

Replay must reconstruct:

residual objects
carriers
restriction behavior
factorization branches
support
locality defects
transport branches
equivalence branches
symmetries
rank profiles
multiplicity
interaction structure
composition defects
strata
singularities
generated requirements
generated operators
ontology deltas

A byte-identical serialization is insufficient if semantic reconstruction differs.

Conversely, different serialization is not a mathematical difference when witnessed residual equivalence is preserved.

9. Residual Geometry Adversarial Court

The architecture rejects the following failure modes.

Scalar residual collapse
A structured residual is replaced by an error magnitude.

Universal-zero collapse
raw = 0 is interpreted as absence independently of the generating operator.

Carrier streetlight
A familiar carrier is selected because existing mathematics is easiest there.

Premature factorization
One decomposition is promoted while valid alternatives remain unresolved.

Local-to-global collapse
Local zero residuals are used to infer global zero.

Pairwise streetlight
Failure to detect pairwise residual structure is treated as absence of higher-order structure.

Transport flattening
Branch-valued transport is replaced by a single convenient lift.

Additive-composition assumption
Composite residuals are assumed to be sums of component residuals.

Unknown-equivalence collapse
Unresolved residual equivalence is silently treated as equality.

Premature stratification
Residuals are forced into a topology, manifold, filtration, or stratification selected beforehand.

Classical singularity streetlight
Only divergent or nondifferentiable behavior is allowed to count as singular.

Relabel-as-operator
A new failure name is presented as a new mathematical operation.

Relabel-as-ontology
A renamed object is presented as a new mathematical type without changing constructible consequences.

Residual suppression
A residual that cannot be expressed by the current language is discarded instead of generating language pressure.

10. Primary Residual Geometry Invariants

The first invariant family should be structural rather than metric:

Carrier(r)
Support(r)
FactorCore(r)
LocalityResidual(r,U)
Symmetry(r)
RankProfile(r)
Multiplicity(r)
InteractionArity(r)
TransportClass(r)
CompositionResidual(G,F,x)
AssociatorResidual(H,G,F,x)
ResidualHolonomy(path,r)
ResidualSignature(r)
Stratum(r)
SingularityType(r)
OntologyPressure(r)

These invariants can later generate metric, topological, categorical, cohomological, or differential structures if the residual behavior forces them.

The direction must not be reversed.

11. System Architecture

                     GRM KERNEL
                         │
          REQ ─ OP ─ RES ─ MOR ─ LIFT ─ BEQ
                         │
                         ▼
                  COMPLETE RESIDUAL
                         │
              RCORE + RG EXTENSIONS
                         │
        ┌────────────────┼────────────────┐
        │                │                │
     Carrier        Factorization      Restriction
        │                │                │
        └───────┬────────┴───────┬────────┘
                │                │
             Support          Locality
                │                │
                └───────┬────────┘
                        │
                    Transport
                        │
                   Equivalence
                        │
          ┌─────────────┼─────────────┐
          │             │             │
       Symmetry        Rank       Interaction
          │             │             │
          └─────────────┼─────────────┘
                        │
                   Composition
                        │
                Residual Signature
                        │
                  Stratification
                        │
                   Singularity
                        │
              Persistent Residual?
                        │
                       yes
                        │
                 Requirement Compiler
                        │
                     Synthesis
                        │
                 Generated Operator
                        │
              Existing ontology enough?
                   │            │
                  yes           no
                   │            │
                Replay     Ontology Delta
                                │
                              Replay

12. Core Architectural Insight

Residual Geometry is not geometry applied to residuals.

That framing is too weak.

Its actual construction is:

execution produces residuals

residual relations produce carriers

carrier-relative residual behavior produces support and locality

transport produces equivalence, symmetry and return defects

joint residual behavior produces interaction structure

composition produces higher-order defects

stable residual signatures produce strata

signature breakdown produces singularities

persistent singular structure produces capability requirements

requirements produce new operators

new operators may force new object types

The architecture therefore reverses a standard mathematical workflow.

Instead of:

choose space
→ define objects
→ define operators
→ measure errors

Residual Geometry uses:

execute operators
→ preserve failures
→ discover structure among failures
→ derive the relevant geometry
→ mutate the mathematical language when required

That is the distinctive architectural claim of GRM Residual Geometry.

13. Canonical Runtime Spine

Σ
→ F
→ RUN
→ R_F(x)

→ Carrier
→ Restrict
→ Factor
→ Support
→ Locality

→ Morphism
→ Lift
→ Transport
→ Equivalence

→ Symmetry
→ Rank
→ Multiplicity
→ Interaction
→ Composition

→ Signature
→ Stratum
→ Singularity

→ Requirement
→ Synthesized Operator
→ Ontology Delta when forced

→ Replay
→ next residual state

The terminal rule is simple:

A residual is never merely something to minimize.

It is information about what the current mathematical construction cannot yet preserve, distinguish, transport, compose, reconstruct, or express.

Residual Geometry exists to turn that information into mathematical structure without allowing the current representation to decide in advance what that structure is.