Introduction to Number Theory EG

Introduction to Number Theory 

2026 Exposure Geometry Edition

Table of Contents

Part I — Arithmetic Distinctions, Frames, and Skeletons

Chapter 1 — Number Theory as Exposure Geometry

1.1 Introduction: Integers as Distinction-Bearing Carriers
1.2 Survey: Local Arithmetic, Global Arithmetic, and Reconstruction
1.3 Arithmetic EG Frames, Skeletons, Residues, and Successor Seeds
1.4 Equality, Congruence, Association, Isomorphism, and Equivalence
1.5 Arithmetic Relations as Typed Transports
1.6 Existence, Construction, Uniqueness, Classification, and Liftback
1.7 Local Validity Versus Global Reconstruction
1.8 Dyadic Data, Higher-Arity Arithmetic, and Cell Residues
1.9 Counterexamples as Minimal Arithmetic Counterkernels
1.10 Formal Theorems, Algorithms, Computations, and Claim Levels
1.11 Noncompensatory Proof Obligations
1.12 The Arithmetic Execution Sequence: Type to Certificate

Chapter 2 — Divisibility, Euclidean Transport, and Prime Skeletons

2.1 Divisibility as a Typed Dyadic Relation
2.2 The Division Algorithm as Canonical Reduction Transport
2.3 Euclid’s Algorithm and Descent Through Remainder Carriers
2.4 Extended Euclid and Exact Bézout Liftback
2.5 Linear Diophantine Equations as Reconstruction Problems
2.6 GCD and LCM as Componentwise Arithmetic Collapse
2.7 Prime Numbers and Irreducible Arithmetic Distinctions
2.8 Fundamental Theorem of Arithmetic as Skeleton Reconstruction
2.9 Prime Support and Valuation Geometry
2.10 Units, Associates, Sign, and Source-Fibre Ambiguity
2.11 Euclid’s Lemma as a Transport-Closure Gate
2.12 Prime Gaps and Worst-Fibre Arithmetic
2.13 Bit Complexity and Euclidean Proof-Carrying Computation
2.14 Divisibility Residues and Minimal Solvability Counterkernels

Part II — Multiplicative Composition and Combinatorial Cells

Chapter 3 — Arithmetic Functions, Convolution, and Carry Geometry

3.1 Multiplicative Functions as Prime-Power Reconstruction Laws
3.2 Completely Multiplicative Versus Coprime-Local Composition
3.3 Divisor Functions and Prime-Support Skeletons
3.4 Euler’s Totient Function as Unit-Carrier Measurement
3.5 The Möbius Function and Inclusion–Exclusion Residue
3.6 Dirichlet Convolution as Arithmetic Composition
3.7 Möbius Inversion as Exact Projection Recovery
3.8 Binomial Coefficients as Finite Interaction Cells
3.9 Pascal and Vandermonde Composition Laws
3.10 Binomial Inversion and Reconstruction from Aggregated Data
3.11 Legendre’s Formula and Factorial Valuation Depth
3.12 Kummer Carry Geometry
3.13 Prime-Power Divisibility and Frobenius Structure
3.14 Combinatorial Explosion, Multiplicity, and Support Reuse
3.15 Average Order Versus Exceptional Arithmetic Fibres
3.16 Forgotten-Distinction Ledgers for Combinatorial Projections

Part III — Quotient Carriers and Local-to-Global Reconstruction

Chapter 4 — Congruences, Units, and Chinese Reconstruction

4.1 Congruences as Quotient Transports
4.2 Residue Classes as Compacted Arithmetic Carriers
4.3 Units, Zero Divisors, and the Cancellation Boundary
4.4 Linear Congruences and Solution-Fibre Multiplicity
4.5 Fermat’s Theorem as Finite-Field Transport Closure
4.6 Euler’s Theorem and Unit-Group Exponents
4.7 Wilson’s Theorem and Inverse-Pair Reconstruction
4.8 The Chinese Remainder Theorem as Exact Local Liftback
4.9 CRT Idempotents and Independent Local Components
4.10 Overlapping Moduli and Boundary Compatibility Residue
4.11 Euler’s Totient Function and Local Unit Capacity
4.12 Carmichael Numbers and Nominal-Theorem Instability
4.13 Fermat Pseudoprimes as Adversarial Counterkernels
4.14 Projection Loss Under Reduction Modulo n
4.15 Local Congruence Data Versus Global Integer Ancestry

Chapter 5 — Computational Number Theory, Certificates, and Cryptography

5.1 Numerical Calculation and Finite Encoding
5.2 Binary Exponentiation as Compressed Transport
5.3 Modular Inversion and Executable Bézout Certificates
5.4 Computational Complexity in Bit Carriers
5.5 Probable Primes and One-Sided Falsifiers
5.6 Miller–Rabin and Structured Adversarial Replay
5.7 Deterministic and Certificate-Based Primality
5.8 Trial Division and Fermat Factorization
5.9 Pollard Rho and Collision Geometry
5.10 Pollard p−1, ECM, Quadratic Sieve, and Number Field Sieve
5.11 Factor Discovery Versus Complete Factorization Liftback
5.12 RSA as Modular Transport and CRT Reconstruction
5.13 Cryptographic Support Duplication and Shared-Prime Ruin
5.14 Nonce Reuse, One-Use Capacity, and Liability Ownership
5.15 Timing, Fault, and Side-Channel Residues
5.16 Mathematical Correctness Versus Implementation Embodiment
5.17 Classical Cryptography Under Quantum Factorization Exposure
5.18 Reproducible Computational Records and Independent Replay

Part IV — Prime-Local Geometry and p-adic Reconstruction

Chapter 6 — Polynomial Congruences, Finite Fields, and p-adic Numbers

6.1 Prime-Valuation Carriers and Ultrametric Distinctions
6.2 p-adic Completion as Successor-Frame Construction
6.3 Compatible Residue Towers and Source Ancestry
6.4 Hensel’s Lemma as Local Liftback
6.5 Newton Transport and Precision Amplification
6.6 Simple Roots, Singular Roots, and Boundary Fracture
6.7 Nonunique Lifts and Singular-Fibre Residues
6.8 Congruences Modulo a Prime
6.9 Finite Fields as Closed Arithmetic Carriers
6.10 Frobenius Transport and Characteristic-p Structure
6.11 Polynomial Interpolation and Exact Reconstruction
6.12 Chevalley–Warning and Variable-Capacity Geometry
6.13 Primitive Roots and Cyclic Multiplicative Skeletons
6.14 Primitive Roots for Prime Powers
6.15 Classification of Moduli with Primitive Roots
6.16 Quadratic Equations Modulo p
6.17 Root Multiplicity and Derivative-Based Separation
6.18 Root Bounds Over Fields and Failure Over Composite Rings
6.19 Local Solvability Versus Global Solvability
6.20 p-adic Precision, Truncation, and Resource Ledgers

Part V — Algebraic Carriers, Orbits, and Quotients

Chapter 7 — Groups, Rings, Fields, and Ideal Reconstruction

7.1 Groups as Arithmetic Transport Systems
7.2 Element Order and Cyclic Reconstruction
7.3 Homomorphisms, Kernels, Images, and Quotients
7.4 Group Actions, Orbits, and Stabilizers
7.5 Invariant Equality Versus Orbit Equality
7.6 Products of Groups and CRT Composition
7.7 Diagonal Coupling and Failure of Componentwise Reconstruction
7.8 Rings as Multi-Operation Arithmetic Carriers
7.9 Units, Zero Divisors, Irreducibles, and Prime Elements
7.10 Ideals as Canonical Relation Carriers
7.11 Prime and Maximal Ideals as Quotient Boundaries
7.12 Euclidean Domains, PIDs, and UFDs
7.13 Failure of Element Factorization and Ideal-Level Repair
7.14 Fields and Finite Extension Carriers
7.15 Frobenius Orbits, Trace, and Norm
7.16 Finite Abelian Group Decomposition
7.17 Proof-Carrying Algebraic Computation
7.18 Source Fibres Under Quotient and Invariant Maps

Part VI — Reciprocity as Path-Comparison Geometry

Chapter 8 — Quadratic Reciprocity, Characters, and Gauss Transport

8.1 Quadratic Residues and the Squaring Projection
8.2 The Legendre Symbol as a Local Exposure Signature
8.3 Euler’s Criterion and Character Evaluation
8.4 Calculation of the Legendre Symbol
8.5 Quadratic Reciprocity as Bidirectional Prime Transport
8.6 Reciprocity Signs as Path-Comparison Residues
8.7 First and Second Supplementary Laws
8.8 Gauss’s Lemma and Boundary-Crossing Counts
8.9 Gauss Sums and Additive–Multiplicative Carrier Interaction
8.10 Fourier Transport Over Finite Fields
8.11 The Jacobi Symbol as a Compacted Prime-Factor Projection
8.12 Jacobi Value Versus Actual Quadratic Solvability
8.13 The Kronecker Symbol and Extended Boundary Typing
8.14 Local Quadratic Symbols and Global Compatibility
8.15 Reciprocity Counterkernels and Exact Symbol Liftback
8.16 Computational Reciprocity and Symbol Certificates

Part VII — Reduction, Equivalence, and Global Orbit Residues

Chapter 9 — Continued Fractions and Binary Quadratic Forms

9.1 Continued Fractions as Recorded Euclidean Transport
9.2 Convergents and Minimal Approximation Residue
9.3 Periodicity and Quadratic-Irrational Source Structure
9.4 Pell Equations and Cyclic Unit Transport
9.5 Binary Quadratic Forms as Arithmetic Carriers
9.6 Discriminants as Invariants but Not Complete Classifiers
9.7 Proper and Improper Equivalence
9.8 Change-of-Variables Move Groupoids
9.9 Reduction as Skeleton Compaction
9.10 Positive Definite Forms and Finite Reduced Regions
9.11 Examples of Positive Definite Forms
9.12 Gauss Composition and Class-Group Structure
9.13 More Examples of Binary Quadratic Forms
9.14 Genus Projection and Global Class Residue
9.15 Local Equivalence Versus Global Equivalence
9.16 Indefinite Binary Quadratic Forms
9.17 Reduction Cycles, Path Order, and Infinite Stabilizers
9.18 Representation Transport and Exact Witness Liftback
9.19 Class Groups as Factorization Counterkernels
9.20 Algorithmic Reduction and Replay Certificates

Part VIII — Algebraic Integers and Parametric Liftback

Chapter 10 — Gaussian Integers and Special Integer Structures

10.1 Gaussian Integers as a Two-Dimensional Arithmetic Carrier
10.2 Norm, Units, and Euclidean Descent
10.3 Gaussian Prime Classification
10.4 Splitting, Inertness, and Ramification
10.5 Sums of Two Squares as Norm Reconstruction
10.6 Element Factorization and Ambient Ring Dependence
10.7 Eisenstein Integers and Cubic Symmetry
10.8 Norm Equations and Algebraic Liftback
10.9 Pythagorean Triangles
10.10 Primitive Triples and Coprimality Gates
10.11 Rational Parameterization as Exact Construction
10.12 Uniqueness, Sign, Ordering, and Source-Fibre Conventions
10.13 Closure of Parametric Families
10.14 Projection from Algebraic Factorization to Integer Solutions

Part IX — Analytic Reconstruction and Nonuniformity Exposure

Chapter 11 — Dirichlet Series, L-functions, and Prime Distribution

11.1 Dirichlet Series as Analytic Arithmetic Carriers
11.2 Coefficient Data and Regions of Convergence
11.3 Products of Dirichlet Series
11.4 Euler Products as Prime-Local Reconstruction
11.5 Absolute Convergence as a Composition Gate
11.6 Analytic Continuation as Carrier Migration
11.7 The Zeta Function and Singular Structure
11.8 The Prime Number Theorem
11.9 Proof of the Prime Number Theorem
11.10 Zero-Free Boundaries and Tauberian Liftback
11.11 Dirichlet’s Theorem on Arithmetic Progressions
11.12 Dirichlet Characters as Residue-Class Projectors
11.13 Primitive Characters, Conductors, and Native Carriers
11.14 Proof of Dirichlet’s Theorem
11.15 Nonvanishing of L-series at s = 1
11.16 Exceptional Real Characters and Worst-Fibre Replay
11.17 Character Orthogonality and Projection Recovery
11.18 Asymptotic Main Terms Versus Local Error Residues
11.19 Nonuniformity in Modulus, Height, and Zero Proximity
11.20 Explicit Formulae as Prime–Zero Transport
11.21 Numerical Verification, Truncation, and Proof Capacity
11.22 Analytic Claims, Computational Claims, and Scope Discipline

Part X — 2026 Arithmetic Extensions

Chapter 12 — Elliptic Curves and Local–Global Arithmetic

12.1 Cubic Curves as Algebraic Arithmetic Carriers
12.2 Singular Versus Nonsingular Cubics
12.3 The Chord-and-Tangent Composition Law
12.4 Rational Points as a Finitely Generated Group
12.5 Reduction Modulo Primes
12.6 Good Reduction, Bad Reduction, and Boundary Residue
12.7 Frobenius and Point Counting
12.8 Heights and Arithmetic Scale
12.9 Descent as Projection, Obstruction, and Liftback
12.10 Local Points, Global Points, and Reconstruction Failure
12.11 Torsion, Rank, and Source-Fibre Structure
12.12 Elliptic-Curve Factorization
12.13 Elliptic-Curve Cryptography
12.14 Subgroup, Cofactor, and Invalid-Curve Counterkernels
12.15 Formal Theorem Versus Executed Cryptographic System

Chapter 13 — Formal and Computational Number Theory in 2026

13.1 Exact Integer, Rational, Polynomial, and Finite-Field Computation
13.2 p-adic and Algebraic-Number Computation
13.3 Computer Algebra Systems as Typed Proof Carriers
13.4 Primality Certificates and Independently Checkable Witnesses
13.5 Certified Factorization Pipelines
13.6 Formalization in Proof Assistants
13.7 Proof Object, Certificate, Replay, and Trusted Kernel
13.8 Randomized Algorithms and Explicit Probability Claims
13.9 Precision, Accuracy, Truncation, and Representation Residues
13.10 Memory, Runtime, Parallelism, and Resource Accounting
13.11 Reproducible Seeds, Logs, Versions, and Hashes
13.12 Structured Perturbation and Adversarial Test Suites
13.13 No-Ghost Computation: Formal Existence Is Not Execution
13.14 Promotion from Φ0 Formal to Φ1 Computational

Part XI — Number-Theoretic Exposure Geometry

Chapter 14 — Arithmetic Frame, Residue, and Successor Construction

14.1 The Maximal Arithmetic EG Frame
14.2 Target-Relative Arithmetic Skeleton Extraction
14.3 Prime, Local, Algebraic, and Analytic Carriers
14.4 Dyadic Relations and Higher-Arity Arithmetic Cells
14.5 Transport, Composition, and Path-Comparison Residue
14.6 Ambient Rings, Fields, Completions, and Embeddings
14.7 Source Ancestry and Reconstruction Fibres
14.8 M0 Worst-Fibre and Tail Exposure in Number Theory
14.9 M1 Parameter Nonuniformity in Bounds and Error Terms
14.10 M2 Boundary Singularity in Roots, Fibres, and Analytic Continuation
14.11 M3 Interaction Multiplicity and Prime-Support Reuse
14.12 M4 Projection Loss in Congruences, Symbols, Invariants, and Averages
14.13 M5 Proof-Carrier Capacity and Verification Complexity
14.14 M6 Structured Perturbation and Adversarial Arithmetic
14.15 M7 Framework-Capacity and Decidability Boundaries
14.16 M8 Ambient Entanglement in Rings, Ideals, Forms, and Global Orbits
14.17 M9 Path-Order Dependence in Descent, Lifting, and Reduction
14.18 M10 Source-Ancestry Fibres Under Quotients and Completions
14.19 Arithmetic Debt and Forgotten-Distinction Ledgers
14.20 Minimal Arithmetic Residue Extraction
14.21 Counterkernel Construction
14.22 Successor-Frame Migration
14.23 Exact Liftback to the Original Number-Theoretic Claim
14.24 Independent Replay and Scoped Certification
14.25 Frontier Payloads for Unresolved Arithmetic Reconstruction

Appendix — Computational and Exposure Tools

A.1 Three Calculators for Number Theorists
A.2 Arbitrary-Precision Integer Calculator
A.3 Modular Arithmetic and CRT Calculator
A.4 Prime, Factorization, and Primality-Certificate Tools
A.5 Finite-Field and Polynomial Calculator
A.6 p-adic Expansion and Hensel-Lifting Calculator
A.7 Legendre, Jacobi, Kronecker, and Character Calculator
A.8 Continued-Fraction and Pell Calculator
A.9 Binary Quadratic Form Reduction and Composition
A.10 Gaussian Integer and Norm Calculator
A.11 Dirichlet Series and L-function Numerical Tools
A.12 Elliptic-Curve Arithmetic Tools
A.13 Proof Certificate and Replay Validator
A.14 Arithmetic EG Frame Builder
A.15 Residue and Counterkernel Registry
A.16 Domain-Specific Number-Theory Constraint Validator
A.17 Final Capstone: TYPE → CARRIER → TRANSPORT → DEBT → RESIDUE → COUNTERKERNEL → LIFTBACK → REPLAY/CERT

Chapter 1 — Number Theory as Exposure Geometry

1.1 Introduction: Integers as Distinction-Bearing Carriers

Number theory studies arithmetic distinctions that survive the severe discreteness of the integers. Equations such as ax + by = c, x² + y² = z², x² − Dy² = 1, and xⁿ + yⁿ = zⁿ may have abundant real solutions but few or no integer solutions because divisibility, parity, prime support, and congruence restrict which configurations can exist. The integer carrier Z preserves addition, multiplication, order, and exact equality simultaneously; changing the carrier to Q, R, Z/nZ, Fp, Zp, or an algebraic number ring changes which distinctions remain visible. Number theory therefore begins by specifying not only an equation but its carrier, admissible objects, equivalence relation, and reconstruction target.

Exposure Geometry treats each arithmetic problem as a distinction-bearing frame. Some distinctions are explicit, such as sign, magnitude, prime multiplicity, or residue class. Others become visible only after transport into a more suitable carrier. For example, n = ∏p p^vp(n) converts multiplication into coordinatewise valuation addition: vp(mn) = vp(m) + vp(n). This representation exposes prime depth but suppresses sign unless sign is retained separately. A useful arithmetic representation is therefore never neutral: it preserves some information, compacts other information, and may create a reconstruction obligation.

1.2 Survey: Local Arithmetic, Global Arithmetic, and Reconstruction

The course develops arithmetic through a sequence of carrier changes. Divisibility and Euclid’s algorithm operate directly in Z. Congruences transport integers into quotient rings Z/nZ. Prime-power decomposition separates a global integer into local valuation data. Finite fields remove zero divisors and permit exact polynomial root bounds. The p-adic numbers complete Q with respect to divisibility by one prime. Quadratic forms organize representation problems into equivalence classes. Gaussian integers enlarge the arithmetic carrier so that sums of two squares become norm factorizations. Dirichlet series transport arithmetic functions into analytic objects whose singularities and zeros encode prime distribution.

The central question is not merely whether a transport simplifies the problem, but whether the original claim can be reconstructed afterward. The Chinese remainder theorem gives exact reconstruction from compatible coprime local residues. Hensel’s lemma gives unique lifting of simple roots but not singular roots. A Jacobi symbol compresses prime-local quadratic information but cannot always reconstruct quadratic residuosity. An Euler product reconstructs a Dirichlet series inside its domain of absolute convergence, while analytic continuation requires a different construction. These differences organize the subject around local closure, global compatibility, residue, and liftback.

1.3 Arithmetic EG Frames, Skeletons, Residues, and Successor Seeds

An arithmetic EG frame is the maximal collection of distinctions relevant to a declared claim. It may include the integer carrier, prime support, valuations, signs, residue classes, units, boundary cases, equivalence relations, algorithms, parameters, and quantifier scope. An EG skeleton is a smaller packet sufficient to reconstruct the target sector. For multiplicative arithmetic, the values on prime powers form a skeleton. For gcd computation, the Euclidean quotient sequence is a skeleton. For a residue class modulo a product of coprime moduli, the component residues and CRT idempotents form a skeleton.

A residue is a distinction that the proposed skeleton does not reconstruct. Reducing an integer modulo n leaves an infinite source fibre because every a + kn has the same image. Replacing a quadratic form by its discriminant leaves a class-group residue because inequivalent forms can share the same discriminant. A successor seed is the least additional structure needed to close the failure. One may add a sign coordinate, an ideal class, a derivative valuation, a closure convention, or an error term. The successor frame is not arbitrary enrichment; it is determined by the minimal surviving reconstruction failure.

1.4 Equality, Congruence, Association, Isomorphism, and Equivalence

Arithmetic uses several nonidentical relations. Integer equality means a = b in Z. Congruence a ≡ b mod n means n | a − b and identifies an entire coset a + nZ. Association in an integral domain means a = ub for a unit u; in Z this identifies a and −a, while in Z[i] it identifies four rotations ±a and ±ia. Group or ring isomorphism identifies structures through an operation-preserving bijection. Equivalence of binary quadratic forms identifies forms related by an admissible integral change of variables.

Confusing these relations causes false reconstruction. The equality of gcds does not imply equality of input pairs. Equal Legendre symbols do not imply equal residues. Isomorphic groups need not be equal subgroups of one ambient group. Equivalent forms represent the same integers under the relevant change of variables, but two forms with the same discriminant need not be equivalent. Every theorem must therefore state its comparison relation explicitly. Exposure Geometry treats the relation itself as a load-bearing field because uniqueness is always uniqueness modulo some declared equivalence.

1.5 Arithmetic Relations as Typed Transports

A transport is a map that carries arithmetic objects from one representation to another while declaring what it preserves. Reduction mod n sends Z to Z/nZ and preserves addition and multiplication but not order or magnitude. The valuation vp sends Q× to Z and turns multiplication into addition, but it does not preserve additive structure: vp(x + y) is usually only bounded below by min(vp(x), vp(y)). Prime factorization sends positive integers to finitely supported exponent vectors and preserves multiplication exactly. A Dirichlet generating function sends an arithmetic function a(n) to Σa(n)n^(−s), converting Dirichlet convolution into multiplication where convergence permits.

Typed transport prevents illegal inference. A theorem proved in Fp cannot automatically be lifted to Z. A factorization over C does not imply a factorization over Q. A numerical approximation does not equal an exact algebraic number. Every transport has a kernel, image, admissibility domain, and possible reconstruction map. The arithmetic content lies as much in these interfaces as in the formulas themselves.

1.6 Existence, Construction, Uniqueness, Classification, and Liftback

Existence asserts that at least one object satisfies stated conditions. Construction produces such an object by a specified operation. Uniqueness shows that all admissible objects coincide under the declared equivalence. Classification gives a complete parameter space of equivalence classes. Liftback reconstructs the result in the original problem carrier.

These obligations are independent. Bézout’s theorem states that gcd(a,b) is an integer combination of a and b, while the extended Euclidean algorithm constructs coefficients. The Chinese remainder theorem proves both existence and uniqueness modulo the product and gives an explicit reconstruction formula. The prime number theorem classifies average prime density asymptotically but does not construct the next prime. A primality test may decide whether n is prime without factoring n − 1 or n + 1 completely. A numerical root finder may construct an approximation without proving uniqueness or exact algebraic identity. A complete argument identifies which of these levels it reaches and does not infer the others.

1.7 Local Validity Versus Global Reconstruction

A local statement concerns one prime, one modulus, one completion, one bounded range, or one neighborhood. A global statement concerns the original integer, rational, or algebraic object across all relevant carriers. Local data can determine the global object when a reconstruction theorem exists. The Chinese remainder theorem reconstructs a class modulo mn from classes modulo coprime m and n. Unique factorization reconstructs a positive integer from all valuations vp(n). A rational number is determined by its ordinary value, but finitely many modular reductions do not determine it without a size bound.

Local solvability can also fail to globalize. A polynomial may have roots modulo many primes but no integer root. A quadratic form may represent an integer over every local completion yet fail globally in more general settings. Pairwise compatibility of data need not control higher-order interaction. Exposure Geometry records the first failed reconstruction map rather than treating abundant local success as evidence that the global obstruction is negligible.

1.8 Dyadic Data, Higher-Arity Arithmetic, and Cell Residues

Dyadic arithmetic concerns relations between two objects: divisibility a | b, congruence a ≡ b mod n, coprimality gcd(a,b) = 1, or a group operation on a pair. Many global structures are generated dyadically, but dyadic closure must be proved rather than assumed. Pairwise coprimality is sufficient for standard CRT reconstruction because the ring decomposition theorem supplies the missing global operation. Pairwise linking information in a more general interaction system may not determine the whole configuration.

In number theory, higher-arity residues occur when a relation depends essentially on three or more components. The equation a + b = c is triadic: prime support in c can arise from cancellation between a and b and is not reconstructible from the factorization of either input alone. The discriminant b² − 4ac combines three coefficients. Character orthogonality may involve an entire group of residues. An n-ary cell residue is the part of the full relation not reconstructible from all certified proper-subset packets.

1.9 Counterexamples as Minimal Arithmetic Counterkernels

A counterexample disproves a universal statement; a counterkernel isolates the smallest distinction pattern responsible for the failure. The congruence 2x ≡ 2 mod 6 is a counterkernel to unrestricted modular cancellation because x = 1 and x = 4 both solve it, while canceling 2 at modulus 6 would retain only one class. The number 341 = 11·31 is a counterkernel to the converse of the base-2 Fermat test because 2^340 ≡ 1 mod 341 despite compositeness. A composite modulus with Jacobi symbol +1 but no square root is a counterkernel to treating the Jacobi symbol as a complete residuosity test.

Minimality matters because it identifies the exact missing hypothesis. Removing the nonunit factor restores cancellation. Replacing the composite modulus by a prime restores the field root bound. Adding prime-factor information restores the meaning of a Jacobi symbol. Counterkernels therefore guide repair: they show which carrier, boundary, parameter, or interaction must be added to the theorem.

1.10 Formal Theorems, Algorithms, Computations, and Claim Levels

A formal theorem is a statement proved from declared axioms and definitions. An algorithm is a finite transition rule acting on encoded inputs. A computation is an executed instance producing a record. These are different claim levels. The formal Euclidean algorithm includes a termination proof and correctness theorem. A software implementation additionally requires a representation of signed integers, division semantics, resource bounds, and error handling. A particular run produces a quotient trace and a reported gcd.

Number-theoretic claims remain formal unless execution is part of the predicate. An existence proof for a polynomial-time algorithm does not certify a particular library implementation. A probable-prime result is not a primality proof unless a suitable theorem converts its evidence into a certificate. Floating-point evaluation of an L-function is not exact analytic continuation. The 2026 course therefore separates Φ0 formal mathematics from Φ1 computational realization and records the exact bridge whenever a theorem becomes executable.

1.11 Noncompensatory Proof Obligations

A mathematical proof is conjunctive: one failed mandatory condition cannot be paid for by strength elsewhere. A highly accurate numerical approximation does not replace a missing existence theorem. A correct local proof at every tested prime does not replace a global reconstruction theorem. A strong average estimate does not control a single catastrophic exceptional fibre unless the target is itself an average. A large collection of examples does not prove an unbounded universal statement.

Noncompensation is particularly important when representations are compressed. A short certificate cannot omit ancestry data if that data are required to prevent support reuse. An asymptotic estimate with a constant depending on the modulus is not uniform merely because the rate is otherwise strong. A p-adic lift with increasing precision does not certify a rational solution unless rational reconstruction and verification succeed. The exact failed gate must remain visible.

1.12 The Arithmetic Execution Sequence: Type to Certificate

The governing arithmetic sequence is:

TYPE → CARRIER → TRANSPORT → DEBT → RESIDUE → COUNTERKERNEL → LIFTBACK → REPLAY/CERT.

TYPE specifies the statement, quantifiers, equivalence, and intended result. CARRIER identifies the integers, quotient rings, fields, completions, groups, forms, or analytic spaces involved. TRANSPORT applies factorization, reduction, lifting, convolution, Fourier transform, reduction theory, or analytic continuation. DEBT records assumptions and information lost. RESIDUE is the surviving obstruction. COUNTERKERNEL isolates it minimally. LIFTBACK reconstructs the original claim after repair. REPLAY tests boundary cases, singular cases, hostile parameters, and alternative representations. CERT is emitted only when the original quantified statement has been reconstructed exactly.

This sequence is not an alternative to proof. It is a discipline for locating what a proof must contain and for preventing representation-level success from being mistaken for theorem-level closure.

Chapter 2 — Divisibility, Euclidean Transport, and Prime Skeletons

2.1 Divisibility as a Typed Dyadic Relation

For integers a and b, a divides b, written a | b, when there exists c ∈ Z such that b = ac. The witness c is part of the relation; divisibility is not merely a Boolean comparison. It is reflexive because a = a·1, transitive because b = ac and d = be imply d = a(ce), and antisymmetric on positive integers but not on all integers, since a | −a and −a | a. Association by units is therefore the natural equivalence induced by mutual divisibility in a domain.

Divisibility behaves differently under carrier changes. In a field every nonzero element divides every element, so the relation loses most of its integer content. In Z/nZ, zero divisors create nonunique quotients. In an integral domain, divisibility is compatible with multiplication but can depend on the ambient ring: 5 is irreducible in Z but factors as (2 + i)(2 − i) in Z[i]. The carrier must always accompany the symbol a | b.

2.2 The Division Algorithm as Canonical Reduction Transport

For integers a and b with b > 0, there exist unique integers q and r such that a = bq + r and 0 ≤ r < b. Existence follows by choosing q = floor(a/b); uniqueness follows because if a = bq + r = bq′ + r′ with both remainders in [0,b), then b(q − q′) = r′ − r, whose absolute value is less than b, forcing q = q′ and r = r′.

The map (a,b) ↦ (b,r) is a reduction transport. It preserves the set of common divisors because d | a and d | b exactly when d | b and d | a − bq = r. The quotient q records how the larger carrier was decomposed, while r is the reduced residue. The strict bound r < b guarantees descent. Without a remainder convention, quotient and remainder are not unique, so the boundary condition is load-bearing.

2.3 Euclid’s Algorithm and Descent Through Remainder Carriers

Euclid’s algorithm repeatedly applies the division algorithm:

r_(i−1) = q_i r_i + r_(i+1), with 0 ≤ r_(i+1) < r_i.

Because the nonnegative remainders strictly decrease, the sequence terminates. The last nonzero remainder is gcd(a,b), since each step preserves common divisors. For 252 and 105:

252 = 2·105 + 42,
105 = 2·42 + 21,
42 = 2·21.

Hence gcd(252,105) = 21.

The sequence is a controlled collapse of magnitude while preserving divisibility structure. Consecutive Fibonacci numbers produce the longest quotient chains for inputs of comparable size, giving a logarithmic bound on the number of steps. Thus Euclid’s algorithm is both a proof of existence and an efficient construction.

2.4 Extended Euclid and Exact Bézout Liftback

Each Euclidean remainder is an integer linear combination of the original inputs. Reversing the recurrence therefore yields integers x and y such that gcd(a,b) = ax + by. In the example,

21 = 105 − 2·42
= 105 − 2(252 − 2·105)
= −2·252 + 5·105.

This back-substitution is exact liftback: the reduced terminal gcd is reconstructed in the original carrier using the recorded quotient ancestry.

The coefficients are not unique. If g = ax₀ + by₀, then all solutions to ax + by = g are x = x₀ + (b/g)t and y = y₀ − (a/g)t. Extended Euclid can be implemented iteratively by maintaining coefficient pairs for each remainder. The invariant r_i = a s_i + b t_i supplies a compact correctness certificate throughout the run.

2.5 Linear Diophantine Equations as Reconstruction Problems

The equation ax + by = c has an integer solution exactly when g = gcd(a,b) divides c. Necessity follows because g divides every integer combination ax + by. Sufficiency follows from Bézout: if g = au + bv and c = gk, then c = a(ku) + b(kv).

When solvable, let (x₀,y₀) be one solution. All solutions are

x = x₀ + (b/g)t,
y = y₀ − (a/g)t,
t ∈ Z.

This classification contains existence, construction, and uniqueness modulo a one-dimensional solution lattice. The obstruction is the residue of c modulo g. If c mod g ≠ 0, no search over x and y can repair the failure; the counterkernel is already present in the quotient Z/gZ.

2.6 GCD and LCM as Componentwise Arithmetic Collapse

If a and b are nonzero and a = ∏p p^αp, b = ∏p p^βp, then

gcd(a,b) = ∏p p^min(αp,βp),
lcm(a,b) = ∏p p^max(αp,βp).

The identities follow because a common divisor may use at most the smaller exponent at each prime, while a common multiple must use at least the larger. Consequently,

gcd(a,b)·lcm(a,b) = |ab|,

since min(α,β) + max(α,β) = α + β.

Prime valuation coordinates turn gcd and lcm into lattice meet and join. This is a canonical arithmetic skeleton: multiplicative comparison becomes coordinatewise order. Sign is deliberately excluded and must be restored separately if the original claim distinguishes a from −a.

2.7 Prime Numbers and Irreducible Arithmetic Distinctions

A prime p is a positive integer greater than 1 whose only positive divisors are 1 and p. An irreducible element in a domain is a nonzero nonunit that cannot be factored into two nonunits. In Z, primes and irreducibles coincide, but the distinction matters in more general rings. A prime element satisfies p | ab ⇒ p | a or p | b; this property controls factorization uniqueness.

Every composite n has a prime divisor at most √n. If n = ab with 1 < a ≤ b, then a ≤ √n, and a has a prime divisor. Trial division therefore decides primality after finitely many tests, though its running time is exponential in the bit length log n. Prime status is formal and exact; computational feasibility is a separate question.

2.8 Fundamental Theorem of Arithmetic as Skeleton Reconstruction

Every integer n > 1 can be expressed as a product of primes, and the multiset of primes is unique. Existence follows by strong induction: a composite n = ab factors because both a and b are smaller. Uniqueness follows from Euclid’s lemma. If p₁···pr = q₁···qs, then p₁ divides some qj and hence equals it; cancellation and induction finish the argument.

Thus every positive integer has a unique finitely supported valuation vector (vp(n))p. The integer is reconstructed by n = ∏p p^vp(n). This is an exact skeleton theorem: the prime-power data are neither merely invariants nor approximations. The only omitted distinctions are sign and the exceptional element 0, whose valuation vector is not finite. These require typed boundary handling.

2.9 Prime Support and Valuation Geometry

The support of a nonzero integer n is supp(n) = {p : vp(n) > 0}. Its radical is rad(n) = ∏p|n p, which records support but forgets multiplicity. The valuation vp(n) measures depth at p. Multiplication satisfies vp(mn) = vp(m) + vp(n), while addition satisfies

vp(m + n) ≥ min(vp(m),vp(n)),

with equality when the two valuations differ. When they are equal, cancellation may increase the output valuation. This is the fundamental singular interaction of p-adic arithmetic.

Support and depth must not be conflated. The integers p and p^100 have the same support and radical but radically different valuation depth. Any theorem transported through rad(n) must account for the forgotten exponent data before exact liftback.

2.10 Units, Associates, Sign, and Source-Fibre Ambiguity

The units of Z are ±1. Integers a and b are associates when a = ub for a unit u, so a and −a are associates. Prime factorization is unique only up to order and units. Restricting to positive integers fixes the unit ambiguity and produces a canonical representative.

Maps often enlarge the source fibre. Absolute value sends a and −a to the same positive integer. The ideal map a ↦ (a) also identifies associates. Valuation vectors identify a and −a unless sign is included. If a theorem concerns divisibility only, this compaction may be harmless. If it concerns additive equations, sign is active because a + b and −a + b behave differently. A source-fibre audit asks which original objects share one compact representation and whether the target distinguishes them.

2.11 Euclid’s Lemma as a Transport-Closure Gate

Euclid’s lemma states that if gcd(a,b) = 1 and a | bc, then a | c. Bézout gives ax + by = 1; multiplying by c yields acx + bcy = c. Since a divides both terms on the left when a | bc, it divides c.

The prime form says p | ab implies p | a or p | b. This gate permits divisibility information to pass through products and is the central closure law behind unique factorization. Without primality or coprimality, the inference fails: 6 | 2·3 but 6 divides neither factor. The counterkernel identifies exactly why composite support cannot be assigned to one branch without additional information.

2.12 Prime Gaps and Worst-Fibre Arithmetic

Let p_n denote the nth prime. The gap g_n = p_(n+1) − p_n measures local sparsity. Gaps are unbounded: for any m, the integers

(m + 1)! + 2, …, (m + 1)! + (m + 1)

are all composite because the kth term is divisible by k. This construction gives m consecutive composite integers.

Unbounded gaps coexist with the prime number theorem π(x) ~ x/log x. The asymptotic density describes global average behavior, not a uniform local spacing law. Exposure Geometry separates the aggregate carrier from its worst fibres: a strong average cannot eliminate large exceptional intervals. The target quantifier determines which geometry matters.

2.13 Bit Complexity and Euclidean Proof-Carrying Computation

An integer of magnitude N requires Θ(log N) bits. Euclid’s algorithm uses O(log min(|a|,|b|)) divisions, but bit complexity also counts the cost of dividing large integers. Classical implementations remain polynomial in the input length, while fast multiplication improves asymptotic performance.

A proof-carrying execution records enough data for independent verification. A quotient trace q₁,…,qk allows the remainders to be replayed. Extended coefficients x,y satisfy ax + by = g, and verification requires only multiplication, addition, and checks that g divides both a and b. The record certifies one computation; the general correctness and termination theorems remain formal obligations.

2.14 Divisibility Residues and Minimal Solvability Counterkernels

Divisibility failures are naturally measured by remainders. The equation a | b fails with residue r = b mod |a|. The equation ax + by = c fails with residue c mod gcd(a,b). A claimed cancellation ca | cb ⇒ a | b fails when c carries prime factors needed by the product but absent from either target factor.

Minimal counterkernels expose these residues with the smallest possible data. For example, 2x = 1 has no integer solution because gcd(2,0) = 2 does not divide 1. The congruence 2x ≡ 2 mod 6 exposes nonunit cancellation. Such examples are not peripheral curiosities; they identify the exact boundary between valid and invalid transport.

Part II — Multiplicative Composition and Combinatorial Cells

Chapter 3 — Arithmetic Functions, Convolution, and Carry Geometry

3.1 Multiplicative Functions as Prime-Power Reconstruction Laws

An arithmetic function is a map f: N → C or another declared codomain. It is multiplicative if f(mn) = f(m)f(n) whenever gcd(m,n) = 1, and completely multiplicative if this holds without the coprimality condition. Because every n has a unique prime-power decomposition,

f(n) = ∏p f(p^vp(n))

for every multiplicative f. Thus prime-power values form a complete skeleton.

The coprimality condition is a boundary law. When m and n share support, their divisor structures interact rather than decompose independently. For instance, τ(p²) = 3 but τ(p)² = 4. The failure is not a defect in τ; it records overlap in prime support. Multiplicativity is exact only across independent local components.

3.2 Completely Multiplicative Versus Coprime-Local Composition

For a completely multiplicative function, f(p^k) = f(p)^k, so the prime values alone determine the function. Examples include n ↦ n^s and Dirichlet characters extended by zero on nonunits. For a merely multiplicative function, each prime-power sequence f(p), f(p²),… contains additional local structure.

This distinction controls Euler factors. A completely multiplicative function often yields a geometric local factor:

Σk≥0 f(p^k)p^(−ks) = 1/(1 − f(p)p^(−s)),

when convergent. A general multiplicative function has an arbitrary prime-power generating series. Treating all multiplicative functions as completely multiplicative projects away local depth and creates false identities.

3.3 Divisor Functions and Prime-Support Skeletons

If n = ∏p p^ap, a positive divisor chooses an exponent ep with 0 ≤ ep ≤ ap. Hence

τ(n) = ∏p (ap + 1).

The sum-of-divisors function is

σ(n) = ∏p (1 + p + ··· + p^ap)
= ∏p (p^(ap+1) − 1)/(p − 1).

More generally, σk(n) = Σd|n d^k is multiplicative with local factor 1 + p^k + ··· + p^(ak).

These formulas demonstrate exact reconstruction from independent exponent intervals. The divisor lattice of n is the product of chains {0,…,ap}. Its cardinality and weighted sums factor because the prime coordinates are independent.

3.4 Euler’s Totient Function as Unit-Carrier Measurement

Euler’s totient φ(n) counts units modulo n. For a prime power p^k, the nonunits are precisely the multiples of p, so

φ(p^k) = p^k − p^(k−1) = p^k(1 − 1/p).

CRT implies multiplicativity and therefore

φ(n) = n∏p|n (1 − 1/p).

The ratio φ(n)/n measures the density of invertible residue classes. It depends only on prime support, not valuation depth beyond the factor n itself. Two moduli can have the same totient but different unit-group structures, such as φ(15) = φ(16) = 8. Thus φ is a capacity measure, not a complete classifier.

3.5 The Möbius Function and Inclusion–Exclusion Residue

The Möbius function is defined by μ(1) = 1; μ(n) = 0 if p² | n for some prime p; and μ(n) = (−1)^r when n is a product of r distinct primes. It satisfies

Σd|n μ(d) = 1 if n = 1, and 0 if n > 1.

This is inclusion–exclusion over prime support. Squareful divisors receive zero because repeated inclusion of the same prime does not create a new subset distinction.

The Möbius function is the convolution inverse of the constant-one function. It removes cumulative divisor aggregation and reconstructs primitive contributions. Its cancellation is exact algebraically, although estimating partial sums Σn≤x μ(n) is a deep analytic problem.

3.6 Dirichlet Convolution as Arithmetic Composition

For arithmetic functions f and g, define

(f * g)(n) = Σd|n f(d)g(n/d).

The operation is commutative and associative, with identity ε defined by ε(1)=1 and ε(n)=0 for n>1. If f and g are multiplicative, then f * g is multiplicative. Important identities are

τ = 1 * 1,
σ = id * 1,
φ = id * μ,
μ * 1 = ε.

Convolution composes data across factor pairs. Its local structure follows from distributing the exponent of each prime between d and n/d. It is the arithmetic counterpart of multiplication of generating functions.

3.7 Möbius Inversion as Exact Projection Recovery

Suppose

F(n) = Σd|n f(d) = (f * 1)(n).

Convolving with μ gives f = F * μ, so

f(n) = Σd|n μ(d)F(n/d).

This is Möbius inversion. It reconstructs a primitive function f from its cumulative divisor projection F.

The theorem generalizes inclusion–exclusion and inversion on partially ordered sets. Its validity depends on the finite divisor lattice and the exact incidence relation d | n. If the aggregation is over a different relation, the Möbius function changes. Projection recovery is therefore carrier-specific rather than a universal symbolic trick.

3.8 Binomial Coefficients as Finite Interaction Cells

The binomial coefficient

C(n,k) = n!/[k!(n−k)!]

counts k-element subsets of an n-element set. Integrality follows from counting, independently of cancellation in the factorial expression. The symmetry C(n,k) = C(n,n−k) reflects complement duality. Pascal’s rule

C(n,k) = C(n−1,k−1) + C(n−1,k)

splits subsets according to whether they contain a distinguished element.

The binomial theorem,

(x + y)^n = Σk=0^n C(n,k)x^k y^(n−k),

assembles n independent binary choices into an n-ary interaction polynomial. Coefficients count paths through the choice cell.

3.9 Pascal and Vandermonde Composition Laws

Pascal’s identity is a local recursion. Vandermonde’s identity is a composition law:

C(m+n,r) = Σk C(m,k)C(n,r−k).

It counts r-element subsets of a disjoint union by the number k chosen from the first component. Algebraically, it is the coefficient of x^r in

(1+x)^(m+n) = (1+x)^m(1+x)^n.

The disjointness of the carriers is essential. If the underlying sets overlap, the same element may be selected through two channels and the formula no longer describes the intended objects without an overlap correction.

3.10 Binomial Inversion and Reconstruction from Aggregated Data

If

F(n) = Σk=0^n C(n,k)f(k),

then

f(n) = Σk=0^n (−1)^(n−k)C(n,k)F(k).

The identity follows from the alternating cancellation

Σj=k^n (−1)^(n−j)C(n,j)C(j,k) = δn,k.

Binomial inversion recovers data mixed over subsets of a finite set. It is the Möbius inversion of the Boolean lattice. The alternating signs encode the correction for repeated inclusion. The transform is exact because the underlying incidence matrix is triangular with diagonal entries 1.

3.11 Legendre’s Formula and Factorial Valuation Depth

For a prime p,

vp(n!) = floor(n/p) + floor(n/p²) + floor(n/p³) + ···.

The kth term counts multiples of p^k, adding one valuation layer for each. The sum terminates when p^k > n. Consequently,

vp(C(n,k)) = vp(n!) − vp(k!) − vp((n−k)!).

This formula exposes prime divisibility of binomial coefficients without computing enormous factorials. It also distinguishes support from depth: p may divide C(n,k) once or many times, and each layer has a precise combinatorial source.

3.12 Kummer Carry Geometry

Kummer’s theorem states that vp(C(n,k)) equals the number of carries when k and n−k are added in base p. Equivalently, it is the number of borrows when subtracting k from n in base p.

The theorem converts valuation depth into a positional interaction count. Carries are not attached to either summand independently; they arise from contact between digit columns. A chain of carries can propagate across many positions, so the valuation is a path-dependent residue of addition. This gives a concrete arithmetic model of higher-order transport: local digit constraints accumulate into global prime depth.

3.13 Prime-Power Divisibility and Frobenius Structure

For a prime p and 0 < k < p, p divides C(p,k). Hence

(x + y)^p ≡ x^p + y^p mod p.

In characteristic p, the Frobenius map x ↦ x^p is a ring homomorphism. Iterating gives (x+y)^(p^r) = x^(p^r) + y^(p^r) in every ring of characteristic p.

Over the integers, the omitted binomial terms are not zero; they are multiples of p. Modulo p² and higher powers, their valuations matter. Thus the freshman’s dream is exact only after transport to a characteristic-p carrier.

3.14 Combinatorial Explosion, Multiplicity, and Support Reuse

A locally simple combinatorial rule can generate globally large multiplicity. The number of subsets is 2^n, the number of divisor choices is ∏(ap+1), and the number of paths in a recursive decomposition can grow exponentially. Counting the same support through multiple branches creates duplication debt.

A proof that assigns one unit of error per subset, divisor, or local cell may fail globally even when every local estimate is bounded. The accumulation law must be explicit: residues must telescope, share a bounded owner, occur in finitely many species, or force descent. A constant per component is not a global constant when the number of components is unbounded.

3.15 Average Order Versus Exceptional Arithmetic Fibres

The divisor function τ(n) is irregular pointwise but has average order log n:

Σn≤x τ(n) = x log x + (2γ − 1)x + error.

This follows by counting lattice points ab ≤ x. The average says nothing directly about the maximum of τ(n) on [1,x], which is driven by integers with unusually rich small-prime support.

Exposure Geometry distinguishes the summatory carrier from the pointwise carrier. An average theorem may be exact for aggregate questions but cannot be exported to every n. Quantifier order matters: “for most n” and “for every n” have different residue structures.

3.16 Forgotten-Distinction Ledgers for Combinatorial Projections

Combinatorial formulas often project rich objects to counts. The number C(n,k) forgets which subset was chosen. The divisor count τ(n) forgets divisor values. The radical rad(n) forgets exponents. A generating function coefficient forgets the individual construction paths contributing to it.

A forgotten-distinction ledger records what has been collapsed and whether the target needs it. Counting may suffice for existence through pigeonhole arguments, but not for explicit construction or uniqueness. If two source objects map to the same count yet behave differently under the next operation, the projection cannot serve as the final carrier.

Part III — Quotient Carriers and Local-to-Global Reconstruction

Chapter 4 — Congruences, Units, and Chinese Reconstruction

4.1 Congruences as Quotient Transports

For n ≥ 1, a ≡ b mod n means n | a − b. This is an equivalence relation whose classes are cosets a + nZ. The quotient Z/nZ inherits addition and multiplication because congruence is compatible with both operations.

The projection πn: Z → Z/nZ is a surjective ring homomorphism with kernel nZ. It preserves polynomial expressions but forgets integer magnitude, order, and the exact multiple of n separating two representatives. A congruence proof therefore proves a necessary condition for an integer equation; liftback requires either explicit representatives or a theorem showing the modular condition is sufficient.

4.2 Residue Classes as Compacted Arithmetic Carriers

Every residue class has a least nonnegative representative in {0,…,n−1}, but the class itself contains infinitely many integers. Arithmetic in Z/nZ is closed and finite. Sequences of powers become eventually periodic, and unit powers are purely periodic.

A residue class is a compact carrier designed for divisibility by n. It is exact for predicates invariant under addition of multiples of n. It is inadequate for inequalities, size bounds, or exact factorization unless additional data are supplied. Choosing a representative is a section of the quotient map, not an inverse ring homomorphism.

4.3 Units, Zero Divisors, and the Cancellation Boundary

A residue class a mod n is a unit exactly when gcd(a,n)=1. If ax ≡ ay mod n and a is a unit, multiplication by a^(−1) gives x ≡ y. If a is not a unit, cancellation can fail.

For example, 2x ≡ 2 mod 6 has solutions x ≡ 1 and 4 mod 6. Dividing the equation and modulus by gcd(2,6)=2 gives x ≡ 1 mod 3, which represents both classes modulo 6. The failure is caused by a nontrivial kernel of multiplication by 2. Fields have no such kernel for nonzero multipliers; composite quotient rings may.

4.4 Linear Congruences and Solution-Fibre Multiplicity

The congruence ax ≡ b mod n is equivalent to ax + ny = b. Let g = gcd(a,n). It has a solution exactly when g | b. If solvable, divide by g:

(a/g)x ≡ b/g mod n/g.

Now a/g is a unit modulo n/g, giving one solution modulo n/g and exactly g solutions modulo n. The solution fibre size equals the kernel size of multiplication by a on Z/nZ.

Thus existence, representative construction, and multiplicity are all determined by the gcd. Illegal cancellation loses the fibre structure and may produce a false uniqueness claim.

4.5 Fermat’s Theorem as Finite-Field Transport Closure

For prime p,

a^p ≡ a mod p

for every integer a, and a^(p−1) ≡ 1 mod p when p ∤ a. One proof observes that multiplication by a permutes the nonzero residue classes. Therefore

a^(p−1)(p−1)! ≡ (p−1)! mod p,

and cancellation gives the result.

The theorem is a closure property of Fp×, a group of order p−1. It produces a fast compositeness falsifier: if a^(n−1) ≠ 1 mod n for some coprime a, then n is composite. The converse fails because composite pseudoprimes can satisfy the congruence.

4.6 Euler’s Theorem and Unit-Group Exponents

If gcd(a,n)=1, then

a^φ(n) ≡ 1 mod n.

The units form a finite group of order φ(n), so the result follows from Lagrange’s theorem. The actual order ordn(a) divides φ(n), and the least universal exponent is the Carmichael function λ(n), the exponent of the unit group.

Using φ(n) is sufficient but not always minimal. For n=15, φ(15)=8, while every unit satisfies a^4 ≡ 1. Distinguishing group order, element order, and group exponent prevents unnecessary or incorrect exponent reductions.

4.7 Wilson’s Theorem and Inverse-Pair Reconstruction

An integer p > 1 is prime exactly when

(p−1)! ≡ −1 mod p.

For prime p, pair each nonzero residue with its inverse. All pairs multiply to 1 except self-inverse elements, which solve x² = 1 and hence are ±1. Their product is −1.

Wilson’s theorem is a complete characterization but an inefficient primality test if implemented by computing the whole factorial. It illustrates the distinction between logical closure and computational feasibility. A theorem can be exact yet unsuitable as a large-input algorithm.

4.8 The Chinese Remainder Theorem as Exact Local Liftback

If n₁,…,nr are pairwise coprime and N = n₁···nr, then

Z/NZ ≅ Z/n₁Z × ··· × Z/nrZ.

Given residues ai, let Ni = N/ni and choose Mi with NiMi ≡ 1 mod ni. Then

x = Σi aiNiMi mod N

is the unique solution to x ≡ ai mod ni.

This is exact local-to-global reconstruction. The local carriers are independent because their moduli are coprime, and the idempotent coefficients isolate each component. The theorem provides existence, construction, uniqueness, and liftback in one packet.

4.9 CRT Idempotents and Independent Local Components

Define ei = NiMi mod N. Then

ei ≡ 1 mod ni,
ei ≡ 0 mod nj for j ≠ i.

Consequently ei² = ei, eiej = 0 for i ≠ j, and Σei = 1. Every x decomposes uniquely as x = Σxiei.

These orthogonal idempotents are the structural reason CRT works. They split the global ring into independent local sectors. Algorithms can compute modulo prime powers separately and recombine, but the factorization of the modulus must be known and the component results must be protected against faults.

4.10 Overlapping Moduli and Boundary Compatibility Residue

For noncoprime moduli m and n, the system

x ≡ a mod m,
x ≡ b mod n

is solvable exactly when a ≡ b mod gcd(m,n). When solvable, the solution is unique modulo lcm(m,n), not mn.

The common divisor is the overlap boundary. The two local specifications must agree on it. The residue a − b mod gcd(m,n) is the exact compatibility obstruction. Pairwise coprimality removes this boundary and yields automatic compatibility.

4.11 Euler’s Totient Function and Local Unit Capacity

The formula

φ(n) = n∏p|n (1 − 1/p)

measures how much of Z/nZ supports invertible multiplication. For prime n, all nonzero classes are units. For highly composite n with many small prime factors, the unit density is much lower.

The identity Σd|n φ(d) = n partitions integers 1,…,n by gcd with n. Möbius inversion gives φ(n) = Σd|n μ(d)n/d. These are two reconstructions of the same capacity: one through gcd fibres, the other through divisor inversion.

4.12 Carmichael Numbers and Nominal-Theorem Instability

A Carmichael number is a composite n satisfying

a^(n−1) ≡ 1 mod n

for every gcd(a,n)=1. Korselt’s criterion says that n is Carmichael exactly when n is squarefree and p−1 | n−1 for every prime p | n. The smallest is 561 = 3·11·17.

Carmichael numbers show that passing every Fermat test on units does not reconstruct primality. The tested property belongs to the unit-group exponent and can occur in composite CRT products. A nominal prime signature is therefore not a complete prime classifier.

4.13 Fermat Pseudoprimes as Adversarial Counterkernels

A composite n is a Fermat pseudoprime to base a if gcd(a,n)=1 and a^(n−1) ≡ 1 mod n. The number 341 = 11·31 is a base-2 pseudoprime. Such examples are minimal hostile inputs for a Fermat primality test.

A test that rejects whenever the congruence fails is sound as a compositeness witness. A test that accepts whenever it passes is unsound. The repair is to use stronger conditions, multiple bases, Miller–Rabin structure, or an actual primality certificate.

4.14 Projection Loss Under Reduction Modulo n

Reduction modulo n forgets all multiples of n. Different integers with distinct signs, magnitudes, and prime supports may share one residue. For example, 1, 1+n, and 1+kn are identical modulo n but can have unrelated factorizations.

A modular obstruction proves nonexistence when no residue solution exists. A modular solution does not generally prove an integer solution. Exact liftback may require a size interval shorter than n, rational reconstruction bounds, Hensel compatibility, or a direct substitution check.

4.15 Local Congruence Data Versus Global Integer Ancestry

Knowing x modulo several moduli determines x modulo their least common multiple, not necessarily as an integer. If |x| is known to be less than half the modulus, a balanced representative reconstructs x uniquely. Without a size bound, infinitely many ancestors remain.

This principle is used in computer algebra: calculate an integer or rational object modulo many primes, combine by CRT, and reconstruct once the modulus exceeds a proven bound. The bound is part of the proof. Premature reconstruction can return a plausible but incorrect ancestor.

Chapter 5 — Computational Number Theory, Certificates, and Cryptography

5.1 Numerical Calculation and Finite Encoding

Computational number theory represents integers as finite bit strings. Input size is measured by bit length, not numerical magnitude. An algorithm taking O(N) steps for an integer N is exponential in log N, while O((log N)^k) is polynomial.

Exact arithmetic avoids rounding but still has resource costs. Addition of b-bit integers is linear in b; multiplication and division depend on the chosen algorithms. Complexity analysis must include operand growth, memory, and conversion between representations. A mathematically finite procedure may be computationally impractical.

5.2 Binary Exponentiation as Compressed Transport

To compute a^m mod n, write m in binary and repeatedly square:

m = Σj εj2^j,
a^m = ∏j (a^(2^j))^εj.

Only O(log m) modular multiplications are needed. Intermediate results are reduced modulo n, preventing exponential growth in represented magnitude.

Correctness follows from the binary decomposition of the exponent. The algorithm is deterministic and exact, but implementation security may require constant-time execution so that the bit pattern of a secret exponent is not leaked through branches or timing.

5.3 Modular Inversion and Executable Bézout Certificates

An inverse of a modulo n exists exactly when gcd(a,n)=1. Extended Euclid constructs x,y with ax + ny = 1, so x mod n is the inverse. Verification requires checking ax ≡ 1 mod n.

The coefficient pair is a compact certificate. It is stronger than a numerical claim because it directly witnesses the defining relation. For large systems, certificate-based computation separates expensive construction from cheap independent verification.

5.4 Computational Complexity in Bit Carriers

Algorithmic complexity must be expressed relative to encoded inputs. Trial division through √N uses exponentially many tests in log N. Euclid’s algorithm is polynomial. General integer factorization is subexponential with the number field sieve but is not known to be polynomial classically. Primality is decidable in deterministic polynomial time.

Complexity also depends on output size. Listing all divisors can require superpolynomial output, so no algorithm can print them faster than the output length. A complexity claim must specify the task: decision, search, counting, enumeration, or certificate verification.

5.5 Probable Primes and One-Sided Falsifiers

A Fermat test computes a^(n−1) mod n. Failure proves compositeness; success gives only probable-prime evidence. Strong probable-prime tests examine the repeated-square chain from n−1 = 2^s d with d odd.

A one-sided test has a trustworthy rejection direction. Its acceptance direction is probabilistic or conditional. The distinction must be carried into the output type: COMPOSITE_WITH_WITNESS differs from PROBABLE_PRIME and PROVED_PRIME.

5.6 Miller–Rabin and Structured Adversarial Replay

For odd n, write n−1 = 2^s d with d odd. A base a passes if a^d ≡ 1 mod n or a^(2^r d) ≡ −1 mod n for some 0 ≤ r < s. Otherwise a is a witness to compositeness.

For every odd composite n, at least three quarters of possible bases are witnesses. Repeating independent tests makes error probability exponentially small. For bounded machine-size ranges, fixed base sets can make the procedure deterministic. The theorem supporting the base set is part of the implementation scope.

5.7 Deterministic and Certificate-Based Primality

A primality certificate supplies data from which primality follows by a theorem. Pocklington-style certificates use a sufficiently factored part of n−1. Pratt certificates recursively certify factors of p−1. ECPP uses elliptic curves and can produce compact certificates efficiently in practice. AKS proves primality in deterministic polynomial time without unproved assumptions.

The construction cost and verification cost differ. A good certificate allows a small trusted verifier to replay the proof independently. This is preferable to trusting a large search implementation.

5.8 Trial Division and Fermat Factorization

Trial division tests possible prime divisors up to √N. It is effective when N has a small factor but scales poorly. Fermat factorization writes

N = x² − y² = (x−y)(x+y).

Starting with x = ceil(√N), one searches for x² − N to become a square. The method is efficient when the factors are close. It performs poorly when they are highly unbalanced.

Each method exposes a different geometry: trial division searches support directly; Fermat searches near-square additive structure.

5.9 Pollard Rho and Collision Geometry

Pollard rho iterates a map such as f(x)=x²+c mod N. Modulo a hidden prime factor p, the sequence eventually cycles. A collision xi ≡ xj mod p may occur before equality modulo N, and then gcd(|xi−xj|,N) reveals p.

The birthday paradox suggests roughly √p steps to find a factor p. The algorithm can fail with gcd 1 or N and then restarts with different parameters. Its success is probabilistic but its returned factor is verified exactly by division.

5.10 Pollard p−1, ECM, Quadratic Sieve, and Number Field Sieve

Pollard p−1 succeeds when a factor p has p−1 composed mostly of small primes. Elliptic-curve factorization replaces p−1 by random elliptic-curve group orders, improving the chance that one hidden factor has smooth group order. The quadratic sieve and number field sieve collect relations whose exponent vectors are linearly dependent mod 2, producing congruent squares and a nontrivial gcd.

These algorithms are not interchangeable implementations of one idea. Each depends on a different carrier: multiplicative groups, elliptic curves, quadratic residues, or algebraic norms. Their performance varies with factor size and total input size.

5.11 Factor Discovery Versus Complete Factorization Liftback

Finding one nontrivial factor d of N produces N = d(N/d), but neither factor is necessarily prime. Complete factorization recursively splits composite factors and proves the terminal factors prime.

A certified factorization packet contains prime factors pi, exponents ei, primality certificates, and a product check N = ∏pi^ei. Probable-prime labels are insufficient if exact factorization is claimed. Multiplicity must also be verified; listing distinct primes without exponents loses valuation depth.

5.12 RSA as Modular Transport and CRT Reconstruction

RSA chooses primes p,q, sets n=pq, and selects e,d with ed ≡ 1 mod λ(n) or φ(n). Encryption is c ≡ m^e mod n; decryption is m ≡ c^d mod n.

Correctness is proved separately modulo p and q and recombined by CRT. If m is divisible by a prime factor, the zero case is handled directly; otherwise Fermat’s theorem applies. Textbook RSA is deterministic and malleable, so practical security requires standardized padding and encoding. The number-theoretic permutation theorem alone does not provide semantic security.

5.13 Cryptographic Support Duplication and Shared-Prime Ruin

If two RSA moduli N₁ = pq and N₂ = pr share a prime, then gcd(N₁,N₂)=p reveals both factorizations. A hard individual problem collapses because secret support was reused across carriers.

This is a one-use accounting failure. Random prime generation must avoid shared factors, and large collections of public keys can be batch-audited by product trees and gcd techniques. Local key validity does not protect against global cross-key interaction.

5.14 Nonce Reuse, One-Use Capacity, and Liability Ownership

Many signature systems require a fresh secret nonce for each message. Reusing it can create two equations with the same unknown support, allowing elimination and private-key recovery. Predictable or biased nonces can produce similar leakage through lattice methods.

Randomness is therefore a consumable resource with one-use semantics. The proof of the signature scheme assumes a distribution; the implementation must realize it. Liability belongs to the randomness generator and state-management boundary, not to the abstract group law.

5.15 Timing, Fault, and Side-Channel Residues

An implementation may leak secrets through execution time, memory access, power use, electromagnetic emission, or induced faults. CRT-accelerated RSA is vulnerable if one branch is corrupted and the faulty output is released; comparing correct and faulty signatures can expose a factor.

Constant-time algorithms, blinding, redundant verification, and fault detection repair these channels. The formal arithmetic remains correct, but physical execution introduces observables absent from the theorem carrier. Security requires a higher claim level than functional correctness.

5.16 Mathematical Correctness Versus Implementation Embodiment

A cryptographic theorem may prove that decryption inverts encryption under exact arithmetic and secret-key assumptions. A program additionally chooses data types, random sources, branch structure, error behavior, and memory management. A deployed device introduces hardware, timing, faults, and attackers.

These layers form an antichain of certificates. Correct algebra does not certify code; correct code does not certify a device; successful tests do not certify all hostile environments. Each promotion requires its own realization and replay packet.

5.17 Classical Cryptography Under Quantum Factorization Exposure

Shor’s algorithm solves integer factorization and discrete logarithms in polynomial time on a sufficiently capable fault-tolerant quantum computer. This changes the security assumption of RSA, finite-field Diffie–Hellman, and elliptic-curve cryptography, not their mathematical correctness.

Post-quantum systems use different hardness carriers, including lattices, codes, and hash structures. Migration requires more than replacing one primitive: key sizes, protocol messages, implementation attacks, and long-lived encrypted data must be retyped. The timeline is driven by viable hardware and deployment constraints, not by the formal existence of the quantum algorithm alone.

5.18 Reproducible Computational Records and Independent Replay

A reproducible number-theory computation records inputs, software version, algorithm, parameters, random seed where applicable, precision, output, and certificate. Independent replay may use a different implementation but must verify the same mathematical predicate.

For primality, the record should include a witness or certificate. For factorization, it includes factors and primality proofs. For p-adic computation, it includes precision and valuation bounds. For L-functions, it includes truncation and error control. A bare decimal or “passed” status is not an independently checkable result.

Part IV — Prime-Local Geometry and p-adic Reconstruction

Chapter 6 — Polynomial Congruences, Finite Fields, and p-adic Numbers

6.1 Prime-Valuation Carriers and Ultrametric Distinctions

For nonzero rational x, write x = p^v a/b with p ∤ ab. Define vp(x)=v and |x|p=p^(−v). Then

|xy|p = |x|p|y|p,
|x+y|p ≤ max(|x|p,|y|p).

Numbers are p-adically close when their difference is divisible by a high power of p. The metric emphasizes prime depth rather than ordinary size. For example, p^n → 0 p-adically although it grows without bound ordinarily.

6.2 p-adic Completion as Successor-Frame Construction

Q is not complete under |·|p. Completing it produces Qp; the closed unit ball is Zp. Every p-adic integer has an expansion

a₀ + a₁p + a₂p² + ···, with 0 ≤ ai < p.

The series converges because p^n → 0. Completion adds limits of compatible residue approximations. It is a successor carrier forced by Cauchy sequences that have no rational limit under the p-adic metric.

6.3 Compatible Residue Towers and Source Ancestry

A p-adic integer corresponds to a compatible sequence

x₁ mod p, x₂ mod p², …,

with x_(k+1) ≡ x_k mod p^k. Thus

Zp ≅ inverse limit of Z/p^kZ.

Compatibility preserves ancestry between precision levels. An arbitrary root modulo each p^k does not define one p-adic root unless the roots form a compatible chain. Branch selection is therefore part of the data.

6.4 Hensel’s Lemma as Local Liftback

If f(a) ≡ 0 mod p and f′(a) not ≡ 0 mod p, then a lifts uniquely to a root in Zp. More generally, if f(a_k) ≡ 0 mod p^k, choose

a_(k+1) = a_k + t p^k,

where t solves

f(a_k)/p^k + t f′(a_k) ≡ 0 mod p.

Derivative invertibility gives a unique t. Hensel lifting reconstructs increasingly precise roots from one simple residue root.

6.5 Newton Transport and Precision Amplification

The p-adic Newton step is

a′ = a − f(a)/f′(a).

When f′(a) is a p-adic unit and a is sufficiently accurate, the valuation of f(a′) is roughly doubled. This permits quadratic precision growth.

Division by f′(a) is legal only if it is invertible at the working precision. Precision tracking must account for any valuation in the denominator. A symbolic Newton formula without this audit may conceal precision loss or nonexistence.

6.6 Simple Roots, Singular Roots, and Boundary Fracture

A root a mod p is simple when f′(a) not ≡ 0 mod p and singular when f′(a) ≡ 0 mod p. Simple roots have stable unique lifts. Singular roots lie on a branching boundary: they may fail to lift, split into many lifts, or require higher-order conditions.

For f(x)=x², the root 0 mod p is singular. Modulo p², every multiple of p is a root, so uniqueness fails maximally. The derivative gate separates stable local geometry from singular fibres.

6.7 Nonunique Lifts and Singular-Fibre Residues

Consider solving f(a + tp^k) ≡ 0 mod p^(k+1). If both f(a)/p^k and f′(a) vanish mod p, every t may lift; if the constant term is nonzero but the derivative vanishes, no t lifts. Intermediate higher-order behavior can produce partial branching.

The residue is not merely “derivative zero.” It is the unresolved higher-order congruence determining branch count. A complete singular Hensel analysis records valuations of f and its derivatives and identifies the first nonzero term in the Taylor expansion.

6.8 Congruences Modulo a Prime

Modulo a prime p, Z/pZ = Fp is a field. Every nonzero class has an inverse, polynomial division works, and a nonzero polynomial of degree d has at most d roots.

Fermat’s identity gives x^p = x for every x ∈ Fp, so

x^p − x = ∏a∈Fp (x−a).

The derivative is −1, so all roots are simple. Finite-field structure is therefore much more rigid than arithmetic modulo composite numbers.

6.9 Finite Fields as Closed Arithmetic Carriers

Every finite field has q=p^n elements. Conversely, for each prime power q there is a field Fq, unique up to isomorphism. One construction is Fp[x]/(g), where g is irreducible of degree n.

Its multiplicative group Fq× is cyclic of order q−1. Every element satisfies x^q=x. The field is closed under its arithmetic operations, but representations depend on a chosen basis or irreducible polynomial. Isomorphic implementations need not use identical coordinates.

6.10 Frobenius Transport and Characteristic-p Structure

The Frobenius map F(x)=x^p is a field automorphism of F_(p^n). Its nth power is the identity, and its fixed field is Fp. The Galois group is cyclic, generated by Frobenius.

Trace and norm are

Tr(x)=x+x^p+···+x^(p^(n−1)),
N(x)=x^((p^n−1)/(p−1)).

These maps transport extension elements into the base field while forgetting orbit position. The full Frobenius orbit is needed to reconstruct the minimal polynomial.

6.11 Polynomial Interpolation and Exact Reconstruction

Given distinct x₁,…,xr in a field and prescribed values y₁,…,yr, the unique polynomial of degree less than r satisfying P(xi)=yi is

P(x)=Σi yi ∏j≠i (x−xj)/(xi−xj).

Every denominator is invertible because the points are distinct. This is Lagrange interpolation.

Over Fp, every function Fp → Fp is represented by a polynomial of degree at most p−1. Higher-degree polynomials may define the same function because x^p−x vanishes everywhere. Function equality and polynomial equality must therefore be separated.

6.12 Chevalley–Warning and Variable-Capacity Geometry

If polynomials f₁,…,fr over Fq in n variables satisfy

Σdeg(fi) < n,

then the number of common zeros is divisible by the characteristic p. For q=p, use the indicator

I(x)=∏i [1−fi(x)^(p−1)].

Summing over Fp^n and expanding shows that every nonconstant monomial vanishes modulo p because its total degree is too small for every variable exponent to be a positive multiple of p−1.

The theorem expresses a capacity inequality: sufficiently many variables force arithmetic cancellation and hence nontrivial solutions for homogeneous systems.

6.13 Primitive Roots and Cyclic Multiplicative Skeletons

A primitive root modulo p is a generator of Fp×. Every finite subgroup of the multiplicative group of a field is cyclic. If g generates Fp×, every nonzero residue is g^k.

To test whether g is primitive, factor p−1 and verify

g^((p−1)/q) not ≡ 1 mod p

for every prime q | p−1. The number of primitive roots is φ(p−1). Discrete logarithms convert multiplicative equations into linear congruences in exponents, although computing them may be difficult.

6.14 Primitive Roots for Prime Powers

For an odd prime p, a primitive root modulo p can be chosen to remain primitive modulo p²; once primitive modulo p², it is primitive modulo every p^k. The unit group modulo p^k is cyclic of order p^(k−1)(p−1).

Primitive roots exist exactly for n = 1,2,4,p^k,2p^k with p odd prime. The obstruction for 2^k, k≥3, is that the unit group is C₂ × C_(2^(k−2)), not cyclic. CRT products with two odd prime factors also fail cyclicity because component orders share a factor 2.

6.15 Classification of Moduli with Primitive Roots

The primitive-root classification combines local unit-group structure and product composition. Each odd prime-power unit group is cyclic. Multiplication by 2 preserves the odd unit group. Other moduli decompose into products whose cyclic components have incompatible orders.

This is a global classification theorem, not a pointwise search. Producing a large-order element for one modulus does not prove cyclicity without showing its order equals φ(n). The factorization of φ(n) supplies the verification skeleton.

6.16 Quadratic Equations Modulo p

For odd prime p and a not ≡ 0 mod p,

ax² + bx + c ≡ 0 mod p

is equivalent to

(2ax+b)² ≡ Δ = b²−4ac mod p.

If Δ is a nonzero square, there are two roots; if Δ=0, one repeated root; if Δ is a nonsquare, no roots. The formula is

x ≡ (−b ± √Δ)(2a)^(−1) mod p.

The discriminant is a complete root-count invariant for quadratic polynomials over fields of odd characteristic.

6.17 Root Multiplicity and Derivative-Based Separation

A root α of f over a field is repeated exactly when f(α)=f′(α)=0. The polynomial gcd(f,f′) contains all repeated factors. A polynomial is squarefree when this gcd is 1.

In characteristic p, f′ may vanish identically for nonconstant f, as with f(x)=g(x^p). Then every exponent is divisible by p, and Frobenius extraction is needed. Derivative-based separation must therefore include the characteristic-p boundary case.

6.18 Root Bounds Over Fields and Failure Over Composite Rings

A nonzero degree-d polynomial over a field has at most d roots. The proof factors out x−α for each root and uses the absence of zero divisors.

Over Z/8Z, x²−1 has four roots: 1,3,5,7. The factorization (x−1)(x+1)=0 does not force either factor to vanish because zero divisors exist. Root bounds are carrier-dependent theorem statements, not purely formal properties of polynomial notation.

6.19 Local Solvability Versus Global Solvability

A global integer or rational solution induces solutions modulo every prime power and over every completion. Failure locally therefore obstructs global existence. The converse requires a local–global theorem and is not universal.

For linear equations and nondegenerate quadratic forms under appropriate hypotheses, strong local–global principles exist. For higher-degree equations, local points can coexist with no rational point. The surviving obstruction belongs to the gluing of local data, not to any isolated local carrier.

6.20 p-adic Precision, Truncation, and Resource Ledgers

A computed p-adic number is represented modulo p^N, often with a valuation offset. Operations change precision. Addition may gain apparent zeros through cancellation but cannot claim them without enough input precision. Division by p^k loses k digits. Newton iteration may gain digits quadratically under a unit-derivative condition.

A precision ledger records absolute precision, relative precision, valuation bounds, and whether digits are certified or provisional. Printing a long p-adic expansion without these fields creates ghost accuracy.

Part V — Algebraic Carriers, Orbits, and Quotients

Chapter 7 — Groups, Rings, Fields, and Ideal Reconstruction

7.1 Groups as Arithmetic Transport Systems

A group G has an associative operation, identity, and inverses. Additive groups such as Z/nZ model repeated addition; multiplicative groups such as Fp× model invertible multiplication. The order of g is the least positive m with g^m=e.

Groups encode transport by reversible operations. Arithmetic statements about powers become orbit statements under repeated multiplication. Fermat’s and Euler’s theorems are consequences of finite group order, while primitive roots are generators whose orbits cover the full carrier.

7.2 Element Order and Cyclic Reconstruction

If G = ⟨g⟩ has order n, then g^k has order n/gcd(n,k). For each divisor d | n, a cyclic group has φ(d) elements of order d.

An element generates G exactly when its order is n. Testing a proposed generator therefore requires enough factorization of n to exclude every proper divisor. Observing many distinct powers is not a proof unless the orbit is shown complete.

7.3 Homomorphisms, Kernels, Images, and Quotients

A homomorphism φ:G→H preserves the group operation. Its kernel contains distinctions collapsed to the identity; its image is the realized target subgroup. The first isomorphism theorem states

G/ker φ ≅ im φ.

The quotient is the exact carrier obtained after identifying elements differing by kernel data. Reconstruction of a source element from its image is possible only modulo the kernel unless a section or additional coordinate is supplied.

7.4 Group Actions, Orbits, and Stabilizers

A group action sends (g,x) to g·x compatibly with multiplication. The orbit Orb(x) contains reachable states; the stabilizer Stab(x) contains operations fixing x. For finite groups,

|Orb(x)| = |G|/|Stab(x)|.

Arithmetic applications include multiplication actions on residues, equivalence actions on forms, and Galois actions on roots. Orbit classification is stronger than invariant computation because an invariant may be constant on several orbits.

7.5 Invariant Equality Versus Orbit Equality

An invariant I satisfies I(g·x)=I(x). Thus different invariant values prove different orbits. Equal values do not prove a common orbit unless the invariant is complete.

Binary quadratic forms with the same discriminant can represent different classes. Group elements with the same order can lie in different conjugacy classes. Polynomials can share discriminant but have different Galois groups. Exposure Geometry treats invariant collisions as source fibres requiring a splitter or a scoped nonuniqueness statement.

7.6 Products of Groups and CRT Composition

The direct product G×H operates componentwise. The order of (g,h) is lcm(ord(g),ord(h)). Therefore Cm×Cn is cyclic exactly when gcd(m,n)=1.

CRT gives

(Z/mnZ)× ≅ (Z/mZ)× × (Z/nZ)×

for coprime m,n. Local group structures reconstruct the global unit group, but cyclicity depends on interaction between component orders. Componentwise cyclicity alone is insufficient.

7.7 Diagonal Coupling and Failure of Componentwise Reconstruction

A subgroup of G×H need not be A×B. The diagonal {(g,g)} couples the coordinates and projects surjectively to each component while remaining smaller than the full product. Thus knowing both projections does not reconstruct the subgroup.

Arithmetic equations can similarly couple local coordinates. Independent local carriers may be available, but a shared global relation selects a diagonal-like subset. Reconstruction needs the compatibility law, not only component descriptions.

7.8 Rings as Multi-Operation Arithmetic Carriers

A commutative ring supports addition, subtraction, and multiplication. An integral domain has no zero divisors; a field makes every nonzero element invertible. Z is a domain but not a field. Z/nZ is a field exactly when n is prime.

The interaction of two operations creates arithmetic richness. Additive structure alone is simple; multiplication introduces primes, ideals, factorization, and zero divisors. Carrier typing must state which ring properties a proof uses.

7.9 Units, Zero Divisors, Irreducibles, and Prime Elements

A unit has a multiplicative inverse. A zero divisor annihilates a nonzero element. An irreducible nonunit cannot be decomposed into nonunits. A prime element divides one factor whenever it divides a product.

Prime implies irreducible in an integral domain, but the converse may fail. In Z[√−5], 2,3,1±√−5 participate in nonunique factorizations, and irreducibility does not guarantee prime behavior. Proofs importing Euclid’s lemma require a prime or UFD hypothesis.

7.10 Ideals as Canonical Relation Carriers

An ideal I is an additive subgroup closed under multiplication by ring elements. The ideal generated by a₁,…,ar is

(a₁,…,ar) = {r₁a₁+···+rrar}.

In Z, (a,b)=gcd(a,b)Z. Ideals collect all consequences of linear combination and are therefore natural carriers for divisibility and congruence. Quotienting by I identifies elements whose difference lies in I.

7.11 Prime and Maximal Ideals as Quotient Boundaries

An ideal P is prime when R/P is a domain; M is maximal when R/M is a field. In Z, pZ is prime and maximal exactly when p is prime.

These definitions shift element properties into quotient-carrier properties. A prime ideal is a boundary across which multiplication remains nondegenerate. A maximal ideal yields a fully invertible nonzero residue system.

7.12 Euclidean Domains, PIDs, and UFDs

A Euclidean domain has a norm supporting division with remainder. Every Euclidean domain is a PID, every PID is a UFD, and every UFD is an integral domain with unique element factorization.

The implications are not reversible in general. Euclidean structure gives an algorithm; PID structure gives principal ideals; UFD structure gives factorization uniqueness. A theorem should use the weakest sufficient level rather than importing unnecessary structure.

7.13 Failure of Element Factorization and Ideal-Level Repair

In rings of algebraic integers, element factorization may fail uniquely while nonzero ideals still factor uniquely into prime ideals. The class group measures the failure of ideals to be principal.

This is carrier mutation forced by residue. The element carrier cannot support unique decomposition; the ideal carrier retains enough structure. Exact liftback to elements is possible only for principal ideal products and therefore carries a class-group obstruction.

7.14 Fields and Finite Extension Carriers

A field extension K/F is a vector space over F with compatible multiplication. Algebraic elements satisfy polynomials over F. The degree [K:F] measures vector-space dimension.

Finite fields, number fields, and p-adic extensions are all extension carriers but have different topology and arithmetic. Minimal polynomials encode one element relative to the base field, not an absolute identity independent of the embedding.

7.15 Frobenius Orbits, Trace, and Norm

In F_(p^n), Frobenius conjugates are x,x^p,…,x^(p^(n−1)). Trace is their sum and norm their product. These lie in Fp and are invariant under the orbit.

Equal trace and norm do not always determine the element when n is large. They are compressed invariants. The minimal polynomial or full orbit may be required for reconstruction.

7.16 Finite Abelian Group Decomposition

Every finite abelian group decomposes uniquely up to isomorphism into cyclic prime-power groups or invariant factors:

G ≅ C_(n₁) × ··· × C_(nr), with n₁ | ··· | nr.

This theorem classifies the abstract group but not a particular embedding or chosen generators. Computing the decomposition of a concrete arithmetic group may require Smith normal form, factorization, or relation matrices.

7.17 Proof-Carrying Algebraic Computation

Computing a group order, Smith normal form, factorization, or ideal class should produce verifiable witnesses. A Smith decomposition supplies unimodular matrices U,V with UAV=D. A polynomial factorization is checked by multiplication and irreducibility certificates. A group relation computation requires generator and relation validation.

The certificate should be smaller or easier to verify than the search. Otherwise the result remains dependent on the implementation rather than the mathematical kernel.

7.18 Source Fibres Under Quotient and Invariant Maps

Every quotient or invariant map creates fibres. The residue map Z→Z/nZ has infinite fibres. The norm map can have many algebraic preimages. The discriminant map groups multiple forms and fields. The trace map collapses Frobenius orbits.

A source-fibre theorem states exactly when the fibre is one equivalence class. Without it, a compact packet cannot support uniqueness. Identifying a fibre’s internal geometry is often the next arithmetic problem.

Part VI — Reciprocity as Path-Comparison Geometry

Chapter 8 — Quadratic Reciprocity, Characters, and Gauss Transport

8.1 Quadratic Residues and the Squaring Projection

For odd prime p, a nonzero a is a quadratic residue if a ≡ x² mod p. The squaring map Fp×→Fp× has kernel {±1}, so its image has (p−1)/2 elements. Every nonzero square has exactly two roots.

The projection loses the sign of the root. Residues and nonresidues form the two cosets of the square subgroup. This index-two structure is encoded by the quadratic character.

8.2 The Legendre Symbol as a Local Exposure Signature

Define

(a/p)=0 if p|a,
(a/p)=1 if a is a nonzero square mod p,
(a/p)=−1 otherwise.

The symbol is multiplicative in a. It records one bit of local solvability for x² ≡ a mod p. It does not record the square roots or distinguish residues within the same class. It is therefore a signature, not a full reconstruction packet.

8.3 Euler’s Criterion and Character Evaluation

For p ∤ a,

a^((p−1)/2) ≡ (a/p) mod p.

If g is a primitive root and a=g^k, then g^((p−1)/2)=−1, so the power equals (−1)^k. This proves the criterion and multiplicativity.

Euler’s criterion gives an efficient symbol computation by modular exponentiation. The output ±1 is exact, while extraction of a square root is a separate algorithmic problem.

8.4 Calculation of the Legendre Symbol

To compute (a/p), reduce a modulo p, factor it when useful, apply multiplicativity, and use the supplementary laws:

(−1/p)= (−1)^((p−1)/2),
(2/p)= (−1)^((p²−1)/8).

Quadratic reciprocity exchanges numerator and denominator, producing a Euclidean-style descent. Efficient symbolic calculation avoids constructing square roots or enumerating all residues.

8.5 Quadratic Reciprocity as Bidirectional Prime Transport

For distinct odd primes p,q,

(p/q)(q/p) = (−1)^(((p−1)/2)((q−1)/2)).

Thus the two symbols agree unless both primes are 3 mod 4, in which case they differ. The theorem transports the question “is p a square mod q?” into the reversed carrier “is q a square mod p?” plus an exact sign correction.

The result is not symmetry without cost; the parity term is the path-comparison residue.

8.6 Reciprocity Signs as Path-Comparison Residues

One may compute a quadratic character by reducing first in one prime carrier or the other. Reciprocity compares these paths. The sign records how lattice points, Gauss sums, or residue permutations cross a boundary under reversal.

Calling the sign a residue is precise: it is the correction required for path-independent reconstruction. Omitting it produces the false law (p/q)=(q/p) for all odd primes.

8.7 First and Second Supplementary Laws

The first law states that −1 is a square mod p exactly when p ≡ 1 mod 4. The second states that 2 is a square mod p exactly when p ≡ ±1 mod 8.

Together with reciprocity, these laws compute every Legendre symbol. They handle boundary numerators not covered by odd-prime exchange and therefore close the recursive descent.

8.8 Gauss’s Lemma and Boundary-Crossing Counts

Gauss’s lemma considers the least residues of a,2a,…,((p−1)/2)a mod p. Let m be the number exceeding p/2. Then

(a/p)= (−1)^m.

The proof pairs residues with signs and compares their product to ((p−1)/2)!. Quadratic character is converted into a count of crossings through the balanced-residue boundary. This supplies a geometric proof carrier for reciprocity.

8.9 Gauss Sums and Additive–Multiplicative Carrier Interaction

For quadratic character χ mod p and ζ=e^(2πi/p), define

τ(χ)=Σa mod p χ(a)ζ^a.

A change of variables gives Σχ(a)ζ^(na)=χ(n)τ(χ). One computes

τ(χ)² = χ(−1)p.

Thus τ(χ)=√p or i√p up to sign according to p mod 4. Gauss sums couple additive characters ζ^a with multiplicative characters χ(a), allowing one carrier to diagonalize operations from the other.

8.10 Fourier Transport Over Finite Fields

Characters form an orthogonal basis for functions on finite abelian groups. Additive Fourier transform converts convolution into pointwise multiplication. Multiplicative characters isolate residue classes and power subgroups.

Orthogonality states

Σχ χ(a) over all characters = |G| if a=e, otherwise 0.

This is an exact projection-recovery formula. Incomplete sums lose orthogonality and require nontrivial estimates.

8.11 The Jacobi Symbol as a Compacted Prime-Factor Projection

For odd positive n = ∏p p^ep, define

(a/n)=∏p (a/p)^ep.

The Jacobi symbol inherits multiplicativity and reciprocity-style computation without requiring factorization. If (a/n)=−1, a is not a square mod n. If it equals +1, a may or may not be a square.

The product compresses prime-local signs. An even number of nonresidue components cancels to +1, erasing the failure locations.

8.12 Jacobi Value Versus Actual Quadratic Solvability

For n=15, the Jacobi symbol (2/15)=(2/3)(2/5)=(-1)(-1)=+1, yet x² ≡ 2 mod 15 has no solution because 2 is a nonsquare modulo both 3 and 5.

This is the minimal projection-loss counterkernel. The repair is to retain each prime-power symbol and CRT root condition. The Jacobi symbol is valuable computationally but cannot replace the full local packet.

8.13 The Kronecker Symbol and Extended Boundary Typing

The Kronecker symbol extends the Jacobi symbol to even and signed denominators by defining local factors at −1 and 2. It yields real primitive Dirichlet characters associated with quadratic discriminants.

The extension requires explicit conventions because parity and sign are active boundaries. A formula valid for odd positive denominators cannot simply be reused without adding these local rules.

8.14 Local Quadratic Symbols and Global Compatibility

Quadratic solvability can be studied over R and Qp using Hilbert symbols. The global product formula constrains the local symbols: their product over all places equals 1.

This is a model local–global architecture. Local data are not independent; one global compatibility law relates them. The product law identifies impossible local packets and prepares the transition to quadratic forms and class field theory.

8.15 Reciprocity Counterkernels and Exact Symbol Liftback

Counterkernels include forgetting the reciprocity sign, treating a Jacobi +1 as solvability, or ignoring the prime 2. Each failure identifies a missing local component or boundary correction.

Exact liftback from symbols to roots requires more than character values. One must construct square roots modulo prime powers and combine them by CRT. Symbol evaluation decides existence locally; root construction realizes it.

8.16 Computational Reciprocity and Symbol Certificates

The Jacobi and Kronecker symbols can be computed by an algorithm analogous to Euclid’s algorithm, repeatedly reducing and swapping arguments while applying sign rules. The running time is polynomial in bit length.

A calculation trace records reductions, extracted powers of 2, reciprocity swaps, and sign changes. Independent replay verifies the final symbol without factoring the denominator. This is a compact arithmetic certificate.

Part VII — Reduction, Equivalence, and Global Orbit Residues

Chapter 9 — Continued Fractions and Binary Quadratic Forms

9.1 Continued Fractions as Recorded Euclidean Transport

A finite continued fraction

[a₀;a₁,…,an] = a₀ + 1/(a₁ + 1/(···+1/an))

records the quotient sequence of Euclid’s algorithm for a rational number. Irrational real numbers produce infinite continued fractions by repeated reciprocal-and-floor operations.

The expansion transports a real number into a discrete sequence of positive integers. Rational numbers terminate; irrational numbers do not. The convergents reconstruct increasingly accurate rational approximations.

9.2 Convergents and Minimal Approximation Residue

Define convergents pk/qk by

p_k = a_k p_(k−1)+p_(k−2),
q_k = a_k q_(k−1)+q_(k−2).

They satisfy p_k q_(k−1) − p_(k−1)q_k = (−1)^(k−1), so consecutive convergents are reduced and tightly spaced. One obtains

|x − p_k/q_k| < 1/q_k².

Convergents are best approximations under precise denominator bounds. The residual error is controlled by the next partial quotient.

9.3 Periodicity and Quadratic-Irrational Source Structure

Lagrange’s theorem states that a real number has an eventually periodic continued fraction exactly when it is a quadratic irrational. For √D, the expansion is periodic after the initial term.

The proof tracks finitely many reduced pairs arising from the reciprocal process. Periodicity is therefore a finite-state consequence of the quadratic relation. The repeating block reconstructs algebraic information such as units in quadratic fields.

9.4 Pell Equations and Cyclic Unit Transport

The Pell equation

x² − Dy² = 1

corresponds to units x+y√D of norm 1 in Z[√D] or the appropriate integer ring. Solutions arise from convergents to √D. A fundamental solution generates infinitely many:

x_n + y_n√D = (x₁+y₁√D)^n.

Continued fractions construct the minimal positive generator. The entire solution set is then a cyclic orbit under multiplication.

9.5 Binary Quadratic Forms as Arithmetic Carriers

A binary quadratic form is

Q(x,y)=ax²+bxy+cy²

with discriminant Δ=b²−4ac. It is positive definite when Δ<0 and a>0, indefinite when Δ>0 and nonsquare, and degenerate when Δ=0.

Representation asks whether Q(x,y)=n for integers x,y. The coefficients are not intrinsic because integral changes of variables can preserve the represented integers. The correct carrier is therefore an equivalence class of forms.

9.6 Discriminants as Invariants but Not Complete Classifiers

Under a determinant-one integral change of variables, the discriminant remains fixed. It determines broad geometry, parity constraints, and the associated quadratic order.

Multiple inequivalent forms can share one discriminant. For Δ=−20, the forms x²+5y² and 2x²+2xy+3y² belong to different classes. The discriminant separates families but not individual orbits. The class group is the surviving residue.

9.7 Proper and Improper Equivalence

Forms Q and Q′ are properly equivalent if Q′(x,y)=Q(αx+βy,γx+δy) for a matrix in SL₂(Z). Allowing determinant −1 gives improper equivalence.

Orientation is therefore a load-bearing distinction. Proper classes form the natural group under Gauss composition. Passing to improper equivalence may identify a class with its inverse and changes the classification.

9.8 Change-of-Variables Move Groupoids

SL₂(Z) is generated by elementary transformations corresponding to substitutions such as (x,y)↦(x+ky,y) and (x,y)↦(−y,x). These moves connect equivalent forms.

The collection of forms and admissible transformations forms a groupoid: objects are forms, morphisms are changes of variables. A reduction algorithm selects representatives but must preserve a record of the transformation for exact liftback of represented solutions.

9.9 Reduction as Skeleton Compaction

For negative discriminant, a reduced positive definite form satisfies inequalities such as |b|≤a≤c with boundary sign conventions. Every class contains a reduced form, and only finitely many reduced forms exist for fixed Δ.

Reduction compacts an infinite orbit into a finite canonical region. Boundary conventions ensure uniqueness except for forms with extra symmetries. The reduced representative is a skeleton for the class, not the original coefficient triple.

9.10 Positive Definite Forms and Finite Reduced Regions

If |b|≤a≤c and Δ=b²−4ac<0, then a is bounded in terms of |Δ|. This makes enumeration finite. One lists admissible a,b,c satisfying the discriminant equation and reduction inequalities.

The class number h(Δ) is the number of proper equivalence classes. Reduction converts an abstract orbit problem into a finite search with a proof of completeness.

9.11 Examples of Positive Definite Forms

For Δ=−4, the principal form x²+y² is the unique reduced class. For Δ=−20, reduced forms include x²+5y² and 2x²+2xy+3y², giving class number 2.

Their represented primes differ according to congruence and splitting conditions. Equal discriminant does not imply identical representation behavior; class identity matters. Concrete examples expose the class-group residue before the general theory is developed.

9.12 Gauss Composition and Class-Group Structure

Gauss composition combines two proper classes of primitive forms of the same discriminant to produce a third. The operation is well defined on equivalence classes, associative, has the principal form as identity, and gives inverses by changing b to −b.

The construction is delicate because coefficient formulas require compatibility and gcd handling. The abstract class-group statement is not justified merely by multiplying represented integers; one must prove independence from representatives and exact discriminant preservation.

9.13 More Examples of Binary Quadratic Forms

Computing class groups for small discriminants reveals cyclic groups, products of cyclic groups, ambiguous classes of order 2, and varying representation patterns. Reduction supplies representatives; composition supplies multiplication.

Examples should include explicit transformations and represented integers, not only class counts. This makes visible the distinction between a numerical invariant h(Δ) and the full group structure.

9.14 Genus Projection and Global Class Residue

Genus theory groups form classes that are locally equivalent at all primes. Genus characters provide a coarse quotient of the class group, often detecting its 2-primary structure.

Different classes can lie in the same genus. Thus local equivalence and congruence conditions may fail to determine global equivalence. The remaining class within a genus is a genuine local-to-global residue.

9.15 Local Equivalence Versus Global Equivalence

Two forms may be equivalent over R and over every Zp yet not over Z. Local transformations exist independently, but they may not glue to one integral matrix.

A local–global theorem must therefore be stated at the correct equivalence level. Representation of numbers, equivalence of forms, and isomorphism of lattices have different obstruction groups. Local success cannot be averaged into global equivalence.

9.16 Indefinite Binary Quadratic Forms

When Δ>0 is nonsquare, a primitive indefinite form takes positive and negative values and has infinitely many automorphisms. Reduction no longer produces a finite set of isolated representatives; it produces cycles.

The automorphism group is related to solutions of Pell equations. Dynamics along the reduction cycle encode the fundamental unit of the associated real quadratic order.

9.17 Reduction Cycles, Path Order, and Infinite Stabilizers

Indefinite reduction repeatedly applies transformations and eventually returns to an equivalent reduced form. The ordered cycle matters: it records transport through neighboring forms and reconstructs an automorphism.

The stabilizer is infinite cyclic up to sign. Treating the set of reduced forms without its cyclic order loses the regulator and fundamental-unit data. Path order is therefore an active distinction.

9.18 Representation Transport and Exact Witness Liftback

If Q′=Q∘M and Q′(u,v)=n, then Q(M(u,v))=n. The matrix M lifts a representation witness from the reduced form to the original form.

A reduction algorithm that reports only the final representative cannot reconstruct solutions. It must retain the transformation matrix or a product of elementary moves. This is exact source liftback.

9.19 Class Groups as Factorization Counterkernels

The class group measures failure of unique factorization into elements. A nonprincipal ideal represents an obstruction to choosing a global generator. Relations among ideal classes explain distinct element factorizations.

Class number 1 means every ideal is principal and the ring of integers is a PID. A nontrivial class is a minimal global residue invisible in local ideal factorization.

9.20 Algorithmic Reduction and Replay Certificates

An implementation of form reduction should output the reduced form, transformation matrix, discriminant check, and reduction inequalities. Replay verifies determinant, coefficient transport, and terminal conditions.

For composition, the certificate includes input representatives, intermediate gcd adjustments, output form, and reduction transform. Exact matrices prevent a black-box class label from becoming the sole authority.

Part VIII — Algebraic Integers and Parametric Liftback

Chapter 10 — Gaussian Integers and Special Integer Structures

10.1 Gaussian Integers as a Two-Dimensional Arithmetic Carrier

The Gaussian integers are Z[i]={a+bi:a,b∈Z}. Addition and multiplication preserve the lattice. The norm

N(a+bi)=a²+b²

is multiplicative: N(zw)=N(z)N(w).

This ring enlarges Z so that sums of two squares become norms and some ordinary primes split. The embedding into C supplies geometry, but arithmetic conclusions depend on the lattice and unit structure.

10.2 Norm, Units, and Euclidean Descent

The units are ±1,±i, the elements of norm 1. Given z,w with w≠0, choose q∈Z[i] nearest to z/w in the complex plane and set r=z−qw. Then N(r)<N(w).

Thus Z[i] is Euclidean, hence a PID and UFD. Euclid’s algorithm, gcds, Bézout identities, and prime factorization all extend to this carrier.

10.3 Gaussian Prime Classification

A rational prime p behaves as follows. The prime 2 ramifies:

2 = −i(1+i)².

If p ≡ 1 mod 4, it splits as p=ππ̄. If p ≡ 3 mod 4, it remains prime in Z[i]. The criterion follows from whether −1 is a square mod p.

Gaussian primes also include a+bi with both coordinates nonzero when a²+b² is an ordinary prime. Prime classification combines norm, reciprocity, and ambient-ring factorization.

10.4 Splitting, Inertness, and Ramification

Splitting means a prime ideal decomposes into distinct factors; inertness means it remains prime; ramification means repeated prime factors occur. In quadratic extensions, discriminants govern ramified primes and residue symbols govern splitting away from the discriminant.

These behaviors are local transport types from Z into a larger integer ring. A prime is not intrinsically split or inert without naming the extension.

10.5 Sums of Two Squares as Norm Reconstruction

A positive integer n is a sum of two squares exactly when every prime q ≡ 3 mod 4 occurs in n with even exponent. Necessity follows because such q remains Gaussian prime and divides a+bi and a−bi in paired fashion. Sufficiency constructs Gaussian factors for primes p ≡ 1 mod 4 and combines them multiplicatively.

The theorem reconstructs additive representations from multiplicative factorization in Z[i]. Counting representations requires units, conjugation, and factor-choice multiplicities.

10.6 Element Factorization and Ambient Ring Dependence

The integer 5 is prime in Z but factors in Z[i]:

5=(2+i)(2−i).

Irreducibility is therefore carrier-relative. A proof using “prime” must specify the ring. Enlarging the carrier may expose a hidden factorization that solves an additive problem, but liftback must return integer coordinates and control units.

10.7 Eisenstein Integers and Cubic Symmetry

The Eisenstein integers Z[ω], with ω²+ω+1=0, form a triangular lattice. The norm is

N(a+bω)=a²−ab+b².

The ring is Euclidean and has six units. Rational primes split according to congruence modulo 3, with 3 ramified. This carrier is natural for cubic symmetries and equations involving x²−xy+y².

10.8 Norm Equations and Algebraic Liftback

A norm equation N(α)=n transforms an integer representation problem into factorization in an extension. Solving it requires identifying allowable prime ideal exponents, class-group obstructions, units, and embeddings.

A factorization of the ideal (n) may not lift to an element α if the selected ideal product is nonprincipal. Thus ideal solvability and element solvability are distinct layers.

10.9 Pythagorean Triangles

Integer solutions to x²+y²=z² correspond to factorizations

(x+iy)(x−iy)=z²

in Z[i]. For primitive triples, gcd(x,y)=1 and one leg is even. Coprimality forces x+iy to be a unit times a square, yielding the standard parameterization.

The Gaussian proof makes the factorization mechanism explicit and explains why parity and coprimality are necessary.

10.10 Primitive Triples and Coprimality Gates

Every primitive Pythagorean triple, after exchanging x and y if necessary, has

x=m²−n²,
y=2mn,
z=m²+n²,

where m>n>0, gcd(m,n)=1, and m,n have opposite parity. These conditions ensure primitiveness. If they fail, a common factor or factor 2 appears.

The gates are not decorative restrictions; they are exactly what makes the inverse factorization unique up to symmetry.

10.11 Rational Parameterization as Exact Construction

The unit circle X²+Y²=1 can be parameterized by lines of rational slope t through (−1,0):

X=(1−t²)/(1+t²),
Y=2t/(1+t²).

Writing t=n/m and clearing denominators gives Pythagorean triples. Every rational point except the base point arises uniquely from its slope.

This is a geometric construction with exact rational liftback. The excluded base point is a typed boundary case, not a failure.

10.12 Uniqueness, Sign, Ordering, and Source-Fibre Conventions

The parameters (m,n) generate multiple signed and permuted versions of the same triangle. Multiplying both by k scales the triple by k² in the raw formulas, while primitive normalization removes common factors.

A classification theorem must choose conventions: m>n>0, opposite parity, gcd 1, even leg designated as y. Without these, the parameter map is surjective but not unique.

10.13 Closure of Parametric Families

A parameterization is complete only if every target object arises and every allowed parameter yields a valid target. For primitive Pythagorean triples, both directions are proved.

For more general Diophantine equations, an attractive formula may cover only one component or miss singular points. Closure requires a source-fibre analysis and boundary audit, not merely substitution verification.

10.14 Projection from Algebraic Factorization to Integer Solutions

Factorization in an extension produces algebraic elements whose coordinates may reconstruct integer solutions. Units, conjugates, denominators, and ideal classes can alter the projection.

The final step must verify the original equation and integrality. A norm factorization that exists only in the field but not in the integer ring may introduce denominators and fail the target claim.

Part IX — Analytic Reconstruction and Nonuniformity Exposure

Chapter 11 — Dirichlet Series, L-functions, and Prime Distribution

11.1 Dirichlet Series as Analytic Arithmetic Carriers

A Dirichlet series has the form

F(s)=Σn≥1 a(n)n^(−s).

It transports a sequence into a complex function. Large n are weighted by n^(−Re(s)), so convergence depends on the real part of s. The coefficients can often be recovered from the function in a right half-plane.

The analytic carrier exposes poles, zeros, and growth that are invisible in the raw sequence. Conversely, analytic manipulation is valid only in declared convergence or continuation domains.

11.2 Coefficient Data and Regions of Convergence

Every Dirichlet series has an abscissa of convergence σc and an abscissa of absolute convergence σa. It converges for Re(s)>σc and diverges for Re(s)<σc; absolute convergence holds to the right of σa.

Termwise addition, multiplication, differentiation, and rearrangement require appropriate convergence. A formal identity outside the valid region must be justified by analytic continuation, not by the original series.

11.3 Products of Dirichlet Series

If F(s)=Σa(n)n^(−s) and G(s)=Σb(n)n^(−s) converge absolutely, then

F(s)G(s)=Σn≥1 (a*b)(n)n^(−s),

where

(a*b)(n)=Σd|n a(d)b(n/d).

Thus analytic multiplication corresponds exactly to Dirichlet convolution. Absolute convergence is the transport gate allowing sums to be rearranged.

11.4 Euler Products as Prime-Local Reconstruction

If a(n) is multiplicative, then in a domain of absolute convergence,

Σn≥1 a(n)n^(−s) = ∏p Σk≥0 a(p^k)p^(−ks).

For ζ(s),

ζ(s)=Σn≥1 n^(−s)=∏p (1−p^(−s))^(−1), Re(s)>1.

Unique factorization is the arithmetic theorem behind the product; absolute convergence is the analytic theorem permitting reconstruction.

11.5 Absolute Convergence as a Composition Gate

Without absolute convergence, rearranging infinitely many terms or factors may change a sum or be undefined. Euler-product manipulations, logarithms, and differentiation must therefore begin in a safe half-plane.

After two analytic functions are shown equal on a connected open region, analytic continuation may extend the identity. The continuation is a new transport based on uniqueness of holomorphic functions, not permission to manipulate the original divergent product everywhere.

11.6 Analytic Continuation as Carrier Migration

The zeta series defines ζ(s) only for Re(s)>1, but integral formulas and functional identities extend it meromorphically to C with a simple pole at s=1. The continued function agrees with the series where both are defined.

Analytic continuation preserves functional identity but changes representation. A value such as ζ(−1)=−1/12 belongs to the continued function; it is not the ordinary sum 1+2+3+···. Conflating these carriers creates false arithmetic statements.

11.7 The Zeta Function and Singular Structure

The Riemann zeta function has Euler product, a simple pole at s=1, trivial zeros at negative even integers, and nontrivial zeros in 0<Re(s)<1. Its logarithmic derivative satisfies

−ζ′(s)/ζ(s)=Σn≥1 Λ(n)n^(−s)

for Re(s)>1, where Λ is the von Mangoldt function.

The pole encodes the main density of integers; zeros govern oscillations in prime-weighted sums. The logarithmic derivative transports prime powers into analytic singularities.

11.8 The Prime Number Theorem

Let π(x) count primes ≤x. The prime number theorem states

π(x) ~ x/log x.

Equivalent forms include ψ(x)~x, where ψ(x)=Σn≤x Λ(n). The theorem describes asymptotic density, not local gap size or the location of an individual next prime.

The main term emerges from the pole of ζ(s) at 1, while controlling errors requires information about zeros near the boundary.

11.9 Proof of the Prime Number Theorem

A classical analytic proof shows ζ(s) has no zeros on Re(s)=1, then applies a Tauberian theorem or contour method to deduce ψ(x)~x. Nonvanishing is established by combining logarithms of ζ at related points with a nonnegative trigonometric polynomial.

The proof’s structure is transport: prime coefficients → Dirichlet series → boundary nonvanishing → asymptotic coefficient sum. Every arrow has domain and growth conditions.

11.10 Zero-Free Boundaries and Tauberian Liftback

A Tauberian theorem converts behavior of a generating function near its boundary of convergence into asymptotics of coefficients, under positivity or regularity assumptions. The theorem provides liftback from analytic singularity to arithmetic summation.

Boundary hypotheses are noncompensatory. Meromorphic continuation alone does not imply a desired asymptotic; zeros, growth, and coefficient signs matter. A hidden boundary singularity can dominate the error.

11.11 Dirichlet’s Theorem on Arithmetic Progressions

If gcd(a,q)=1, there are infinitely many primes p ≡ a mod q. The proof uses characters modulo q to isolate the residue class and shows that the corresponding prime reciprocal sum diverges.

This is a global distribution theorem across a quotient carrier. Coprimality is necessary because a nonunit residue class contains at most one prime divisor of q.

11.12 Dirichlet Characters as Residue-Class Projectors

A Dirichlet character χ mod q is a group character of (Z/qZ)× extended by zero to nonunits. Character orthogonality gives

1_(n≡a mod q) = 1/φ(q) Σχ χ(n)overline{χ(a)}

for gcd(a,q)=1.

Thus residue-class selection is decomposed into multiplicative spectral components. Each character creates an L-function

L(s,χ)=Σχ(n)n^(−s).

11.13 Primitive Characters, Conductors, and Native Carriers

A character modulo q may factor through a smaller modulus f. The least such f is its conductor, and the corresponding character is primitive.

The nominal modulus can therefore contain redundant carrier structure. Functional equations and Gauss sums are naturally expressed using the conductor. Failing to separate modulus from conductor introduces incorrect local factors and scale dependence.

11.14 Proof of Dirichlet’s Theorem

For Re(s)>1,

log L(s,χ) = Σp χ(p)p^(−s) + bounded prime-power terms.

Character orthogonality isolates primes in a chosen reduced class. The principal character contributes a logarithmic divergence as s→1+, while nonprincipal L-functions remain finite and nonzero at s=1. Hence the prime sum in every reduced class diverges, proving infinitely many primes.

The proof depends critically on nonvanishing at 1.

11.15 Nonvanishing of L-series at s = 1

For nonreal characters, algebraic combinations and positivity arguments control L(1,χ). For real nonprincipal characters, the issue is subtler because cancellation is weaker. Classical proofs use positivity of suitable zeta products or class-number formulas.

Nonvanishing is not a technical side lemma; it is the exact gate preventing a character component from canceling the principal divergence needed for Dirichlet’s theorem.

11.16 Exceptional Real Characters and Worst-Fibre Replay

Real primitive characters can have a zero unusually close to 1, producing distorted prime distribution in finite ranges. Even when global theorems remain true, constants and error terms may deteriorate with the conductor and zero location.

A uniform statement must audit this worst fibre. Average behavior over characters cannot pay for one exceptional character if the target quantifies over every modulus.

11.17 Character Orthogonality and Projection Recovery

For a finite abelian group G,

Σχ∈Ĝ χ(g) = |G| if g=e, otherwise 0.

Dually,

Σg∈G χ(g)overline{ψ(g)} = |G| if χ=ψ, otherwise 0.

These identities invert finite Fourier transforms and reconstruct residue indicators. The full character group is required. Omitting components produces a projection with a nontrivial kernel.

11.18 Asymptotic Main Terms Versus Local Error Residues

An asymptotic f(x)~g(x) means f(x)/g(x)→1. It permits substantial finite-range deviation. Error terms quantify the surviving residue:

f(x)=g(x)+E(x).

Uniformity requires explicit dependence of E on parameters such as modulus, character, degree, or height. Writing O(1) without declaring its dependencies can conceal unbounded nonuniformity.

11.19 Nonuniformity in Modulus, Height, and Zero Proximity

A theorem for each fixed q may have constants C(q) growing rapidly. It cannot be exported to q varying with x without controlling that growth. Similar issues arise with field discriminant, polynomial degree, conductor, and zero-free margins.

The quantifier order

for every q, as x→∞

is weaker than

uniformly for q≤Q(x).

Exposure Geometry records the parameter carrier and tests whether constants remain bounded in the intended regime.

11.20 Explicit Formulae as Prime–Zero Transport

Explicit formulas relate sums over prime powers to sums over zeros of ζ or L-functions, plus pole and archimedean terms. Schematically,

ψ(x) = x − Σρ x^ρ/ρ + boundary corrections.

The equality exhibits prime irregularity as interference among zero contributions. Truncation requires error terms, and zeros near Re(s)=1 dominate long-range deviations.

11.21 Numerical Verification, Truncation, and Proof Capacity

Numerically evaluating ζ, L-functions, or prime sums requires finite truncation, precision control, and rigorous error bounds. A list of computed zeros is evidence only within the verified height and region. Turing-style methods can certify that no zeros were missed up to a height when all hypotheses are checked.

A decimal approximation without an enclosure is observational output, not an exact theorem. Computation becomes proof only through a validated certificate and a theorem connecting the finite record to the target claim.

11.22 Analytic Claims, Computational Claims, and Scope Discipline

An analytic continuation theorem is formal. A numerical value is computational. An empirical timing benchmark is implementation-specific. These claim types must not be conflated.

The analytic domain, parameter range, branch convention, truncation bound, and precision must accompany each result. Scope discipline prevents a correct local computation from being advertised as a global analytic theorem.

Part X — 2026 Arithmetic Extensions

Chapter 12 — Elliptic Curves and Local–Global Arithmetic

12.1 Cubic Curves as Algebraic Arithmetic Carriers

An elliptic curve over a field K is a smooth projective genus-one curve with a chosen K-rational point. In characteristic not 2 or 3, it can often be written

y² = x³ + Ax + B

with discriminant Δ = −16(4A³+27B²) nonzero.

The chosen point at infinity serves as the identity. The curve’s arithmetic depends on the base field: rational points, finite-field points, real components, and p-adic points are distinct carriers.

12.2 Singular Versus Nonsingular Cubics

If Δ=0, the cubic has a node or cusp and is not an elliptic curve. Its geometry and group structure degenerate to multiplicative or additive types after parameterization.

Nonsingularity is the boundary gate ensuring tangent lines are well defined and the chord-and-tangent law is globally regular. A formula for point addition may still be written on a singular cubic, but the elliptic-curve theorems no longer apply.

12.3 The Chord-and-Tangent Composition Law

Given points P and Q, draw the line through them, or the tangent when P=Q. It meets the cubic at a third point R; reflect R across the x-axis to obtain P+Q. Algebraically, if xP≠xQ,

λ=(yQ−yP)/(xQ−xP),
x_(P+Q)=λ²−xP−xQ,
y_(P+Q)=λ(xP−x_(P+Q))−yP.

The law extends projectively and is associative, though associativity requires a genuine theorem rather than the picture alone.

12.4 Rational Points as a Finitely Generated Group

Mordell’s theorem states

E(Q) ≅ E(Q)_tors ⊕ Z^r.

The finite torsion subgroup and nonnegative rank r describe the rational-point group. The theorem proves finite generation but does not automatically compute generators or rank.

Descent, height bounds, and local information provide algorithms and conditional enclosures. Existence of infinitely many points is equivalent to positive rank.

12.5 Reduction Modulo Primes

For an elliptic curve with integral model, reducing coefficients mod p gives a curve over Fp. If p does not divide the discriminant, reduction is good and maps rational points with suitable denominators into E(Fp).

Reduction provides finite-group information about torsion and local behavior. It is not injective on all rational points, and bad primes require separate analysis.

12.6 Good Reduction, Bad Reduction, and Boundary Residue

At a good prime, the reduced cubic is nonsingular. At a bad prime, it becomes singular. Bad reduction can be multiplicative or additive and contributes local conductor and discriminant data.

The discriminant boundary is therefore an active arithmetic carrier. Ignoring bad primes can invalidate global formulas for L-functions, heights, or torsion.

12.7 Frobenius and Point Counting

For E/Fq,

#E(Fq)=q+1−a_q,

where |a_q|≤2√q by Hasse’s theorem. Frobenius acts on the Tate module with characteristic polynomial T²−a_qT+q.

Point counts determine local Euler factors. Algorithms such as Schoof’s compute them in polynomial time. The trace a_q is an invariant of the finite-field curve, not a complete description of its point group.

12.8 Heights and Arithmetic Scale

The naive height measures coordinate magnitude; the canonical height satisfies

hat h(nP)=n² hat h(P)

and differs from the naive height by a bounded amount. It turns the free part of E(Q) into a quadratic lattice.

Heights make infinite search finite by bounding possible generators or preimages. The constants depend on the curve and model and must be controlled in uniform statements.

12.9 Descent as Projection, Obstruction, and Liftback

An n-descent maps E(Q)/nE(Q) into a finite cohomological or explicit algebraic carrier. The Selmer group contains the image and is computable from local conditions. Its size bounds the rank.

The difference between the Selmer group and actual rational images is measured by the Tate–Shafarevich group. Thus descent produces a finite projection with a possible global liftback residue.

12.10 Local Points, Global Points, and Reconstruction Failure

A genus-one curve may have points over R and every Qp but no rational point. Such a curve is a torsor under its Jacobian elliptic curve and represents a nontrivial global obstruction.

This is an exact local–global counterkernel. Every isolated local carrier is valid; the failure lies in global gluing. The obstruction cannot be removed by testing more finite primes without a theorem controlling the global class.

12.11 Torsion, Rank, and Source-Fibre Structure

Torsion points have finite order; the free generators determine rank. Reduction modulo good primes constrains torsion because prime-to-p torsion injects into E(Fp). Heights and descent constrain the free part.

Knowing the rank does not identify generators, and knowing several point counts does not determine E(Q). Each invariant has a nontrivial source fibre.

12.12 Elliptic-Curve Factorization

ECM chooses a random elliptic curve modulo N and computes a large scalar multiple of a point. Modulo a hidden factor p, failure to invert a denominator can reveal gcd(denominator,N)=p.

The method succeeds when #E(Fp) is smooth. Random curves vary the hidden group order, making ECM effective for medium-size factors regardless of the total size of N.

12.13 Elliptic-Curve Cryptography

ECC uses the presumed hardness of discrete logarithms in a large prime-order subgroup of E(Fq). Scalar multiplication is efficient; inversion of Q=kP is believed hard for suitable curves and parameters.

Security requires point validation, subgroup checks, constant-time arithmetic, secure nonce generation, and resistance to side channels. The abstract group alone is not a deployed cryptosystem.

12.14 Subgroup, Cofactor, and Invalid-Curve Counterkernels

If a protocol accepts points outside the intended subgroup, an attacker may force computations in a small subgroup and recover secret residues. Invalid-curve attacks exploit points satisfying a related curve equation with weaker group structure.

The repair is complete input validation, cofactor handling, and protocol design that binds the exact curve and subgroup. These are carrier-boundary obligations.

12.15 Formal Theorem Versus Executed Cryptographic System

The elliptic-curve group law and discrete-log assumptions are formal. A concrete scheme selects a curve, encoding, hash-to-curve rule, random generator, and protocol. Software and hardware realize these choices.

Security certification therefore requires separate proofs and tests at mathematical, algorithmic, implementation, and physical levels. Success at one level cannot compensate for a failed gate at another.

Chapter 13 — Formal and Computational Number Theory in 2026

13.1 Exact Integer, Rational, Polynomial, and Finite-Field Computation

Modern systems support arbitrary-precision integers, reduced rationals, symbolic polynomials, quotient rings, and finite fields. Exactness means operations satisfy algebraic identities without rounding, not that resource use is unbounded.

Representations matter. Dense and sparse polynomials have different costs. Finite fields may use polynomial bases, normal bases, or tower fields. Conversion must preserve the declared element and modulus.

13.2 p-adic and Algebraic-Number Computation

p-adic computation uses finite precision with valuation-aware error propagation. Algebraic numbers may be represented by minimal polynomials plus isolating data or by coordinates in a number field.

Two symbolic expressions may denote the same algebraic number but require field embeddings and reduction to compare. Equality testing is exact only when the representation includes sufficient embedding or canonicalization data.

13.3 Computer Algebra Systems as Typed Proof Carriers

A computer algebra system can construct factors, Gröbner bases, class groups, or normal forms. Its output should be interpreted through the theorem implemented and the certificate emitted.

A large trusted code base is not equivalent to a small proof kernel. For high-assurance claims, one exports checkable relations: multiplication identities, transformation matrices, modular witnesses, or proof terms.

13.4 Primality Certificates and Independently Checkable Witnesses

A primality certificate reduces the claim “n is prime” to smaller prime claims and modular identities. Verification should be deterministic and substantially cheaper than discovery.

Certificate chains terminate at small primes checked directly. The complete chain owns every dependency. A single probable-prime leaf prevents exact promotion.

13.5 Certified Factorization Pipelines

A certified pipeline alternates factor search, probable-prime screening, rigorous primality proof, multiplicity extraction, and product verification. Failure to split a composite creates a frontier, not permission to label it prime.

The final record includes all prime powers and validates ∏p^e=N exactly. Independent software should replay the certificate without reproducing the search.

13.6 Formalization in Proof Assistants

Proof assistants encode definitions, theorem statements, and deductions in a small trusted kernel. Number theory formalization includes induction, divisibility, finite groups, fields, valuations, and analytic estimates.

Formalization exposes hidden coercions and missing hypotheses but does not automatically find proofs. Library choices and representation design affect proof complexity. The kernel certificate is formal, not an empirical claim about software correctness below the trusted base.

13.7 Proof Object, Certificate, Replay, and Trusted Kernel

A proof object contains derivation data accepted by a kernel. A certificate may be domain-specific and checked by a verified theorem. Replay reconstructs the conclusion from the certificate and inputs.

Trust is concentrated in the checker and its execution environment. Minimizing the trusted kernel reduces ghost assumptions. A human-readable explanation and a machine-checkable object serve complementary purposes.

13.8 Randomized Algorithms and Explicit Probability Claims

A Monte Carlo algorithm may return an incorrect answer with bounded probability. A Las Vegas algorithm always returns a correct answer but has random running time. Randomized search can produce an exact witness whose verification is deterministic.

Probability bounds require assumptions about independent uniform random bits. Reusing seeds, biased generators, or adversarial randomness changes the theorem. The randomness source belongs to the computational packet.

13.9 Precision, Accuracy, Truncation, and Representation Residues

Precision measures resolution; accuracy measures closeness to truth. A high-precision decimal can be inaccurate if the algorithm or input is wrong. Truncation error, rounding error, conditioning, and representation error must be separated.

Exact modular arithmetic has no rounding but may lose global information by quotient projection. Exactness inside one carrier does not guarantee exact liftback to another.

13.10 Memory, Runtime, Parallelism, and Resource Accounting

Algorithms consume time, memory, communication, and sometimes randomness. Parallel speedup may duplicate memory or coordination costs. Asymptotic complexity does not determine practical crossover points.

A computational theorem should state resource dependence on bit length, degree, field size, precision, and output size. Hidden dependence on an unbounded parameter is nonuniformity debt.

13.11 Reproducible Seeds, Logs, Versions, and Hashes

A reproducible experiment records source data, software and library versions, compiler or interpreter, parameters, random seeds, precision, hardware-relevant options, and cryptographic hashes.

A hash identifies a byte sequence but does not validate its mathematical meaning. Provenance and theorem-level verification remain separate fields.

13.12 Structured Perturbation and Adversarial Test Suites

Testing should include boundary values, singular inputs, maximal valuations, repeated support, balanced and unbalanced cases, random instances, and independently generated known answers. Metamorphic tests verify invariants under transformations.

Adversarial testing is generative: the first failure identifies which carrier or assumption is missing. Passing a test suite does not prove a universal theorem, but failing one supplies a concrete counterkernel.

13.13 No-Ghost Computation: Formal Existence Is Not Execution

A formally defined algorithm does not imply that an implementation exists, terminates under actual resource limits, or returns the claimed record. An executable program does not imply empirical security or physical reliability.

No-ghost computation requires finite encodings, explicit transitions, termination or limit semantics, resource accounting, error handling, and reproducible output. Uninstantiated oracles and infinite-precision steps block Φ1 promotion.

13.14 Promotion from Φ0 Formal to Φ1 Computational

To promote a number-theoretic construction from formal to computational, specify data representation, algorithm, termination, complexity, precision or exactness, exceptional inputs, output schema, and verifier. Execute representative and adversarial cases and preserve records.

The computational certificate is scoped to the implementation and resource model. It includes the formal theorem but does not imply physical or cryptographic adequacy outside its tested environment.

Part XI — Number-Theoretic Exposure Geometry

Chapter 14 — Arithmetic Frame, Residue, and Successor Construction

14.1 The Maximal Arithmetic EG Frame

The maximal frame contains every distinction that could affect the target: carrier, signs, units, prime support, valuations, modulus, local completions, equivalence, parameter ranges, boundary cases, algorithmic encoding, and quantifier order.

Maximal does not mean indiscriminately infinite. It is target-relative: distinctions are included when changing them can alter the conclusion or its reconstruction. Frame construction precedes compression so that omitted data are deliberate rather than invisible.

14.2 Target-Relative Arithmetic Skeleton Extraction

A skeleton is the least packet from which the declared target can be reconstructed. For gcd, it may be the final gcd plus Bézout coefficients. For multiplicative functions, prime-power values suffice. For a quadratic-form class, a reduced representative plus equivalence convention suffices.

Minimality is relative to the target. A radical is enough to identify prime support but not valuation depth. A discriminant locates a quadratic-form family but not a class. Skeleton claims require a reconstruction theorem.

14.3 Prime, Local, Algebraic, and Analytic Carriers

Prime carriers include valuation coordinates and finite fields. Local carriers include Zp and Qp. Algebraic carriers include number fields, integer rings, ideals, forms, and elliptic curves. Analytic carriers include Dirichlet series and L-functions.

The same arithmetic object can inhabit several carriers, but maps between them have distinct kernels and scopes. Carrier plurality is productive only when each interface is typed.

14.4 Dyadic Relations and Higher-Arity Arithmetic Cells

Divisibility, congruence, and coprimality are dyadic. Addition a+b=c is triadic. Polynomial evaluation combines coefficients and powers in a higher-arity cell. Euler products assemble infinitely many local factors under convergence control.

A higher-arity residue appears when the full interaction is not reconstructible from pairwise reports. Additive cancellation and global class obstructions are central examples.

14.5 Transport, Composition, and Path-Comparison Residue

Arithmetic transports compose only when interfaces match. One may reduce modulo p, lift p-adically, and reconstruct rationally, but each step has compatibility and precision conditions.

If two admissible transport paths produce different reconstructed objects, their difference is a path-comparison residue. Reciprocity signs, carry chains, branch choices, and analytic continuation monodromy are examples of correction data needed for coherent composition.

14.6 Ambient Rings, Fields, Completions, and Embeddings

Factorization, irreducibility, and unit structure depend on the ambient ring. A polynomial can split over C, remain irreducible over Q, and factor differently over Qp. An algebraic number has multiple embeddings into C.

Ambient data are not background notation. They determine what operations exist and which equivalences are valid. Exact liftback must return to the original ambient carrier rather than stop in a convenient extension.

14.7 Source Ancestry and Reconstruction Fibres

A compact object may have many ancestors: one residue class has infinitely many integer lifts; one norm has many algebraic preimages; one invariant has several orbits. Source ancestry records the original object and every transport.

A reconstruction fibre is the set of source states compatible with the compact packet. Uniqueness requires this fibre to be one declared equivalence class. Otherwise the correct result is a family or a new splitter.

14.8 M0 Worst-Fibre and Tail Exposure in Number Theory

Worst-fibre exposure asks whether a theorem survives the most adverse admissible input. Prime gaps, exceptional characters, high valuation depth, singular roots, and dense support overlap can dominate a universal claim despite benign average behavior.

Tail and average estimates are valid for their quantifiers. They become invalid when exported to every input without a firewall. The audit searches for ruin fibres rather than typical cases.

14.9 M1 Parameter Nonuniformity in Bounds and Error Terms

A bound C may depend on modulus q, degree d, discriminant Δ, precision N, rank r, or valuation depth. If the theorem requires one constant across a growing family, these dependencies must be controlled.

Nonuniformity is detected by scaling the carrier while keeping the nominal statement fixed. A constant per prime, cell, or component may accumulate without bound. Uniformity requires explicit supremum control.

14.10 M2 Boundary Singularity in Roots, Fibres, and Analytic Continuation

Singular boundaries include derivative zero in Hensel lifting, discriminant zero for curves and forms, bad reduction, poles of L-functions, repeated roots, and noninvertible denominators.

Generic theorems may fail exactly there. A phrase such as “except finitely many cases” is acceptable only when those cases are owned and either excluded by the target or handled separately.

14.11 M3 Interaction Multiplicity and Prime-Support Reuse

Prime support can be reused across several local operations, creating double counting. Combinatorial decompositions can assign one divisor, prime, or certificate to multiple owners. Local capacity then exceeds global capacity only on paper.

One-use accounting attaches every support resource to an owner and records when it is consumed, shared, or reset. Multiplicity residue survives when no globally feasible allocation exists.

14.12 M4 Projection Loss in Congruences, Symbols, Invariants, and Averages

Reduction modulo n forgets magnitude. Radicals forget exponents. Jacobi symbols forget component signs. Discriminants forget classes. Averages forget exceptional inputs.

Projection is admissible only when a recovery theorem reconstructs every distinction needed by the target. Otherwise the forgotten-distinction ledger becomes an active residue.

14.13 M5 Proof-Carrier Capacity and Verification Complexity

A theorem may have a concise statement but require a certificate too large or complex for the proposed proof carrier. Compressing the certificate by deleting ancestry, overlap, or boundary information can invalidate it.

Proof capacity concerns whether the carrier can express and verify every load-bearing relation. A flow certificate, for example, can verify a constructed incidence but cannot create missing arithmetic incidence by syntax alone.

14.14 M6 Structured Perturbation and Adversarial Arithmetic

A construction should be replayed under scaling, prime-support concentration, balanced cancellation, unit boundaries, singular derivatives, modulus changes, branch permutations, and inverse liftback.

The perturbations preserve admissibility while stressing hidden assumptions. Failure fingerprints determine the next carrier mutation. Adversarial replay is therefore a discovery operation, not only a final audit.

14.15 M7 Framework-Capacity and Decidability Boundaries

Some arithmetic problems are decidable but computationally hard; others become undecidable in sufficiently general Diophantine settings. A framework may classify one family while lacking the language to express another.

At a capacity boundary, one must distinguish unknown result, resource exhaustion, representation failure, and proven undecidability. HALT is justified only by an exact boundary theorem, not by search fatigue.

14.16 M8 Ambient Entanglement in Rings, Ideals, Forms, and Global Orbits

Global arithmetic structure can survive every local or componentwise projection. Class groups, nonprincipal ideals, form classes, Galois groups, and local–global obstructions are ambient residues.

A list of local invariants does not reconstruct the global orbit unless a completeness theorem is supplied. Ambient entanglement is the arithmetic analogue of global organization not reducible to constituent labels.

14.17 M9 Path-Order Dependence in Descent, Lifting, and Reduction

Euclidean quotients, continued-fraction steps, Hensel branches, form-reduction cycles, and analytic continuation paths carry ordered ancestry. Reordering operations may alter intermediate denominators, precision, or branch identity.

Order is not necessarily physical time. It is a dependency structure. A valid proof must show path independence or retain the correction residue generated by alternative paths.

14.18 M10 Source-Ancestry Fibres Under Quotients and Completions

A quotient, completion, or invariant can erase the original source. Multiple rationals can approximate one p-adic truncation; multiple integers share modular data; multiple ideals share a norm.

Ancestry liftback requires bounds, compatibility, or additional labels. Without them, the compact object certifies an equivalence class or fibre, not one source.

14.19 Arithmetic Debt and Forgotten-Distinction Ledgers

Debt is any unresolved obligation introduced by an arithmetic transport: omitted signs, untracked units, unknown prime factors, parameter-dependent constants, nonprincipal ideals, branch ambiguity, truncation errors, or unverified software assumptions.

A ledger assigns each debt an owner, scope, propagation law, and discharge operation. Debt cannot disappear because later notation is cleaner. It must be proved irrelevant, bounded, reconstructed, or quarantined.

14.20 Minimal Arithmetic Residue Extraction

After all licensed identities and reconstructions are applied, the remaining obstruction is minimized. For ax+by=c it is c mod gcd(a,b). For singular Hensel lifting it is the first nonzero Taylor coefficient at the required valuation. For quadratic forms it may be an ideal class. For analytic estimates it may be a zero-free margin.

Minimality excludes redundant descriptions and points directly to the missing theorem or field.

14.21 Counterkernel Construction

A counterkernel contains the fewest distinctions needed to reproduce the failure. It should survive the proposed method but collapse when one load-bearing distinction is removed.

Examples include a composite modulus for cancellation failure, a pseudoprime for primality-test failure, a singular root for unique-lift failure, and two same-discriminant inequivalent forms for invariant incompleteness. The counterkernel determines repair scope.

14.22 Successor-Frame Migration

The residue is promoted to the least-powerful sufficient addition. Add gcd data to a linear equation; prime-factor symbols to a Jacobi result; derivative valuations to a singular lift; ideal classes to element factorization; error bounds to a numerical computation.

Migration preserves the valid prefix and invalidates only dependent claims. The successor frame must strictly reduce unresolved debt rather than rename it.

14.23 Exact Liftback to the Original Number-Theoretic Claim

Liftback reconstructs the original integer, rational point, form, factorization, or quantified theorem from the transformed carrier. Every auxiliary variable, unit, denominator, modulus, and parameter must be eliminated or interpreted.

A proof that ends with an object in a larger ring, a flow network, a numerical approximation, or a local completion is incomplete unless the target was explicitly stated in that carrier.

14.24 Independent Replay and Scoped Certification

Independent replay rechecks the theorem using the sealed hypotheses, certificate, and transformation record. It includes boundary, singular, scaling, and inverse-path tests.

Certification is scoped: a formal theorem, a finite algorithm, and a particular computation receive different certificates. The certificate states exactly what has been reconstructed and what remains frontier.

14.25 Frontier Payloads for Unresolved Arithmetic Reconstruction

When a claim cannot be certified, the correct endpoint is a frontier payload containing the valid prefix, first failed gate, minimal residue, counterkernel, owner, repair cone, next construction, and falsifier.

This preserves progress without inflating status. A frontier is not “unknown” in the vague sense; it is an exact statement of what has been built and what mathematical object is still absent.

Appendix — Computational and Exposure Tools

A.1 Three Calculators for Number Theorists

A basic number-theory toolset should include an exact integer calculator, a modular arithmetic calculator, and a symbolic or arbitrary-precision system. The first handles gcds, factorizations, and exact powers; the second handles residues, inverses, and CRT; the third handles polynomials, fields, and analytic approximations. Results should be treated as computations with declared scope, not as theorem authority.

A.2 Arbitrary-Precision Integer Calculator

An arbitrary-precision calculator supports integers larger than machine-word limits. It should provide gcd, extended gcd, modular powers, integer roots, valuation, factor trial division, and exact quotient-remainder operations. Outputs such as primality or factorization should identify whether they are probable or certified.

A.3 Modular Arithmetic and CRT Calculator

The tool should solve ax ≡ b mod n, list solution multiplicity, compute inverses only for units, and solve compatible systems with coprime or overlapping moduli. A CRT result should include its reconstruction modulus and component verification.

A.4 Prime, Factorization, and Primality-Certificate Tools

A complete tool distinguishes trial division, probable-prime testing, factor search, complete factorization, and primality proof. It should export witnesses or certificates and verify that prime powers multiply to the original input.

A.5 Finite-Field and Polynomial Calculator

The system should construct F_(p^n) from an irreducible polynomial, perform polynomial gcds and factorization, compute Frobenius, trace, norm, and roots, and distinguish polynomial equality from functional equality over finite fields.

A.6 p-adic Expansion and Hensel-Lifting Calculator

A p-adic calculator should track prime, valuation, absolute precision, relative precision, branch ancestry, and derivative conditions. Singular roots must not be passed through a simple-root routine without a branching audit.

A.7 Legendre, Jacobi, Kronecker, and Character Calculator

The calculator should show reciprocity reductions and sign corrections. A Jacobi +1 result should be labeled as inconclusive for residuosity unless factor-local roots are constructed. Character tables should declare modulus and conductor.

A.8 Continued-Fraction and Pell Calculator

The tool should compute continued fractions, convergents, approximation errors, periodic blocks for quadratic irrationals, and Pell solutions. Transformation ancestry should be retained so that generated solutions can be verified directly.

A.9 Binary Quadratic Form Reduction and Composition

A form calculator should verify discriminants, reduce forms under explicit boundary conventions, output SL₂(Z) transformation matrices, compose classes, and identify the resulting class-group structure where certified.

A.10 Gaussian Integer and Norm Calculator

The tool should handle norms, Euclidean division, gcds, units, Gaussian prime factorization, and sums-of-two-squares reconstruction. It must distinguish ordinary and Gaussian primality.

A.11 Dirichlet Series and L-function Numerical Tools

Numerical analytic tools should state the represented function, branch, precision, truncation, error enclosure, conductor, and functional equation normalization. A plotted or decimal value without an error model is exploratory evidence.

A.12 Elliptic-Curve Arithmetic Tools

The system should verify nonsingularity, perform point addition and scalar multiplication, count finite-field points, validate subgroup membership, and distinguish rational, finite-field, and p-adic points. Cryptographic modes require input validation and constant-time considerations.

A.13 Proof Certificate and Replay Validator

A validator accepts a theorem-specific certificate and checks it through a small trusted kernel. Examples include Bézout identities, CRT reconstructions, primality chains, factor products, Smith decompositions, form transformations, and interval enclosures.

A.14 Arithmetic EG Frame Builder

The frame builder records target statement, quantifiers, carrier, equivalence, prime support, parameters, boundaries, transports, and intended liftback. It prevents calculations from beginning on an under-typed problem.

A.15 Residue and Counterkernel Registry

The registry stores failed gates, minimal residues, counterexamples, activation conditions, and repairs. Repeated failures with the same structural fingerprint can then be recognized across arithmetic domains.

A.16 Domain-Specific Number-Theory Constraint Validator

The validator checks divisibility hypotheses, unit conditions, characteristic assumptions, convergence regions, derivative gates, coprimality, local compatibility, parameter uniformity, and exact reconstruction. It is domain-aware rather than a generic syntax checker.

A.17 Final Capstone: TYPE → CARRIER → TRANSPORT → DEBT → RESIDUE → COUNTERKERNEL → LIFTBACK → REPLAY/CERT

The capstone selects one substantial theorem or algorithm and prosecutes the full sequence. Suitable examples include certified primality, singular Hensel lifting, quadratic-form class computation, RSA shared-prime detection, elliptic-curve descent, or an explicit prime-counting estimate. The final submission must contain the exact theorem, carrier definitions, construction, proof, computation where applicable, counterkernel suite, liftback, independent replay, and scoped certificate or frontier payload.

This establishes the complete continuous course text at dense textbook level while preserving the exact upgraded TOC and integrating Exposure Geometry into the mathematical substance rather than treating it as commentary.