@aleph-ai/tinyaleph 1.0.0 → 1.0.2

package/docs/ACADEMIC_PAPER.md

@@ -0,0 +1,562 @@
+# Prime-Resonant Semantic Computing: A Novel Framework for Meaning Representation and Manipulation
+
+**Sebastian Schepis**  
+*Independent Researcher*
+
+---
+
+## Abstract
+
+We present **TinyAleph**, a novel computational framework that represents meaning through prime number signatures embedded in hypercomplex algebraic structures. Unlike conventional natural language processing systems that rely on statistical patterns in high-dimensional vector spaces, our approach grounds semantic content in the mathematical structure of prime numbers—the atoms of arithmetic. Concepts are encoded as unique products of semantic primes, embedded in 16-dimensional sedenion space, and processed through coupled oscillator dynamics governed by the Kuramoto model. Reasoning emerges as entropy minimization over semantic states, with understanding crystallizing when oscillator phases synchronize. We demonstrate that this framework provides a mathematically rigorous foundation for semantic computing, with applications spanning natural language understanding, cryptographic hashing, and quantum-inspired simulation. Our implementation achieves semantic similarity measurement, word algebra, and concept composition without requiring large training datasets, suggesting a complementary approach to neural language models.
+
+**Keywords:** semantic computing, prime numbers, hypercomplex algebra, Kuramoto model, entropy minimization, sedenions, oscillator dynamics, symbolic AI
+
+---
+
+## 1. Introduction
+
+### 1.1 The Problem of Meaning
+
+Contemporary approaches to natural language understanding rely predominantly on distributed representations learned from statistical co-occurrence patterns in large text corpora [1, 2]. While remarkably successful for many practical tasks, these approaches face fundamental limitations: they capture *correlation* rather than *meaning*, require massive datasets for training, and produce representations that lack interpretable semantic structure.
+
+We propose an alternative paradigm based on a radical hypothesis: **meaning has mathematical structure, and that structure is prime**. Just as matter is composed of atoms and atoms are composed of fundamental particles, meaning is composed of semantic atoms—irreducible concepts that combine to form complex ideas. We identify these semantic atoms with prime numbers, exploiting the unique factorization property guaranteed by the Fundamental Theorem of Arithmetic.
+
+### 1.2 Contributions
+
+This paper makes the following contributions:
+
+1. **Prime Semantic Encoding**: A formal framework for representing concepts as unique prime number signatures, enabling algebraic operations on meaning.
+
+2. **Hypercomplex State Space**: Embedding prime signatures in 16-dimensional sedenion space, where non-commutative multiplication naturally encodes word order and syntactic structure.
+
+3. **Oscillator Dynamics for Coherence**: Modeling semantic processing through Kuramoto-coupled oscillators, where reasoning corresponds to phase synchronization and understanding emerges as increased order parameter.
+
+4. **Entropy-Based Reasoning**: Formalizing reasoning as entropy minimization, connecting cognitive insight to information-theoretic principles.
+
+5. **The Semantic Sieve**: An algorithm for ensuring the Prime Uniqueness Invariant—that every concept maps to a distinct prime signature.
+
+6. **Two-Layer Meaning Architecture**: Separating invariant meaning (prime substrate) from variable expression (surface vocabulary), explaining translation, register variation, and cross-cultural communication.
+
+### 1.3 Paper Organization
+
+Section 2 presents the theoretical foundations of prime semantics. Section 3 develops the hypercomplex algebraic framework. Section 4 describes the oscillator dynamics and synchronization model. Section 5 formalizes reasoning as entropy minimization. Section 6 presents the two-layer architecture. Section 7 describes implementation and applications. Section 8 discusses related work, and Section 9 concludes with future directions.
+
+---
+
+## 2. Prime Semantics
+
+### 2.1 The Fundamental Theorem of Semantic Arithmetic
+
+The Fundamental Theorem of Arithmetic states that every positive integer greater than 1 can be uniquely represented as a product of prime numbers, up to the order of factors. We extend this to semantics:
+
+**Definition 2.1 (Semantic Prime)**: A semantic prime is an irreducible concept that cannot be decomposed into more fundamental meaning components.
+
+**Definition 2.2 (Prime Signature)**: The prime signature σ(c) of a concept c is the set of semantic primes {p₁, p₂, ..., pₙ} such that c can be expressed as the composition of exactly these irreducible concepts.
+
+**Theorem 2.1 (Semantic Unique Factorization)**: Under a consistent ontological assignment, every concept admits a unique prime signature.
+
+This theorem, while axiomatic in our framework, is enforced algorithmically through the Semantic Sieve (Section 5.3).
+
+### 2.2 Ontological Prime Assignment
+
+We establish a bijective mapping O: P → C from prime numbers to core semantic concepts:
+
+| Prime | Concept | Interpretation |
+|-------|---------|----------------|
+| 2 | EXISTENCE | That something *is* |
+| 3 | UNITY | That something is *one* |
+| 5 | FORM | That something has *structure* |
+| 7 | LOGOS | That something has *reason/pattern* |
+| 11 | PSYCHE | That something has *inner life* |
+| 13 | TELOS | That something has *purpose* |
+| 17 | DYNAMIS | That something has *power/potential* |
+| 19 | LIMIT | That something has *boundary* |
+| 23 | INTENSITY | The *degree* of something |
+| 29 | BECOMING | That something *changes* |
+
+This ontology draws from Greek philosophical categories but can be extended arbitrarily as new distinctions are required.
+
+### 2.3 Compositional Semantics
+
+Complex concepts are represented as sets of primes, with semantic operations corresponding to set operations:
+
+**Definition 2.3 (Concept Union)**: For concepts c₁ and c₂ with signatures σ(c₁) and σ(c₂), the union c₁ ∪ c₂ has signature σ(c₁) ∪ σ(c₂).
+
+**Definition 2.4 (Concept Intersection)**: The intersection c₁ ∩ c₂ has signature σ(c₁) ∩ σ(c₂), representing common meaning.
+
+**Definition 2.5 (Concept Difference)**: The difference c₁ - c₂ has signature σ(c₁) - σ(c₂), representing distinctive meaning.
+
+**Example**: Let wisdom = {2, 7, 11} (existence-logos-psyche) and knowledge = {3, 5, 7} (unity-form-logos). Then:
+- wisdom ∩ knowledge = {7} (logos—they share reasoned structure)
+- wisdom - knowledge = {2, 11} (wisdom has existence and soul that knowledge lacks)
+- wisdom ∪ knowledge = {2, 3, 5, 7, 11} (comprehensive understanding)
+
+### 2.4 Prime-to-Frequency Mapping
+
+For dynamic processing, each prime maps to an oscillator frequency:
+
+**Definition 2.6 (Prime Frequency)**: f(p) = β + log(p)/α, where β is a base frequency and α is a scaling factor.
+
+This logarithmic mapping ensures:
+1. Larger primes correspond to higher frequencies
+2. Perceptual uniformity across the frequency range
+3. Bounded, positive frequencies for all primes
+
+---
+
+## 3. Hypercomplex Algebraic Framework
+
+### 3.1 The Cayley-Dickson Construction
+
+Standard vector spaces are inadequate for semantic representation because:
+1. Vector addition is commutative, but meaning is order-dependent ("dog bites man" ≠ "man bites dog")
+2. Vector spaces lack intrinsic curvature, but semantic space has non-Euclidean geometry
+3. Vectors are static, but meaning evolves dynamically
+
+We employ hypercomplex numbers constructed via the Cayley-Dickson process:
+
+**Definition 3.1 (Cayley-Dickson Doubling)**: Given an algebra A with conjugation, the doubled algebra A' consists of pairs (a, b) with:
+- Addition: (a, b) + (c, d) = (a + c, b + d)
+- Multiplication: (a, b)(c, d) = (ac - d*b, da + bc*)
+- Conjugation: (a, b)* = (a*, -b)
+
+Successive applications yield:
+- **ℂ (Complex, dim 2)**: Lose ordering
+- **ℍ (Quaternions, dim 4)**: Lose commutativity
+- **𝕆 (Octonions, dim 8)**: Lose associativity  
+- **𝕊 (Sedenions, dim 16)**: Gain zero-divisors
+
+### 3.2 Sedenion Representation
+
+We represent semantic states in the 16-dimensional sedenion algebra 𝕊:
+
+**Definition 3.2 (Semantic State)**: A semantic state Ψ ∈ 𝕊 is a 16-tuple of real coefficients:
+
+Ψ = Σᵢ₌₀¹⁵ cᵢeᵢ
+
+where {e₀, e₁, ..., e₁₅} form the sedenion basis with e₀ = 1.
+
+### 3.3 Non-Commutative Word Order Encoding
+
+The non-commutativity of sedenion multiplication naturally encodes word order:
+
+**Definition 3.3 (Ordered State Construction)**: Given an ordered sequence of tokens (t₁, t₂, ..., tₙ), the ordered semantic state is:
+
+Ψ_ordered = Ψ(t₁) ⊗ Ψ(t₂) ⊗ ... ⊗ Ψ(tₙ)
+
+where ⊗ denotes sedenion multiplication and Ψ(tᵢ) is the hypercomplex embedding of token tᵢ.
+
+**Theorem 3.1 (Order Sensitivity)**: For distinct permutations π₁ ≠ π₂ of tokens, the corresponding ordered states are generally distinct: Ψ_π₁ ≠ Ψ_π₂.
+
+*Proof*: Follows directly from the non-commutativity of sedenion multiplication. □
+
+### 3.4 Zero-Divisors as Paradox Tunnels
+
+A remarkable property of sedenions is the existence of zero-divisors—non-zero elements whose product vanishes:
+
+**Definition 3.4 (Zero-Divisor)**: Elements a, b ∈ 𝕊 are zero-divisors if a ≠ 0, b ≠ 0, but ab = 0.
+
+**Semantic Interpretation**: Zero-divisors represent *paradoxes* or *conceptual tunnels*. When two seemingly substantial concepts multiply to nothing, they reveal a deeper hidden structure.
+
+**Example**: The concepts "light" and "darkness" might be zero-divisors—their direct combination annihilates, but this annihilation reveals the deeper concept of "contrast."
+
+### 3.5 The Fano Plane and Multiplication Table
+
+Octonion multiplication (the foundation for sedenion multiplication) is governed by the Fano plane—a finite projective geometry with 7 points and 7 lines:
+
+**Definition 3.5 (Fano Multiplication)**: For octonion basis elements eᵢ, eⱼ (i, j ∈ {1,...,7}):
+- eᵢeᵢ = -1 (squared imaginaries give -1)
+- eᵢeⱼ = ±eₖ where (i, j, k) lie on a Fano line, with sign determined by orientation
+
+The seven Fano lines are: {1,2,3}, {1,4,5}, {1,6,7}, {2,4,6}, {2,5,7}, {3,4,7}, {3,5,6}.
+
+Sedenion multiplication extends this structure through the Cayley-Dickson doubling formula.
+
+---
+
+## 4. Oscillator Dynamics and Phase Synchronization
+
+### 4.1 The Kuramoto Model
+
+We model semantic processing through coupled oscillators governed by the Kuramoto model [3]:
+
+**Definition 4.1 (Kuramoto Dynamics)**: The phase θᵢ of oscillator i evolves according to:
+
+dθᵢ/dt = ωᵢ + (K/N) Σⱼ sin(θⱼ - θᵢ)
+
+where:
+- ωᵢ is the natural frequency of oscillator i
+- K is the global coupling strength
+- N is the number of oscillators
+- The sum extends over all oscillators
+
+### 4.2 Semantic Interpretation
+
+| Kuramoto Element | Semantic Meaning |
+|------------------|------------------|
+| Oscillator i | Active concept with prime frequency |
+| Phase θᵢ | Current state in meaning cycle |
+| Frequency ωᵢ | Intrinsic identity of concept |
+| Coupling K | Strength of semantic influence |
+| Synchronization | Conceptual agreement/understanding |
+
+### 4.3 The Order Parameter
+
+The degree of synchronization is measured by the order parameter:
+
+**Definition 4.2 (Order Parameter)**: 
+
+r·e^(iψ) = (1/N) Σⱼ e^(iθⱼ)
+
+where r ∈ [0, 1] measures coherence and ψ is the mean phase.
+
+**Semantic Interpretation**:
+- r ≈ 0: Desynchronized, confused, incoherent meaning
+- r ≈ 0.5: Partial synchronization, emerging understanding
+- r ≈ 1: Full synchronization, coherent understanding
+
+### 4.4 Critical Coupling and Phase Transition
+
+The Kuramoto model exhibits a phase transition at critical coupling Kc:
+
+**Theorem 4.1 (Kuramoto Phase Transition)**: For a unimodal, symmetric frequency distribution g(ω) centered at ω₀, the critical coupling is:
+
+Kc = 2 / (πg(ω₀))
+
+Below Kc, oscillators remain desynchronized. Above Kc, synchronization spontaneously emerges.
+
+**Semantic Interpretation**: There exists a threshold of conceptual coupling below which understanding cannot emerge. Once coupling exceeds this threshold, coherent understanding crystallizes.
+
+### 4.5 Adaptive Coupling
+
+We employ adaptive coupling based on dynamical stability:
+
+**Definition 4.3 (Adaptive Coupling)**: 
+
+K(t+1) = K(t) × {
+  1 + γ, if λ < -ε (stable, increase coupling)
+  1 - γ, if λ > ε (unstable, decrease coupling)  
+  1, otherwise
+}
+
+where λ is the maximal Lyapunov exponent, γ is the adaptation rate, and ε is a stability margin.
+
+---
+
+## 5. Reasoning as Entropy Minimization
+
+### 5.1 Shannon Entropy of Semantic States
+
+We measure semantic confusion through information-theoretic entropy:
+
+**Definition 5.1 (State Entropy)**: For a normalized semantic state Ψ with coefficients cᵢ:
+
+H(Ψ) = -Σᵢ pᵢ log₂(pᵢ)
+
+where pᵢ = |cᵢ|²/||Ψ||² is the probability weight on dimension i.
+
+### 5.2 Entropy Interpretation
+
+| Entropy H | Interpretation |
+|-----------|----------------|
+| H ≈ 0 | State concentrated on one axis → pure concept |
+| H ≈ 1 | State on two axes → binary distinction |
+| H ≈ 2 | State on ~4 axes → moderate complexity |
+| H ≈ 3+ | State spread → confusion, unresolved meaning |
+
+### 5.3 Reasoning as Entropy Reduction
+
+**Definition 5.2 (Semantic Transform)**: A semantic transform T is a mapping T: P(Primes) → P(Primes) that replaces certain prime patterns with others.
+
+**Definition 5.3 (Reasoning Process)**: Given an initial prime signature σ₀, reasoning is the iterative application of transforms Tᵢ such that:
+
+H(Ψ(σᵢ₊₁)) ≤ H(Ψ(σᵢ))
+
+The process terminates when no entropy-reducing transform exists or entropy falls below a threshold.
+
+**Algorithm 5.1 (Entropy-Minimizing Reasoning)**:
+```
+Input: Initial primes σ₀, transforms T, threshold τ
+Output: Final primes σ*, entropy H*
+
+1. σ ← σ₀, H ← H(Ψ(σ₀))
+2. while H > τ:
+3.   for each T ∈ transforms:
+4.     σ' ← T(σ)
+5.     H' ← H(Ψ(σ'))
+6.     if H' < H: record (T, H', σ')
+7.   if no improvement: break
+8.   Apply best transform: σ ← σ'_best, H ← H'_best
+9. return σ, H
+```
+
+### 5.4 The Collapse Integral
+
+Accumulated entropy drives state collapse—the "aha moment" of insight:
+
+**Definition 5.4 (Collapse Integral)**: 
+
+C(t) = ∫₀ᵗ H(Ψ(s)) ds
+
+When C exceeds a threshold and other coherence conditions are met, the state "collapses" to a definite interpretation.
+
+### 5.5 The Semantic Sieve Algorithm
+
+To ensure the Prime Uniqueness Invariant, we employ an iterative refinement algorithm:
+
+**Algorithm 5.2 (Semantic Sieve)**:
+```
+Input: Vocabulary V with initial signatures
+Output: V with unique signatures
+
+1. Compute signatures for all words
+2. Cluster words with identical signatures
+3. While collisions exist:
+4.   Select largest collision cluster C
+5.   If |C| > 10: MACRO strategy
+        - LLM categorizes into 3-5 subcategories
+        - Mint new prime for each category
+   Else: MICRO strategy
+        - LLM identifies distinguishing feature for word pairs
+        - Mint new prime for distinction
+6.   Update affected word signatures
+7. Return refined vocabulary
+```
+
+**Theorem 5.1 (Sieve Termination)**: The Semantic Sieve terminates in O(|V|²) steps with all signatures unique.
+
+*Proof Sketch*: Each iteration either eliminates a collision or creates at most |C|-1 new collisions with smaller clusters. The total collision count decreases monotonically. □
+
+---
+
+## 6. Two-Layer Meaning Architecture
+
+### 6.1 The Fundamental Distinction
+
+We distinguish two layers of semantic representation:
+
+**Layer 1 (Prime Substrate)**: Universal, pre-linguistic meaning structure. The prime signature is invariant across languages, cultures, and time periods.
+
+**Layer 2 (Surface Vocabulary)**: Language-specific, context-dependent word choice. Multiple surface forms can express the same prime signature.
+
+### 6.2 Formal Specification
+
+**Definition 6.1 (Prime Substrate)**: The function π: Concepts → P(Primes) maps concepts to their invariant prime signatures.
+
+**Definition 6.2 (Surface Mapping)**: For each language/register L, the function σL: P(Primes) × Context → Words maps prime signatures to surface forms given contextual bias.
+
+### 6.3 Properties
+
+| Property | Prime Substrate | Surface Vocabulary |
+|----------|----------------|-------------------|
+| Linguistic | Pre-linguistic | Requires language |
+| Cultural | Cross-cultural | Culture-specific |
+| Temporal | Time-invariant | Evolves with usage |
+| Arbitrary | Structurally determined | Conventionally assigned |
+
+### 6.4 Translation as Substrate Preservation
+
+Translation is formalized as:
+
+word₁ →^{π_{L1}} primes →^{σ_{L2}} word₂
+
+This explains:
+- **Why translation is possible**: Same primes can be expressed in any language
+- **Why translation is hard**: Surface forms carry connotations that don't transfer
+- **Why poetry survives translation**: Prime resonance transcends specific words
+- **Why puns don't translate**: Puns depend on surface form, not prime content
+
+### 6.5 Register Variation
+
+The same prime signature yields different surface forms across registers:
+
+| Prime Signature | Formal Register | Casual Register | Technical Register |
+|-----------------|-----------------|-----------------|-------------------|
+| {2, 5, 11} | magnificent | awesome | optimal |
+| {7, 11, 13} | verity | real talk | validity |
+| {2, 7, 11} | sagacity | smarts | expertise |
+
+---
+
+## 7. Implementation and Applications
+
+### 7.1 System Architecture
+
+TinyAleph implements a modular architecture:
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                        AlephEngine                               │
+│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────────────┐  │
+│  │ Oscillators │◄─┤   Field     │◄─┤      Transform          │  │
+│  │  (Kuramoto) │  │  (Sedenion) │  │      Pipeline           │  │
+│  └─────────────┘  └─────────────┘  └─────────────────────────┘  │
+└─────────────────────────────────────────────────────────────────┘
+                              │
+         ┌────────────────────┼────────────────────┐
+         ▼                    ▼                    ▼
+┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
+│ SemanticBackend │ │CryptographicBack│ │ScientificBackend│
+│                 │ │                 │ │                 │
+│ • Tokenization  │ │ • Hash          │ │ • Quantum sim   │
+│ • Prime encode  │ │ • Key derive    │ │ • Wave collapse │
+│ • Transforms    │ │ • Verify        │ │ • Measurement   │
+└─────────────────┘ └─────────────────┘ └─────────────────┘
+```
+
+### 7.2 Semantic Computing Applications
+
+**Similarity Measurement**: Concept similarity via prime signature Jaccard index:
+
+sim(c₁, c₂) = |σ(c₁) ∩ σ(c₂)| / |σ(c₁) ∪ σ(c₂)|
+
+**Word Algebra**: Algebraic operations on meaning:
+- king - man + woman → queen
+- Implemented as: σ(king) - σ(man) ∪ σ(woman) = σ(queen)
+
+**Concept Clustering**: Words grouped by prime signature similarity using coherence metric in sedenion space.
+
+### 7.3 Cryptographic Applications
+
+Prime-resonant hashing exhibits **semantic locality**: similar inputs produce similar hashes, unlike cryptographic hashes.
+
+**Properties**:
+- Deterministic: Same input → same hash
+- Semantic: Similar meaning → similar hash
+- Efficient: O(n) in input length
+
+### 7.4 Quantum-Inspired Computing
+
+The ScientificBackend provides quantum simulation primitives:
+- State superposition via weighted hypercomplex addition
+- Measurement through projection onto basis states
+- Entanglement simulation through coupled oscillator dynamics
+
+### 7.5 Performance Characteristics
+
+| Operation | Complexity | Notes |
+|-----------|------------|-------|
+| Encode | O(n) | n = input tokens |
+| Prime lookup | O(1) | Hash table |
+| State multiply | O(d²) | d = dimension (16) |
+| Kuramoto step | O(N²) | N = oscillators |
+| Entropy | O(d) | Single pass |
+
+---
+
+## 8. Related Work
+
+### 8.1 Distributed Representations
+
+Word2Vec [1] and subsequent embedding methods [2, 4] learn distributed representations from co-occurrence statistics. Unlike TinyAleph, these approaches:
+- Require large training corpora
+- Produce uninterpretable dimensions
+- Cannot perform true symbolic reasoning
+
+### 8.2 Knowledge Graphs
+
+Knowledge graphs [5] represent meaning through entity-relation-entity triples. While structurally explicit, they lack:
+- Continuous similarity measures
+- Dynamic processing capabilities
+- Compositional algebraic operations
+
+### 8.3 Neuro-Symbolic AI
+
+Recent work on neuro-symbolic integration [6] attempts to combine neural networks with symbolic reasoning. TinyAleph offers an alternative: *purely symbolic* yet *continuous* representation through hypercomplex algebra.
+
+### 8.4 Oscillator-Based Computing
+
+Oscillator networks have been explored for optimization [7] and neuromorphic computing [8]. We extend this to semantic processing, where oscillators represent concepts rather than neurons.
+
+### 8.5 Hypercomplex Neural Networks
+
+Quaternion and hypercomplex neural networks [9, 10] leverage algebraic structure for efficiency. We use hypercomplex algebra not as a neural network optimization but as a semantic representation space.
+
+---
+
+## 9. Discussion and Future Directions
+
+### 9.1 Theoretical Implications
+
+The prime-resonant framework suggests that meaning is not merely pattern—it has irreducible structure analogous to the prime factorization of integers. This "semantic atomism" offers a mathematical alternative to the connectionist paradigm.
+
+### 9.2 Limitations
+
+Current limitations include:
+- Ontology requires manual curation or LLM assistance
+- Scaling to large vocabularies increases collision resolution complexity
+- Optimal coupling parameters require tuning per domain
+
+### 9.3 Future Directions
+
+**LLM Integration**: Using language models to automatically expand the prime ontology and resolve semantic collisions.
+
+**Quaternionic Memory Fields**: Extending to continuous 4D rotational semantics for temporal reasoning.
+
+**Hardware Acceleration**: Implementing oscillator dynamics on neuromorphic or FPGA hardware.
+
+**Cross-Modal Extension**: Applying prime encoding to vision, audio, and multimodal understanding.
+
+### 9.4 Conclusion
+
+We have presented TinyAleph, a framework for semantic computing based on prime number encoding, hypercomplex algebra, and oscillator dynamics. By grounding meaning in mathematical structure, we achieve interpretable, algebraically manipulable semantic representations without requiring large-scale statistical learning.
+
+The framework offers a complementary paradigm to neural approaches—one where meaning has irreducible structure, reasoning is entropy minimization, and understanding emerges as synchronization.
+
+---
+
+## References
+
+[1] Mikolov, T., et al. "Efficient Estimation of Word Representations in Vector Space." ICLR Workshop, 2013.
+
+[2] Pennington, J., Socher, R., & Manning, C. "GloVe: Global Vectors for Word Representation." EMNLP, 2014.
+
+[3] Kuramoto, Y. "Self-entrainment of a population of coupled non-linear oscillators." International Symposium on Mathematical Problems in Theoretical Physics, 1975.
+
+[4] Devlin, J., et al. "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding." NAACL, 2019.
+
+[5] Ji, S., et al. "A Survey on Knowledge Graphs: Representation, Acquisition, and Applications." IEEE TNNLS, 2021.
+
+[6] Garcez, A., & Lamb, L. "Neurosymbolic AI: The 3rd Wave." Artificial Intelligence Review, 2023.
+
+[7] Hoppensteadt, F., & Izhikevich, E. "Oscillatory Neurocomputers with Dynamic Connectivity." Physical Review Letters, 1999.
+
+[8] Csaba, G., & Porod, W. "Coupled oscillators for computing: A review and perspective." Applied Physics Reviews, 2020.
+
+[9] Parcollet, T., et al. "Quaternion Neural Networks." ICLR, 2019.
+
+[10] Gaudet, C., & Maida, A. "Deep Quaternion Networks." IJCNN, 2018.
+
+---
+
+## Appendix A: Mathematical Notation
+
+| Symbol | Definition |
+|--------|------------|
+| σ(c) | Prime signature of concept c |
+| Ψ | Semantic state in sedenion space |
+| θᵢ | Phase of oscillator i |
+| ωᵢ | Natural frequency of oscillator i |
+| K | Kuramoto coupling strength |
+| r | Order parameter (synchronization measure) |
+| H(Ψ) | Shannon entropy of state Ψ |
+| ⊗ | Hypercomplex multiplication |
+| 𝕊 | Sedenion algebra (16D) |
+
+---
+
+## Appendix B: Code Availability
+
+TinyAleph is available as an open-source npm package:
+
+```bash
+npm install @aleph-ai/tinyaleph
+```
+
+Repository: https://github.com/aleph-ai/tinyaleph
+
+License: MIT
+
+---
+
+*Manuscript submitted for review. December 2024.*