Leanpub Header

Skip to main content
Go to Leanpub.com

From Stochastic Chaos to Deterministic Certainty

From Stochastic Chaos to Deterministic Certainty
This book is 100% completeLast updated on 2026-09-14
Grigorios Agathangelidis

Stop choosing between an AI that's powerful and one you can certify. From Shannon's entropy to a reproducible, auditable Industry Language Model — built in C#, wrapped in Deterministic Islands, with every number independently verified.

Minimum price

$9.99

$12.00

You pay

Author earns

$

Also available for 1 book credit with a Reader Membership

PDF
About

About

About the Book

A Stochastic Model and an Absolute Guarantee, in One Architecture

Most books about AI in industry pick a side. Either they teach the mathematics of deep learning in full rigour and leave "what happens when a regulator asks for proof" as an exercise for later — or they teach deployment and governance and treat the model itself as an unopenable black box. Neither gives an engineer at a nuclear plant, a grid operator, or a hospital what they actually need: a system that is both genuinely intelligent and capable of an absolute, certifiable guarantee about the one class of question where "probably right" is not good enough.

This book builds that system from first principles. It begins with Claude Shannon asking how much information a noisy channel can carry, and ends with a quantum circuit's hash sitting in the same lookup table as a pump's rated pressure. In between, it constructs — with full mathematical rigour, real C# code, and dozens of independently verified worked examples — an Industry Language Model (ILM): a Transformer-based system wrapped in Deterministic Islands, a formally proven mechanism that lets a small, enumerable class of safety-critical facts be answered identically, every time, no matter what the model's stochastic side has learned or how it has been retrained.

Every worked number in this book was computed, not guessed — pump pressures, gradient-descent iterations, Bell-state measurement statistics, SHA-256 collision probabilities — and checked against a script before it went on the page. Where the book cites a regulation or a hardware milestone, it was verified against current sources, not recalled from memory.

The book does not stop at the model. It formalises the entire architecture in Domain-Driven Design, applies the identical discipline to a second Bounded Context (quantum computing) to prove the method generalises, wires everything into a real, runnable C# solution with repositories, application services, and a test suite — and closes by mapping the whole thing against the EU AI Act, the NIST AI RMF, and the sector-specific frameworks of the IAEA, EASA, and FDA, current as of 2026.

What the Book Covers

  • Information theory and statistics, built for AI — Shannon entropy and the Noisy Channel Coding Theorem, a nine-level staged treatment of probability from Kolmogorov's axioms to Bayesian inference, bias-variance and regularisation.
  • Neural networks from first principles — the perceptron and Novikoff's convergence proof, XOR and the Universal Approximation Theorem, backpropagation derived by hand on a real 2-2-1 network, then CNNs, RNNs/LSTMs, and the full Transformer with a worked attention computation.
  • The Industry Language Model and Deterministic Islands — why a general LLM hallucinates and cannot be certified; the Static Vault, Frozen Snapshot, and Intra-Vector Routing mechanisms, the Arbiter, and a formal, proven definition of what "deterministic" means for a stochastic system.
  • Applications in critical infrastructure — predictive maintenance, anomaly detection, control and optimisation (with a Shield that reuses the Deterministic Island architecture directly), decision support, and safety and compliance, grounded in current 2025–2026 regulatory practice.
  • Software modeling — DDD in C# — Entities, Value Objects, Aggregates, Repositories, and Domain Events, applied twice: once to quantum computing, once to the full AI/ILM architecture, then integrated into one real solution.
  • Towards Artificial Certainty — the mathematics behind the guarantee (birthday-bound collision resistance, IEEE-754 reproducibility), a Neural Constitution naming the system's non-negotiable rules, and a forward look at Deterministic Islands for quantum neural networks.

How It Is Written

Every chapter follows the same honest blueprint. It opens with "What You Already Know" — a bridge from an engineer's existing intuition (a feedback loop, a conservation law, a state machine) into the new material, so nothing is ever presented as if it fell from the sky. It climbs through the concept with worked examples computed and verified inline, not merely asserted. And it closes with an Honest Boundary: a section that states plainly what the chapter's methods do not guarantee, where the open engineering problems still are, and which claims are mathematically certain versus which are current best practice, subject to change.

Honest Boundary, borrowed from the book itself: "Verification is not proof, for the system as a whole… Certification is a process, not a stamp… no regulatory approval, however rigorously obtained, substitutes for the continuous evaluation, monitoring, and accountable human judgement this entire book has argued for."

Written for engineers, data scientists, physicists, and technical leaders with a university-level exact-science background — no prior machine learning or software architecture assumed. If you can follow a proof, read a block of code, and reason about a system's failure modes, you have what this book requires to start on page one.

Part of the NEXUS-1 series, and a direct companion to From Grid to Core (Volume I) and From Core to Quantum (Volume II) — Chapter 21 revisits the quantum-mechanical Bounded Context built in Volume II and applies this book's own Domain-Driven Design discipline to it, side by side with the AI Context this volume builds.

Share this book

Categories

Artificial IntelligenceC#Neural NetworksQuantum ComputingSoftware Architecturenuclear physics

Feedback

Email the Author

Author

About the Author

Grigorios Agathangelidis

I am a software and .NET engineer, not an AI safety researcher and not a career regulatory specialist — and I say so here rather than leave it for a reader to discover.

From Stochastic Chaos to Deterministic Certainty grew out of the same impulse as the rest of the NEXUS-1 series: take a domain I do not start out an expert in, learn its fundamentals honestly and rigorously, and then bring the disciplines I do know — software architecture, systems engineering, and a stubborn insistence on checking every number — to make that domain legible to other engineers. Every worked example in this book was computed and verified before it was written down; every regulatory claim was checked against a current source, not recalled from memory; and every Honest Boundary says plainly where the settled engineering stops and the open problem begins.

The architecture this book builds — Deterministic Islands, the Arbiter, the Neural Constitution — is a teaching model. It is not validated, licensed, or certified, and is not intended for operational use in any real facility without the certification process the book itself describes in full. The point of this book is understanding the pattern well enough to build the real thing properly, not shortcutting the real thing.

Contents

Table of Contents

  • Introduction: The Bridge from Quantum to Intelligence (p. iii)
  • Part I: Foundations — Information, Uncertainty, Probability (p. 1)
    • Chapter 1: Information, Uncertainty, and the Shannon Limit (p. 2)
      • 1.1 What You Already Know (p. 2)
      • 1.2 The Birth of Information Theory (p. 2)
        • 1.2.1 The Problem of Compression and Transmission (p. 2)
        • 1.2.2 Information as "Surprise" (p. 2)
        • 1.2.3 Entropy (p. 3)
      • 1.3 The Shannon Theorem — A Detailed Treatment (p. 4)
        • 1.3.1 The Noisy Channel Coding Theorem (p. 4)
        • 1.3.2 Interpretation (p. 4)
        • 1.3.3 The Proof, Skeletally (p. 5)
        • 1.3.4 The Converse Theorem (p. 5)
        • 1.3.5 Worked Examples: AWGN, BSC, and the Shannon Limit in dB (p. 5)
        • 1.3.6 Connection to Critical Infrastructure (p. 6)
      • 1.4 Rate-Distortion Theory (p. 6)
        • 1.4.1 The Theorem (p. 7)
        • 1.4.2 Lossy Compression (p. 7)
        • 1.4.3 Connection to Feature Extraction (p. 7)
      • 1.5 The Shannon Theorem as the Forerunner of AI (p. 7)
        • 1.5.1 Entropy as a Loss Function (p. 7)
        • 1.5.2 Mutual Information as a Feature Selection Criterion (p. 8)
        • 1.5.3 The Channel as a Neuron Model (p. 8)
        • 1.5.4 The Shannon Limit as a Learning Limit (p. 8)
        • 1.5.5 The Information Bottleneck (p. 8)
      • 1.6 Worked Example: Shannon in an Industrial Sensor (p. 9)
    • Chapter 2: Probability and Statistics — A Staged Introduction (p. 11)
      • 2.1 What You Already Know (p. 11)
      • 2.2 Level 1 — Intuition: Sample Spaces, Events, and Kolmogorov's Axioms (p. 11)
      • 2.3 Level 2 — Conditional Probability and Bayes' Theorem (p. 12)
      • 2.4 Level 3 — Random Variables and Distributions (p. 13)
      • 2.5 Level 4 — Multivariate Distributions (p. 14)
      • 2.6 Level 5 — Statistical Inference (p. 14)
      • 2.7 Level 6 — Hypothesis Testing (p. 14)
      • 2.8 Level 7 — Stochastic Processes (p. 15)
      • 2.9 Level 8 — Information Theory Meets Statistics (p. 15)
      • 2.10 Level 9 — Towards Machine Learning (p. 16)
      • 2.11 Worked Examples (p. 17)
    • Chapter 3: From Data to Models (p. 19)
      • 3.1 What You Already Know (p. 19)
      • 3.2 Maximum Likelihood and Maximum A Posteriori Estimation, Applied (p. 19)
      • 3.3 The Bias– Variance Tradeoff (p. 20)
      • 3.4 Cross-Validation and Regularization (p. 21)
        • 3.4.1 Cross-Validation (p. 21)
        • 3.4.2 Regularization (p. 21)
        • 3.4.3 The Closed-Form Ridge Solution (p. 21)
        • 3.4.4 Why L1 Produces Sparsity: The Geometric Picture (p. 22)
      • 3.5 The Minimum Description Length Principle (p. 22)
      • 3.6 Learning as Compression: The Shannon Connection Revisited (p. 23)
      • 3.7 The Curse of Dimensionality and the No Free Lunch Theorem (p. 24)
        • 3.7.1 The Curse of Dimensionality (p. 24)
        • 3.7.2 The No Free Lunch Theorem (p. 25)
    • Chapter 4: Towards Machine Learning (p. 26)
      • 4.1 What You Already Know (p. 26)
      • 4.2 Linear Regression as a Single Neuron (p. 26)
      • 4.3 Logistic Regression as a Classifier (p. 27)
        • 4.3.1 The Log-Odds Interpretation (p. 28)
        • 4.3.2 The Cross-Entropy Loss, Concretely (p. 29)
      • 4.4 Gradient Descent: From Calculus to Optimization (p. 29)
        • 4.4.1 Batch, Stochastic, and Mini-Batch Gradient Descent (p. 30)
      • 4.5 The Problem of Non-linearity: Why a Single Layer Is Not Enough (p. 31)
      • 4.6 Introduction to Neural Networks: The Perceptron as a Building Block (p. 32)
  • Part II: Neural Networks — From Perceptron to Deep Networks (p. 34)
    • Chapter 5: The Perceptron — The Foundation (p. 35)
      • 5.1 What You Already Know (p. 35)
      • 5.2 The Biological Neuron (p. 35)
      • 5.3 The McCulloch-Pitts Model (p. 36)
      • 5.4 The Artificial Neuron (Perceptron) (p. 36)
        • 5.4.1 Inputs, Weights, and Bias (p. 36)
        • 5.4.2 Linear Combination (p. 36)
        • 5.4.3 Activation: The Step Function (p. 36)
        • 5.4.4 Output (p. 36)
      • 5.5 The Perceptron Learning Algorithm (p. 37)
        • 5.5.1 The Update Rule (p. 37)
        • 5.5.2 Geometric Interpretation: Rotating the Hyperplane (p. 37)
        • 5.5.3 The Perceptron Convergence Theorem (p. 37)
      • 5.6 Worked Example: Linear Separation in Industrial Data (p. 38)
      • 5.7 The Limitation: The XOR Problem (p. 39)
        • 5.7.1 The Perceptron Cannot Solve XOR (p. 40)
        • 5.7.2 Proof: XOR Is Not Linearly Separable (p. 40)
        • 5.7.3 Geometric Interpretation, Restated (p. 40)
        • 5.7.4 Historical Note: Minsky, Papert, and the First AI Winter (p. 40)
    • Chapter 6: Solving XOR — Hidden Layers and Non-linearity (p. 42)
      • 6.1 What You Already Know (p. 42)
      • 6.2 The Solution to XOR (p. 42)
        • 6.2.1 XOR as a Composition of Simpler Gates (p. 42)
        • 6.2.2 Two-Layer Implementation (p. 43)
        • 6.2.3 Visualization: Transforming the Input Space (p. 44)
      • 6.3 Non-linear Activation Functions (p. 44)
        • 6.3.1 Why Non-linearity Is Essential (p. 45)
      • 6.4 The Universal Approximation Theorem (p. 45)
        • 6.4.1 Statement (p. 45)
        • 6.4.2 Proof Sketch, via Step Functions (p. 45)
      • 6.5 Worked Example: XOR in C# (p. 46)
      • 6.6 From XOR to Deep Networks (p. 48)
        • 6.6.1 Hierarchical Feature Learning (p. 48)
        • 6.6.2 Each Layer, a More Abstract Representation (p. 49)
        • 6.6.3 Examples: From Pixels to Objects (p. 49)
    • Chapter 7: Backpropagation — The Engine of Learning (p. 50)
      • 7.1 What You Already Know (p. 50)
      • 7.2 The Backpropagation Algorithm (p. 50)
        • 7.2.1 Forward Pass: Computing the Output (p. 50)
        • 7.2.2 Loss Functions: MSE and Cross-Entropy, Recalled (p. 50)
        • 7.2.3 Backward Pass: Computing Gradients (p. 51)
        • 7.2.4 Weight Update (p. 51)
      • 7.3 Detailed Derivation for a 2-2-1 Network (p. 51)
        • 7.3.1 Step-by-Step Gradient Computation (p. 51)
        • 7.3.2 Table of Variables and Partial Derivatives (p. 52)
      • 7.4 Worked Example: Training XOR with Backprop in C# (p. 52)
        • 7.4.1 Code, Annotated (p. 52)
        • 7.4.2 Visualization of the Loss Evolution (p. 54)
        • 7.4.3 Final Weight Values (p. 54)
      • 7.5 Challenges (p. 55)
        • 7.5.1 Vanishing and Exploding Gradients (p. 55)
        • 7.5.2 Local Minima (p. 55)
        • 7.5.3 Overfitting (p. 56)
        • 7.5.4 Learning Rate Selection (p. 56)
      • 7.6 Optimization Techniques (p. 57)
        • 7.6.1 Momentum and Nesterov Momentum (p. 57)
        • 7.6.2 AdaGrad, RMSProp, and Adam (p. 57)
        • 7.6.3 Batch Normalization (p. 57)
        • 7.6.4 Dropout (p. 57)
    • Chapter 8: Modern Architectures (p. 59)
      • 8.1 What You Already Know (p. 59)
      • 8.2 Convolutional Neural Networks (CNNs) (p. 59)
        • 8.2.1 Locality, Weight Sharing, and Pooling (p. 59)
        • 8.2.2 Applications: Image and Signal Processing in Industry (p. 61)
      • 8.3 Recurrent Neural Networks (RNNs) (p. 61)
        • 8.3.1 Sequences, Hidden State, and Backpropagation Through Time (p. 61)
        • 8.3.2 LSTM and GRU: Gates and Memory (p. 61)
        • 8.3.3 Applications: Time-Series Prediction and Anomaly Detection (p. 62)
      • 8.4 The Attention Mechanism (p. 62)
        • 8.4.1 The Key-Value-Query Formulation (p. 62)
        • 8.4.2 Self-Attention and Multi-Head Attention (p. 63)
        • 8.4.3 Positional Encoding (p. 63)
      • 8.5 The Transformer Architecture (p. 64)
        • 8.5.1 Encoder-Decoder Structure (p. 64)
        • 8.5.2 Why Transformers Dominate Modern AI (p. 64)
        • 8.5.3 Applications: Language Models, Time-Series, and Multimodal Data (p. 64)
      • 8.6 Autoencoders and Variational Autoencoders (VAEs) (p. 65)
      • 8.7 Generative Adversarial Networks (GANs) (p. 65)
      • 8.8 Architecture Selection for Industrial Data (p. 66)
    • Chapter 9: Supervised, Unsupervised, and Reinforcement Learning (p. 67)
      • 9.1 What You Already Know (p. 67)
      • 9.2 Supervised Learning (p. 67)
      • 9.3 Unsupervised Learning (p. 68)
        • 9.3.1 Clustering (p. 68)
        • 9.3.2 Dimensionality Reduction (p. 69)
        • 9.3.3 Anomaly Detection (p. 69)
      • 9.4 Reinforcement Learning (p. 70)
        • 9.4.1 Markov Decision Processes (p. 70)
        • 9.4.2 Q-Learning and Policy Gradients (p. 70)
        • 9.4.3 Applications in Control: When RL Is Appropriate, and When It Is Dangerous (p. 71)
      • 9.5 Semi-supervised and Self-supervised Learning (p. 71)
      • 9.6 Transfer Learning and Fine-Tuning (p. 72)
      • 9.7 Applications in Critical Infrastructure (p. 72)
  • Part III: The Industry Language Model (ILM) (p. 74)
    • Chapter 10: Why an ILM? (p. 75)
      • 10.1 What You Already Know (p. 75)
      • 10.2 The Limits of General LLMs in Critical Applications (p. 75)
        • 10.2.1 Hallucinations: The Model "Confabulates" Facts (p. 75)
        • 10.2.2 Lack of Domain Knowledge (p. 75)
        • 10.2.3 Non-Deterministic Behaviour (p. 76)
        • 10.2.4 Inability to Verify (p. 76)
        • 10.2.5 Safety: One Mistake Can Be Catastrophic (p. 77)
      • 10.3 Definition of the ILM (Industry Language Model) (p. 77)
        • 10.3.1 Data (p. 77)
        • 10.3.2 Goal (p. 77)
        • 10.3.3 Requirements: Reliability, Verifiability, Determinism Where Needed (p. 77)
      • 10.4 Architecture of the ILM (p. 77)
        • 10.4.1 Transformer-Based (p. 78)
        • 10.4.2 Adaptations (p. 78)
      • 10.5 Training Data (p. 80)
        • 10.5.1 Sources (p. 80)
        • 10.5.2 Preprocessing (p. 80)
        • 10.5.3 Handling Sparse Data and Class Imbalance (p. 80)
        • 10.5.4 Data Ethics and Security (p. 81)
    • Chapter 11: Training and Adaptation of the ILM (p. 82)
      • 11.1 What You Already Know (p. 82)
      • 11.2 Pre-training (p. 82)
      • 11.3 Masked Language Modeling and Causal Language Modeling (p. 82)
      • 11.4 Fine-Tuning (p. 83)
      • 11.5 Parameter-Efficient Fine-Tuning: LoRA and Adapters (p. 83)
      • 11.6 Retrieval-Augmented Generation (p. 84)
      • 11.7 Prompt Engineering for Industrial Contexts (p. 85)
      • 11.8 Evaluation: Perplexity, BLEU, ROUGE, and Human Evaluation (p. 85)
      • 11.9 Ethical Issues: Bias, Fairness, and Transparency (p. 86)
    • Chapter 12: Deterministic Islands (p. 88)
      • 12.1 What You Already Know (p. 88)
      • 12.2 The Problem (p. 88)
        • 12.2.1 Neural Networks Are Stochastic (p. 88)
        • 12.2.2 Two Identical Inputs, Two Different Outputs (p. 88)
        • 12.2.3 In Critical Applications, This Is Unacceptable (p. 88)
      • 12.3 The Deterministic Islands (p. 89)
        • 12.3.1 Static Vault: Hash-Based Routing (p. 89)
        • 12.3.2 Frozen Snapshot: Locked Software, Seed, and CPU Inference (p. 90)
        • 12.3.3 Intra-Vector Routing: Islands the Model Creates for Itself (p. 91)
      • 12.4 Architecture of the Islands (p. 93)
        • 12.4.1 How They Are Integrated into the ILM (p. 93)
        • 12.4.2 Code Examples: The Arbiter (p. 93)
        • 12.4.3 Conflict Resolution Between Islands (p. 94)
        • 12.4.4 Updating Islands: When and How (p. 94)
      • 12.5 Worked Example: An Island in Industrial Control (p. 95)
    • Chapter 13: Evaluation and Verification of the ILM (p. 96)
      • 13.1 What You Already Know (p. 96)
      • 13.2 Test Sets, Validation Sets, Cross-Validation (p. 96)
      • 13.3 Metrics: Accuracy, Precision, Recall, F1, and AUC-ROC (p. 96)
      • 13.4 Robustness: Adversarial Examples and Distribution Shift (p. 98)
      • 13.5 Explainability: SHAP, LIME, and Attention Visualization (p. 99)
      • 13.6 Uncertainty Quantification (p. 99)
      • 13.7 Formal Verification: Specifications and Model Checking (p. 100)
      • 13.8 Certification: Standards (p. 100)
    • Chapter 14: Ethics, Safety, and Governance (p. 102)
      • 14.1 What You Already Know (p. 102)
      • 14.2 Bias and Fairness (p. 102)
      • 14.3 Transparency and Explainability (p. 103)
      • 14.4 Security: Adversarial Attacks and Data Poisoning (p. 103)
      • 14.5 Privacy: Federated Learning and Differential Privacy (p. 104)
      • 14.6 Regulatory Framework (p. 104)
      • 14.7 Human Oversight (p. 105)
      • 14.8 Responsibility and Accountability (p. 105)
  • Part IV: Applications in Critical Infrastructure (p. 107)
    • Chapter 15: Predictive Maintenance (p. 108)
      • 15.1 What You Already Know (p. 108)
      • 15.2 Data: Vibration, Temperature, Pressure, Current (p. 108)
      • 15.3 Feature Engineering: Time-Domain and Frequency-Domain (p. 109)
        • 15.3.1 Time-Domain Features (p. 109)
        • 15.3.2 Frequency-Domain Features (p. 110)
      • 15.4 Models: RNN, LSTM, and Transformer (p. 110)
      • 15.5 Remaining Useful Life (RUL) Estimation (p. 110)
      • 15.6 Examples: Turbines, Pumps, and Transformers (p. 111)
      • 15.7 Predictive Maintenance Inside the ILM (p. 111)
    • Chapter 16: Anomaly Detection (p. 113)
      • 16.1 What You Already Know (p. 113)
      • 16.2 What Is an Anomaly? Point, Contextual, Collective (p. 113)
      • 16.3 Methods: Statistical, Distance-Based, Density-Based, Reconstruction-Based (p. 113)
        • 16.3.1 Statistical Methods (p. 114)
        • 16.3.2 Distance-Based and Density-Based Methods (p. 114)
        • 16.3.3 Reconstruction-Based Methods (p. 114)
      • 16.4 Autoencoders for Anomaly Detection (p. 115)
      • 16.5 One-Class SVM and Isolation Forest (p. 115)
      • 16.6 Deep Learning: GANs and VAEs (p. 116)
      • 16.7 Evaluation: Precision@k and AUPRC (p. 116)
      • 16.8 Examples: Leak Detection, Sensor Failure, and Cyberattack (p. 117)
      • 16.9 Anomaly Detection Inside the ILM (p. 117)
    • Chapter 17: Control and Optimization (p. 119)
      • 17.1 What You Already Know (p. 119)
      • 17.2 Model Predictive Control (MPC) with Neural Networks (p. 119)
      • 17.3 Reinforcement Learning for Control (p. 120)
      • 17.4 Safe RL: Constrained RL and Shielding (p. 120)
      • 17.5 Digital Twins (p. 122)
      • 17.6 Examples: Reactor Control, Combustion, and Energy Management (p. 122)
    • Chapter 18: Decision Support (p. 123)
      • 18.1 What You Already Know (p. 123)
      • 18.2 Decision Support Systems (p. 123)
      • 18.3 Human-AI Collaboration (p. 124)
      • 18.4 Explainable AI for Decisions (p. 124)
      • 18.5 Scenario Analysis (p. 124)
      • 18.6 Examples: Fault Diagnosis, Maintenance Planning, and Crisis Management (p. 125)
    • Chapter 19: Safety and Compliance (p. 127)
      • 19.1 What You Already Know (p. 127)
      • 19.2 Regulations: Nuclear (IAEA), Aerospace (FAA, EASA), Medical (FDA) (p. 127)
      • 19.3 Certification of AI Systems (p. 128)
      • 19.4 Safety Cases (p. 128)
      • 19.5 Auditing and Traceability (p. 129)
      • 19.6 Examples: AI Certification for Medical Devices and Autonomous Vehicles (p. 129)
  • Part V: Software Modeling — DDD in C# (p. 131)
    • Chapter 20: Introduction to Software Modeling for Engineers (p. 132)
      • 20.1 What You Already Know (p. 132)
      • 20.2 Why Model Physics and AI with Code? (p. 132)
        • 20.2.1 Simulations and Digital Twins (p. 132)
        • 20.2.2 Educational Tools and Interactive Learning (p. 132)
        • 20.2.3 Data Analysis and Experimental Validation (p. 132)
        • 20.2.4 Control Automation and Safety (p. 133)
      • 20.3 Basic Principles of Object-Oriented Programming (OOP) in C# (p. 133)
      • 20.4 Introduction to Domain-Driven Design (DDD) (p. 134)
        • 20.4.1 Ubiquitous Language (p. 134)
        • 20.4.2 Bounded Contexts (p. 134)
        • 20.4.3 Entities vs. Value Objects (p. 134)
        • 20.4.4 Aggregates and Aggregate Roots (p. 135)
        • 20.4.5 Repositories (p. 136)
        • 20.4.6 Domain Services (p. 136)
        • 20.4.7 Domain Events (p. 136)
      • 20.5 The Translation Table: Physics $ $ DDD (p. 137)
    • Chapter 21: Bounding the Quantum Context (Revisited) (p. 138)
      • 21.1 What You Already Know (p. 138)
      • 21.2 Value Objects: StateVector, Operator, Basis (p. 138)
      • 21.3 Entities: Qubit and QuantumSystem (p. 139)
      • 21.4 Modeling Quantum Collapse and Entanglement (p. 140)
      • 21.5 Code Example: Gate Application, Measurement, and Time Evolution (p. 141)
    • Chapter 22: Bounding the AI Context (p. 143)
      • 22.1 What It Includes: Models, Training, Inference, Prompts, Deterministic Islands (p. 143)
      • 22.2 Domain Terminology (p. 143)
      • 22.3 Value Objects (p. 143)
      • 22.4 Entities (p. 144)
      • 22.5 Aggregate and the ILM Aggregate Root (p. 145)
      • 22.6 Domain Services (p. 146)
      • 22.7 Domain Events (p. 147)
      • 22.8 Code Example: Building a Minimal ILM in C# (p. 148)
    • Chapter 23: Integration and Application Architecture (p. 150)
      • 23.1 Connecting the Three Contexts (p. 150)
      • 23.2 Use of Repositories (p. 150)
      • 23.3 Application Creation (p. 151)
        • 23.3.1 Application Layer: Coordination of Domain Services (p. 151)
        • 23.3.2 Infrastructure Layer: Repository Implementation, Logging, Monitoring (p. 152)
        • 23.3.3 Presentation Layer: CLI or Web API (p. 152)
      • 23.4 Design Patterns Used (p. 152)
      • 23.5 Creating a Console or API for Interacting with the Model (p. 153)
      • 23.6 Testing (p. 154)
      • 23.7 Future Extensions: Digital Twin of the Reactor (p. 155)
  • Part VI: Towards Artificial Certainty (p. 157)
    • Chapter 24: Deterministic Islands — Deep Dive (p. 158)
      • 24.1 What You Already Know (p. 158)
      • 24.2 Mathematical Foundations (p. 158)
        • 24.2.1 Hashing: SHA-256 and Collision Resistance (p. 158)
        • 24.2.2 Reproducibility: Seeds, Determinism, and Bitwise Accuracy (p. 159)
        • 24.2.3 Formal Definition of a Deterministic Island (p. 159)
      • 24.3 Implementation, Revisited Through This Lens (p. 160)
      • 24.4 Management (p. 160)
        • 24.4.1 Updating and Extending Islands (p. 160)
        • 24.4.2 Conflict Resolution, Recalled (p. 160)
      • 24.5 Code Examples in C# (p. 161)
      • 24.6 Case Study: A Deterministic Island for Reactor Safety (p. 161)
    • Chapter 25: Governance and Regulation (p. 163)
      • 25.1 What You Already Know (p. 163)
      • 25.2 Regulatory Frameworks, Mapped to This Book's Architecture (p. 163)
        • 25.2.1 EU AI Act (p. 163)
        • 25.2.2 NIST AI Risk Management Framework (p. 163)
        • 25.2.3 Sector-Specific Regulations (p. 164)
      • 25.3 Responsibility and Accountability, Made Queryable (p. 164)
      • 25.4 Ethical Issues, Extended: Autonomous Decision-Making (p. 165)
    • Chapter 26: The Future — Artificial Certainty (p. 166)
      • 26.1 What You Already Know (p. 166)
      • 26.2 The Vision: AI That Is Not Merely Intelligent, but Reliable (p. 166)
      • 26.3 Deterministic Islands as the Foundation of Trustworthy AI (p. 166)
      • 26.4 The Neural Constitution: Rules, Constraints, Values (p. 167)
      • 26.5 Ethical AI (p. 168)
      • 26.6 The Future of Critical Infrastructure (p. 168)
      • 26.7 Quantum Neural Networks with Deterministic Islands (p. 168)
        • 26.7.1 Introduction to Quantum Neural Networks (QNNs) (p. 168)
        • 26.7.2 The Challenge of Noise and Decoherence (p. 168)
        • 26.7.3 Deterministic Islands in the Quantum Regime (p. 169)
        • 26.7.4 A Hybrid Architecture: Classical, Quantum, and Deterministic (p. 169)
        • 26.7.5 Potential Applications (p. 170)
        • 26.7.6 The Road Ahead (p. 170)
      • 26.8 The Long View: From Artificial Intelligence to Artificial Certainty (p. 171)
  • Epilogue: From Certainty to Responsibility (p. 172)
  • Appendix A.1: Mathematical Review (p. 174)
  • Appendix A.2: Table of Physical Constants (p. 176)
  • Appendix A.3: Glossary of Terms (p. 177)
  • Appendix A.4: Keepin Six-Group Delayed Neutron Data (p. 183)
  • Appendix A.5: Architecture of a Deterministic Island (p. 184)
  • Appendix A.6: Comparison Table of AI Paradigms in Critical Systems (p. 185)
  • Appendix A.7: Installation and Execution Guide for the Code (C#) (p. 186)
  • Bibliography (p. 188)
  • Index (p. 190)

Get the free sample chapters

Click the buttons to get the free sample in PDF or EPUB, or read the sample online here

Download Sample PDF

The Leanpub 60 Day 100% Happiness Guarantee

Within 60 days of purchase you can get a 100% refund on any Leanpub purchase, in two clicks.

See full terms...

Earn $8 on a $10 Purchase, and $16 on a $20 Purchase

We pay 80% royalties on purchases of $7.99 or more, and 80% royalties minus a 50 cent flat fee on purchases between $0.99 and $7.98. You earn $8 on a $10 sale, and $16 on a $20 sale. So, if we sell 5000 non-refunded copies of your book for $20, you'll earn $80,000.

(Yes, some authors have already earned much more than that on Leanpub.)

In fact, authors have earned over $15 million writing, publishing and selling on Leanpub.

Learn more about writing on Leanpub

Free Updates. DRM Free.

If you buy a Leanpub book, you get free updates for as long as the author updates the book! Many authors use Leanpub to publish their books in-progress, while they are writing them. All readers get free updates, regardless of when they bought the book or how much they paid (including free).

Most Leanpub books are available in PDF (for computers) and EPUB (for phones, tablets and Kindle). The formats that a book includes are shown at the top right corner of this page.

Finally, Leanpub books don't have any DRM copy-protection nonsense, so you can easily read them on any supported device.

Learn more about Leanpub's ebook formats and where to read them

Write and Publish on Leanpub

You can use Leanpub to easily write, publish and sell in-progress and completed ebooks and online courses!

Leanpub is a powerful platform for serious authors, combining a simple, elegant writing and publishing workflow with a store focused on selling in-progress ebooks.

Leanpub is a magical typewriter for authors: just write in plain text, and to publish your ebook, just click a button. (Or, if you are producing your ebook your own way, you can even upload your own PDF and/or EPUB files and then publish with one click!) It really is that easy.

Learn more about writing on Leanpub