Front Matter
9 entries- F1Why This Book ExistsThe evaluation-first stance, the seven-forms method, and why quantum AI is not magic AI.
- F2What This Book CoversThe eleven-part arc from qubits to production-grade quantum AI software.
- F3Who Should Read This BookAnyone who wants both the theory and the build: AI/ML engineers, researchers, and practitioners moving from classical to quantum.
- F4Diagnostic Self-TestShort checks on Python, linear algebra, probability, and ML to pick your onramp path.
- F5Notation GuideDirac notation, state vectors, operators, and conventions, all in one place.
- F6Chapter-Dependency GraphA DAG of prerequisites with suggested paths for the three course modes.
- F7Running the Code and Coding With a CopilotOne-click environment setup and the Build-With-AI discipline that runs throughout.
- F8About the AuthorsWho wrote this book and how.
- F9About the Hands-On AI Science SeriesThe nine-book series and where this volume fits.
Foundations · Mathematical Onramp
1 chapterThe mathematics the rest of the book uses: complex numbers, vectors, matrices, tensor products, probability, and Dirac notation.
- 0Mathematical Foundations You Actually Need NewComplex numbers, inner products, unitarity, tensor products, eigenvalues, probability, and gentle Dirac notation.
Part I · The Rules of the Quantum Computing Game
6 chaptersQuantum computing as a rule-based computational model: qubits, measurement, gates, rotation gates, and circuits.
- 1Quantum Computing Without PhysicsThe computational model, not the physics story. Why it matters for AI, and what "Quantum AI" can and cannot mean.
- 2Qubits: The Smallest Quantum Data ObjectThe qubit as a vector of amplitudes. Amplitude vs probability. Normalization. Why amplitudes can be negative or complex.
- 3Measurement: From Quantum State to Classical OutputThe Born rule, sampling, shots, and collapse. Measuring in a chosen basis and expectation values.
- 4Quantum Gates: Legal Moves in the GameUnitarity, reversibility, and the common one-qubit gates: X, Y, Z, H, S, T.
- 5The Bloch Sphere and Rotation Gates NewThe Bloch sphere, RX/RY/RZ as parameterized rotations, and the trainable-weight interpretation that all of Part VI needs.
- 6Quantum Circuits: Programs Made of GatesCircuit notation, translating diagrams to code, circuits as model architectures. PennyLane and Qiskit first builds.
Part II · Essential Quantum Ideas for AI
5 chaptersSuperposition, interference, tensor products, multi-qubit gates, entanglement, and reversibility.
- 7Superposition: Representation Before ComputationSuperposition without exaggeration. Not "all answers at once"; a distributed representation with careful output extraction.
- 8Interference: Making Some Answers LouderAmplitudes adding and canceling. Phase kickback as the recurring mechanism behind algorithms.
- 9Multiple Qubits, Tensor Products, and Multi-Qubit GatesExponential state spaces, tensor products, CNOT, controlled-U, SWAP, and Toffoli.
- 10Entanglement: Dependencies Beyond Classical CorrelationProduct vs entangled states, Bell pairs, measurement dependency, and what distinguishes entanglement from correlation.
- 11Reversibility, No-Cloning, and Information FlowWhy quantum gates are reversible, why unknown states cannot be copied, and what that means for algorithm design.
Part III · Building Quantum Algorithms
7 chaptersThe structure of quantum algorithms and their AI/ML relevance: search, optimization, feature extraction, and objective minimization.
- 12Quantum Algorithm Design PatternPrepare, transform, interfere, measure, repeat. Oracles, phase kickback revisited, uncomputation.
- 13Deutsch-Jozsa AlgorithmThe first clean demonstration of algorithmic advantage: constant vs balanced functions in one query.
- 14Grover Search and Amplitude AmplificationQuadratic speedup for unstructured search. Amplitude amplification as a general primitive.
- 15Quantum Fourier Transform, Phase Estimation, and Shor in BriefPeriodic structure, phase estimation, and the factoring connection (post-quantum security implication noted).
- 16Quantum Linear Algebra and the HHL Algorithm NewHHL, QRAM, the fine print on exponential speedup claims, and the dequantization that followed.
- 17Hamiltonians, Energy, and Objective FunctionsHamiltonians as cost functions. Ground state as optimum. Preparing for VQE and QAOA.
- 18VQE: Variational Quantum EigensolverThe hybrid quantum-classical loop. Ansatz, energy expectation, classical optimizer, and the ML connection.
Part IV · Classical Machine Learning Refresher
4 chaptersA compact ML onramp focused on concepts that map directly to quantum ML. Explicitly skippable for ML-literate readers.
- 19Machine Learning as Function LearningData, labels, models, loss, generalization. Classical baselines for toy datasets.
- 20Feature Spaces and KernelsFeature maps, the kernel trick, SVMs, and kernel matrices. Preparing for quantum kernels.
- 21Neural Networks and Trainable ModelsParameters, gradients, backpropagation. Preparing for variational circuits.
- 22Generative Models and SamplingDistribution learning, generative vs discriminative. Preparing for quantum generative models.
Part V · Encoding Data into Quantum Circuits
3 chaptersThe data-loading problem, feature maps, QRAM, and measurement as feature extraction.
- 23The Data Loading ProblemWhy encoding is not free. Basis, angle, amplitude encoding. QRAM and the open feasibility question.
- 24Quantum Feature MapsData-dependent circuits, data re-uploading, depth, and expressivity.
- 25Measurement as Feature ExtractionObservables, expectation values, shot noise, and classical post-processing.
Part VI · Quantum Machine Learning Models
13 chaptersThe full QML model zoo: kernels, variational circuits, QNNs, gradients, barren plateaus, expressivity, equivariant models, tensor networks, reservoir computing, generative models, and RL.
- 26Quantum KernelsQuantum feature-space similarity, kernel estimation, SVM, and exponential concentration.
- 27Variational Quantum CircuitsParameterized circuits as trainable models, ansatz design, hybrid optimization.
- 28Variational Models Are Kernel Methods NewThe encoding-induced kernel, why ansatz choices matter less than encoding choices, and what this means for advantage.
- 29Quantum Neural NetworksQNN structure, hybrid models, and comparison to classical neural networks.
- 30Gradients, Parameter-Shift, and TrainingThe parameter-shift rule, automatic differentiation, shot noise, and optimization difficulty.
- 31Barren Plateaus and TrainabilityVanishing gradients, the Lie-algebraic DLA criterion (2024), and the barren-plateau/simulability link.
- 32What Can Quantum Models Learn? Expressivity and Generalization NewCircuits as truncated Fourier series. ICLR 2025 generalization bounds vs classical RFF. Sample complexity.
- 33Quantum Convolutional and Structured ModelsQCNNs, local inductive bias, barren-plateau-free design, and the simulability tradeoff (PRX Quantum 2025).
- 34Geometric and Equivariant Quantum Machine Learning NewSymmetry-aware ansatze, equivariant gates, inductive bias, and improved trainability.
- 35Tensor Networks: The Bridge Between Quantum and Classical ML NewMPS, bond dimension, classical tractability, and the tensor-network classifier.
- 36Quantum Reservoir Computing NewFixed quantum reservoirs sidestep barren plateaus. Temporal tasks and linear readouts.
- 37Quantum Generative ModelsBorn machines, quantum Boltzmann machines, distribution learning, and the 68-qubit advantage claim.
- 38Quantum Reinforcement LearningQuantum parameterized policies and optimization inside RL loops.
Part VII · Quantum Optimization for AI
4 chaptersQAOA, QUBO, annealing, and optimization pipelines for AI tasks.
- 39QAOA: Quantum Approximate Optimization AlgorithmCost and mixer Hamiltonians, alternating layers, the classical loop, and a MaxCut example.
- 40QUBO and Ising FormulationsBinary objectives, QUBO, Ising models, and mapping AI problems to energy functions.
- 41Quantum Annealing and Quantum-Inspired OptimizationAnnealing vs gate-based, and where classical heuristics already win.
- 42Optimization in AI PipelinesFeature selection, clustering, hyperparameter search, and resource allocation via QUBO.
Part VIII · Reality Check: Noise, Hardware, and Limits
7 chaptersDensity matrices, noise, error correction, the 2026 hardware landscape (six modalities including topological), and a rigorous framework for evaluating advantage.
- 43Density Matrices and Mixed States NewPure vs mixed states, partial trace, reduced states, and why noise turns pure states mixed.
- 44Simulators Versus Real Quantum HardwareIdeal vs noisy simulation, shots, connectivity, queues, and circuit depth limits.
- 45Noise and Error MitigationNoise channels, decoherence, mitigation techniques, and the mitigation-vs-correction distinction.
- 46Error Correction and the Road to Fault Tolerance NewLogical qubits, threshold theorem, Willow below-threshold, IBM BB qLDPC, Quantinuum Helios 48-logical, AlphaQubit 2.
- 47The 2026 Hardware Landscape NewSix modalities: superconducting, trapped-ion, neutral-atom, photonic, silicon-spin, and topological (Majorana 1).
- 48Scaling Problems and Simulation LimitsExponential state spaces, tensor-network simulation boundaries, and practical limits.
- 49What Would Quantum Advantage Mean? Complexity, Dequantization, and Honest BaselinesBQP, dequantization, classical surrogates, Bowles-Ahmed-Schuld benchmarking, entanglement-induced advantage (2025), robust dequantization (2024).
Part IX · Advanced Topics in Quantum AI
8 chaptersQuantum data and learning separations, QNLP, quantum transformers, graph learning, scientific ML, neural quantum states, AI for quantum, and domain applications.
- 50Quantum Data and Learning Quantum SystemsPower of data, learning from quantum experiments, classical shadows, and quantum dynamics learning.
- 51Quantum Natural Language ProcessingCompositionality, grammar-and-meaning, tensor structure, and sentence circuits.
- 52Quantum Transformers and Attention NewQuantum attention blocks, and the quantum-hardware vs quantum-inspired-classical distinction (applied skeptically).
- 53Quantum Graph LearningQuantum walks, graph kernels, molecular graphs, and graph classification.
- 54Quantum AI for Scientific Machine LearningChemistry, materials, and simulation: the most plausible near-term advantage domain.
- 55Neural Quantum States NewClassical neural networks representing quantum wavefunctions. VMC, transformer wavefunctions, foundation NQS (2025). The baseline for Ch 54.
- 56AI for Quantum ComputingAlphaQubit 2 (near-optimal, real-time decoding), calibration, circuit search, and LLM-assisted coding.
- 57Quantum AI Across Domains NewDrug discovery (fastest-growing), finance, high-energy physics, and industrial optimization, all read through the claim-vs-proven lens.
Part X · Quantum AI Software, Hardware Access, and Research Practice New
4 chaptersThe engineering and research-practice spine: reproducible environments, AI-assisted coding, real-hardware access, GPU/JIT acceleration, benchmarking, and statistics.
- 58The Quantum AI Development Environment and AI-Assisted Coding NewPinned envs, SDK decision map, testing quantum code, CI for notebooks, and the Build-With-AI (vibe coding) discipline.
- 59Compiling, Transpiling, and Running on Real Hardware NewTranspilation, OpenQASM 3 / QIR / primitives, Qiskit 2.3 C++ transpiler, Mitiq, and access recipes for IBM/Braket/IonQ.
- 60Performance, Scale, and Differentiable Quantum Programming NewBackend selection, Catalyst/JAX/lightning.gpu/CUDA-Q, shot-budget optimization, and circuit cutting.
- 61Datasets, Benchmarks, and the Statistics of Quantum Experiments NewShared benchmark harness, shot-noise statistics, bootstrap CIs, tests for advantage, and experiment tracking.
Part XI · Capstone Projects
8 chaptersDesign, implement, evaluate, and explain complete quantum AI experiments with honest baselines, reproducible code, and the research-practice tools of Part X.
- 62How to Build a Quantum ML ExperimentResearch question, dataset, baseline, model, metrics, ablation, noise analysis, and reproducibility. The shared harness.
- 63Capstone 1: Quantum Kernel ClassifierDataset, classical baselines, quantum feature map, kernel SVM, paper-style report.
- 64Capstone 2: Variational Quantum ClassifierAnsatz design, training curves, classical comparison. Reproducible notebook and report.
- 65Capstone 3: Quantum Generative ModelBorn machine, distribution learning, sample evaluation, and failure analysis.
- 66Capstone 4: Quantum Optimization for AIQUBO / QAOA / quantum-inspired solver for feature selection, clustering, or scheduling, with a classical comparison.
- 67Capstone 5: Noise-Aware Quantum ML Study NewIdeal vs noisy vs hardware runs, mitigation applied and measured, shot-count sensitivity analysis.
- 68Capstone 6: Reproduce a Published Result with an AI Copilot NewEnd-to-end reproduction with error bars, a tuned classical baseline, and an explicit dequantization assessment. The copilot's contributions and corrections both documented.
- 69Final Perspective: What Quantum AI Is TodayMature, experimental, overhyped, and worth learning: an honest map of the field and how to keep studying.
Appendices & Back Matter
6 appendices- ALinear Algebra and Probability ReferenceFull treatment behind Ch 0: vector spaces, unitarity, spectral decomposition, tensor products, and the book's key proof sketches.
- BNotation, Symbols, and Dirac Notation ReferenceComplete symbol table, Dirac notation alongside vector/matrix forms.
- CEnvironment Setup and the Companion RepositoryPinned versions, test suite, CI setup, hardware access, and how to use the companion repo.
- DSolutions and Instructor ResourcesSolution sketches, grading rubrics, the dependency graph in full, slides, and problem sets.
- EThe Quantum AI Cookbook NewAll Recipe Boxes consolidated and indexed by task: encode, estimate, diagnose, run, align, track.
- FReproduce-the-Paper Index and Research-Practice Checklist NewCurated reproduction exercises and a one-page checklist: pinned env, tests, seeds, construct-matched metrics, error bars, honest baselines.