Skip to content

Notation and Conventions

This chapter summarizes the mathematical notation used throughout the book. Familiarity with this notation will accelerate reading and reduce ambiguity.

General Mathematical Notation

Symbol Meaning
\(\mathbb{R}\), \(\mathbb{C}\) Real and complex number fields
\(\mathbb{Z}\), \(\mathbb{N}\) Integer and natural number sets
\(\mathcal{H}\) Hilbert space
\(\mathbb{C}^n\) \(n\)-dimensional complex vector space
\(\mathbf{x} \in \mathbb{R}^d\) Column vector \(\mathbf{x}\) of dimension \(d\)
\(\mathbf{A} \in \mathbb{C}^{m \times n}\) Matrix \(\mathbf{A}\) of size \(m \times n\)
\(\mathbf{A}^\top\) Transpose of matrix \(\mathbf{A}\)
\(\mathbf{A}^\dagger\) Conjugate transpose (adjoint) of \(\mathbf{A}\)
\(\mathbf{A}^{-1}\) Inverse of matrix \(\mathbf{A}\)
\(\|\mathbf{x}\|\) Euclidean (2-norm) of vector \(\mathbf{x}\)
\(\|\mathbf{A}\|_F\) Frobenius norm of matrix \(\mathbf{A}\)
\(\|\mathbf{A}\|_*\) Nuclear norm of matrix \(\mathbf{A}\)
\(\langle \mathbf{u}, \mathbf{v} \rangle\) Inner product of \(\mathbf{u}\) and \(\mathbf{v}\)
\(\mathbf{u} \otimes \mathbf{v}\) Tensor (Kronecker) product
\(\mathbf{I}_n\) \(n \times n\) identity matrix
\(\mathbf{0}\) Zero vector or zero matrix
\(\text{tr}(\mathbf{A})\) Trace of matrix \(\mathbf{A}\)
\(\text{Tr}[\cdot]\) Trace operator (physics convention)
\(\det(\mathbf{A})\) Determinant of \(\mathbf{A}\)
\(\text{rank}(\mathbf{A})\) Rank of matrix \(\mathbf{A}\)
\(\text{diag}(a_1,…,a_n)\) Diagonal matrix with entries \(a_1,…,a_n\)
\(\sigma_{\max}(\mathbf{A})\) Largest singular value of \(\mathbf{A}\)
\(\lambda_i(\mathbf{A})\) \(i\)-th eigenvalue of \(\mathbf{A}\)
\(\text{span}(S)\) Linear span of set \(S\)
\(\mathcal{O}(f(n))\) Big-O asymptotic complexity
\(\Omega(f(n))\) Big-Omega (lower bound)
\(\tilde{\mathcal{O}}(f(n))\) Big-O ignoring logarithmic factors
\(\nabla f(\mathbf{x})\) Gradient of \(f\) at \(\mathbf{x}\)
\(\nabla^2 f(\mathbf{x})\) Hessian matrix of \(f\)
\(\partial f / \partial x_i\) Partial derivative
\(\mathbb{E}[X]\) Expectation of random variable \(X\)
\(\text{Var}[X]\) Variance of \(X\)
\(\text{Cov}[X,Y]\) Covariance of \(X\) and \(Y\)
\(P(A)\) Probability of event \(A\)
\(P(A|B)\) Conditional probability of \(A\) given \(B\)
\(\mathcal{N}(\mu, \sigma^2)\) Normal distribution
\(H(X)\) Shannon entropy of \(X\)
\(D_{\text{KL}}(P\|Q)\) KL divergence from \(P\) to \(Q\)
\(I(X;Y)\) Mutual information

Quantum Mechanics Notation (Dirac/Bra-Ket)

Symbol Meaning
\(|\psi\rangle\) Ket: quantum state vector in \(\mathcal{H}\)
\(\langle\psi|\) Bra: dual vector \(|\psi\rangle^\dagger\)
\(\langle\phi|\psi\rangle\) Inner product: \(\langle\phi||\psi\rangle \in \mathbb{C}\)
\(|\psi\rangle\langle\phi|\) Outer product: rank-1 operator
\(|0\rangle, |1\rangle\) Computational basis states of a qubit
\(|+\rangle, |-\rangle\) Hadamard basis states \((|0\rangle \pm |1\rangle)/\sqrt{2}\)
\(|\Phi^+\rangle\) Bell state \((|00\rangle + |11\rangle)/\sqrt{2}\)
\(|x\rangle\) Computational basis state for bit string \(x\)
\(|\psi\rangle^{\otimes n}\) \(n\)-fold tensor product of \(|\psi\rangle\)
\(\hat{U}\) Unitary operator
\(\hat{H}\) Hamiltonian operator
\(\hat{\rho}\) Density matrix (operator)
\(\hat{O}\) Observable (Hermitian operator)
\(\langle \hat{O} \rangle_\psi\) Expectation value \(\langle\psi|\hat{O}|\psi\rangle\)
\(\sigma_x, \sigma_y, \sigma_z\) Pauli matrices (also \(X, Y, Z\))
\(R_x(\theta), R_y(\theta), R_z(\theta)\) Single-qubit rotation gates
\(H\) Hadamard gate
\(\text{CNOT}\), \(\text{CX}\) Controlled-NOT gate
\(\text{CZ}\) Controlled-Z gate
\(\text{SWAP}\) SWAP gate
\(S(\hat{\rho})\) von Neumann entropy: \(-\text{Tr}[\hat{\rho}\log\hat{\rho}]\)
\(F(\rho,\sigma)\) Quantum fidelity
\(D(\rho,\sigma)\) Trace distance
\(\mathcal{E}\) Quantum channel (CPTP map)
\(K_i\) Kraus operators of a channel
\(n\) Number of qubits
\(N = 2^n\) Hilbert space dimension

Machine Learning Notation

Symbol Meaning
\(\mathcal{X}\) Input (feature) space
\(\mathcal{Y}\) Output (label) space
\(\mathcal{D} = \{(\mathbf{x}_i, y_i)\}_{i=1}^N\) Training dataset
\(N\) Number of training samples
\(d\) Input feature dimension
\(f: \mathcal{X} \to \mathcal{Y}\) Hypothesis/model function
\(\mathcal{L}(\cdot)\) Loss function
\(\mathcal{C}(\boldsymbol{\theta})\) Cost function (parameterized)
\(\boldsymbol{\theta} \in \mathbb{R}^p\) Trainable parameters
\(\boldsymbol{\phi}(\mathbf{x})\) Feature map / embedding
\(k(\mathbf{x}, \mathbf{x}')\) Kernel function
\(\mathbf{K}\) Kernel (Gram) matrix
\(\eta\) Learning rate
\(\lambda\) Regularization parameter
\(\epsilon\) Small positive constant / tolerance
\(\mathcal{R}[\hat{f}]\) Generalization risk
\(\hat{\mathcal{R}}_N[\hat{f}]\) Empirical risk
\(\text{VC-dim}(\mathcal{H})\) VC dimension of hypothesis class
\(\mathcal{R}_N(\mathcal{H})\) Rademacher complexity

QML-Specific Notation

Symbol Meaning
\(U(\boldsymbol{\theta})\) Parameterized unitary (ansatz)
\(U_\phi(\mathbf{x})\) Feature map unitary
\(|\psi(\boldsymbol{\theta})\rangle\) Parameterized quantum state
\(\mathcal{C}(\boldsymbol{\theta})\) Quantum cost function
\(\frac{\partial \mathcal{C}}{\partial \theta_i}\) Gradient via parameter shift rule
\(\mathbf{F}(\boldsymbol{\theta})\) Quantum Fisher information matrix
\(g_{ij}\) Fubini-Study metric tensor
\(\text{Ent}(|\psi\rangle)\) Entanglement entropy
\(\text{Expr}(U)\) Expressibility of ansatz \(U\)
\(\text{Ent}(U)\) Entanglement capability of \(U\)
\(k_Q(\mathbf{x},\mathbf{x}')\) Quantum kernel
\(\mathbf{K}_Q\) Quantum kernel matrix
\(\bar{e}_\infty\) Average gate fidelity
\(\epsilon_{\text{depol}}\) Depolarizing error rate
\(T_1, T_2\) Relaxation / coherence times
\(p\) Number of QAOA layers
\(r\) Number of data re-uploading layers

Abbreviations Used Frequently

See the Abbreviations section for a complete list. Common abbreviations encountered from the first chapter include: QML (quantum machine learning), PQC (parameterized quantum circuit), VQC (variational quantum circuit), NISQ (noisy intermediate-scale quantum), QNN (quantum neural network), QSVM (quantum support vector machine), VQE (variational quantum eigensolver), QAOA (quantum approximate optimization algorithm), HHL (Harrow-Hassidim-Lloyd).