Skip to content

Mathematical Reference Tables

Linear Algebra Identities

Table 1. Key matrix identities used in QML
Identity Condition
\((AB)^\dagger = B^\dagger A^\dagger\) Any matrices
\(\mathrm{Tr}[AB] = \mathrm{Tr}[BA]\) Square matrices
\(\mathrm{Tr}[A\otimes B] = \mathrm{Tr}[A]\cdot\mathrm{Tr}[B]\) Any matrices
\(\|A\|_F^2 = \mathrm{Tr}[A^\dagger A]\) Frobenius norm
\((A\otimes B)(C\otimes D) = (AC)\otimes(BD)\) Compatible dims
\(e^{iA} = \cos A + i\sin A\) Hermitian \(A\)
\(e^{-i\theta P/2} = \cos(\theta/2)I - i\sin(\theta/2)P\) Pauli \(P\), \(P^2=I\)
\(U^\dagger = U^{-1}\) Unitary \(U\)

Pauli Matrix Algebra

\[\begin{align*} X &= \begin{pmatrix}0&1\\1&0\end{pmatrix}, \quad Y = \begin{pmatrix}0&-i\\i&0\end{pmatrix}, \quad Z = \begin{pmatrix}1&0\\0&-1\end{pmatrix} \\ XY &= iZ, \quad YZ = iX, \quad ZX = iY \\ \{X,Y\} &= 0, \quad [X,Y] = 2iZ \\ X^2 &= Y^2 = Z^2 = I \\ \mathrm{Tr}[P_\alpha P_\beta] &= 2\delta_{\alpha\beta} \quad (P\in\{I,X,Y,Z\}) \end{align*}\]

Quantum Gate Reference

Table 2. Common quantum gates: matrix, circuit symbol, and properties
Gate Matrix Inverse Generator
\(H\) \(\frac{1}{\sqrt{2}}\begin{pmatrix}1&1\\1&-1\end{pmatrix}\) \(H\) \(-(X+Z)/\sqrt{2}\)
\(S\) \(\begin{pmatrix}1&0\\0&i\end{pmatrix}\) \(S^\dagger\) \(Z\)
\(T\) \(\begin{pmatrix}1&0\\0&e^{i\pi/4}\end{pmatrix}\) \(T^\dagger\) \(Z\)
\(R_x(\theta)\) \(\begin{pmatrix}\cos\frac{\theta}{2}&-i\sin\frac{\theta}{2}\\-i\sin\frac{\theta}{2}&\cos\frac{\theta}{2}\end{pmatrix}\) \(R_x(-\theta)\) \(X\)
\(R_y(\theta)\) \(\begin{pmatrix}\cos\frac{\theta}{2}&-\sin\frac{\theta}{2}\\\sin\frac{\theta}{2}&\cos\frac{\theta}{2}\end{pmatrix}\) \(R_y(-\theta)\) \(Y\)
\(CNOT\) \(\begin{pmatrix}1&0&0&0\\0&1&0&0\\0&0&0&1\\0&0&1&0\end{pmatrix}\) \(CNOT\) \(\frac{I-Z}{2}\otimes X\)

Entropy and Information Measures

Table 3. Information-theoretic quantities used in QML
Quantity Classical Quantum
Entropy \(H(X) = -\sum_x p_x\log p_x\) \(S(\rho) = -\mathrm{Tr}[\rho\log\rho]\)
Relative entropy \(D(p\|q) = \sum_x p_x\log\frac{p_x}{q_x}\) \(D(\rho\|\sigma) = \mathrm{Tr}[\rho(\log\rho-\log\sigma)]\)
Mutual info \(I(X;Y) = H(X) - H(X|Y)\) \(I(A;B) = S(A) + S(B) - S(AB)\)
Fidelity \(F(p,q) = \sum_x\sqrt{p_xq_x}\) \(F(\rho,\sigma) = \mathrm{Tr}\sqrt{\sqrt{\rho}\sigma\sqrt{\rho}}\)
Trace distance \(D(p,q) = \frac{1}{2}\sum_x|p_x-q_x|\) \(D(\rho,\sigma) = \frac{1}{2}\|\rho-\sigma\|_1\)
Fisher info \(F_\theta = \mathbb{E}\left[(\partial_\theta\log p)^2\right]\) \(F_{jk} = 4\,\mathrm{Re}\langle\partial_j\psi|\partial_k\psi\rangle_c\)

Taylor and Fourier Series

For a quantum circuit \(f(\theta) = \langle O\rangle_{U(\theta)}\) where \(U(\theta)\) has generator with spectrum \(\{\pm r\}\):

\[f(\theta) = c_0 + c_1\cos(r\theta) + s_1\sin(r\theta)\]

For a circuit with \(K\) distinct frequencies \(\{r_k\}_{k=1}^K\):

\[f(\theta) = c_0 + \sum_{k=1}^K [c_k\cos(r_k\theta) + s_k\sin(r_k\theta)]\]

The gradient via PSR: \(\frac{df}{d\theta} = r\sum_{k}[c_k(-\sin(r\theta)) + s_k\cos(r\theta)] = r[f(\theta+\frac{\pi}{4r}) - f(\theta-\frac{\pi}{4r})]\).

Useful Inequalities

\[\begin{align*} \text{Cauchy-Schwarz:}& \quad |\langle\phi|\psi\rangle|^2 \leq \langle\phi|\phi\rangle\langle\psi|\psi\rangle \\ \text{Hoeffding's:}& \quad P(|\hat{f} - f| \geq \epsilon) \leq 2e^{-2M\epsilon^2} \\ \text{Markov's:}& \quad P(X \geq t) \leq \mathbb{E}[X]/t \quad (X \geq 0) \\ \text{Fuchs-van de Graaf:}& \quad 1 - F(\rho,\sigma) \leq D(\rho,\sigma) \leq \sqrt{1-F(\rho,\sigma)^2} \\ \text{Operator norm:}& \quad \|AB\| \leq \|A\|\cdot\|B\| \end{align*}\]