Trainability ยท Algorithms
Locality forbids the barren plateau
Theorem statement
For the $n$-qubit family built from single-qubit gates, whose dynamical Lie algebra is the product $\mathfrak g=\bigoplus_{j=1}^{n}\mathfrak{su}(2)_j$, fix the input state $\rho=|0\rangle\!\langle0|^{\otimes n}$ and the local observable $O=X_1$ (Pauli $X$ on the first qubit), with loss $\ell_\theta=\operatorname{Tr}[U_\theta\,\rho\,U_\theta^{\dagger}O]$. The loss variance is the closed-form constant $$\operatorname{Var}_\theta[\ell]=\frac{P_{\mathfrak g_1}(\rho)\,P_{\mathfrak g_1}(O)}{\dim\mathfrak{su}(2)_1}=\frac{2^{-n}\cdot 2^{n}}{3}=\frac13,$$ independent of the qubit count $n$, where $\mathfrak g_1=\mathfrak{su}(2)_1$ is the single-qubit ideal of the observable and $P_{\mathfrak g_1}$ its purity. A constant variance does not decay as $C\,b^{-n}$ with $b>1$, so the family has no barren plateau.
Sources
- A Lie Algebraic Theory of Barren Plateaus for Deep Parameterized Quantum Circuits
Michael Ragone, Bojko N. Bakalov, Frederic Sauvage, Alexander F. Kemper, Carlos Ortiz Marrero, Martin Larocca, M. Cerezo, 2023
Lean context
- Lean declaration
QuantumAlg.localObs_not_hasBarrenPlateau
Copy a short prompt with the import, theorem name, citations, and public source link.