Trainability · Algorithms

Simple-DLA barren plateau (g ≃ su(d))

Theorem statement

For a parametrized quantum circuit whose dynamical Lie algebra $\mathfrak{g}$ is simple, $\mathfrak{g} \simeq \mathfrak{su}(d)$ (dimension $d^2-1$, centerless), the loss variance reduces to a single term $$\operatorname{Var}_\theta[\ell] = \frac{P_{\mathfrak{g}}(\rho)\, P_{\mathfrak{g}}(O)}{d^2-1}.$$ In particular, for $d = 2^n$ the dimension $\dim\mathfrak{g} = 4^n-1$ grows exponentially in the qubit count $n$, so the loss exhibits an exponential barren plateau (under the Haar second-moment / Schur hypothesis carried as a named input).

Sources

  1. 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

  2. Barren plateaus in quantum neural network training landscapes

    Jarrod R. McClean, Sergio Boixo, Vadim N. Smelyanskiy, Ryan Babbush, Hartmut Neven, 2018

Lean context

Copy a short prompt with the import, theorem name, citations, and public source link.

Open Lean source