A sharper Magnus expansion bound woven in binary branches

Abstract:The Magnus expansion provides an exponential representation of one-parameter operator families, expressed as a series expansion in their generators. This is particularly relevant in quantum mechanics for describing a unitary evolution determined by a time-dependent Hamiltonian generator of the dynamics. The solution is constructed as a series expansion in terms of increasingly complex nested commutators that rapidly become challenging to compute directly. This work establishes a universal upper bound scaling as $O\big((A(t)/\xi)^n n^{-3/2}\big)$ on the error incurred when the Magnus expansion is truncated at an arbitrary order. The factor $n^{-3/2}$ sharpens the exponential suppression governed by the integrated norm of the generator $A(t)$ and the convergence radius $\xi$, and we prove that this dependence on the order $n$ is optimal for the corresponding majorant. The main technical ingredient of the proof is the binary tree representation introduced by Iserles and Norsett from which we derive a recursion formula to delimit the magnitude of any term in the expansion. We complement our analytic bound with explicit computation of the first tree coefficients. With these findings we aim to contribute to the understanding of the accuracy and limitations of the Magnus expansion technique, and to provide a sharper bound for approximating quantum dynamics when information on the generator is not available.

arXiv:2509.18312 [quant-ph]

(or arXiv:2509.18312v2 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2509.18312 Focus to learn more arXiv-issued DOI via DataCite

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.