The central question the authors tackle is deceptively simple: when two perturbative schemes claim to describe the same experiment, are they actually computing the same thing? The answer, worked out in detail in their manuscript posted on 11 September 2026 as a discussion preprint under review for the journal Magnetic Resonance, is a qualified yes. Bock, Ernst and Uhrig identify the Floquet–Magnus expansion as the essential bridge between the two formalisms. By expressing the Floquet solution in the language of the Magnus expansion, the standard tool for time-dependent perturbation theory, they are able to prove the mathematical equivalence of Floquet theory and average Hamiltonian theory term by term. The equivalence, however, is formal rather than practical. Two calculational routes that lead to identical expressions in principle can still differ enormously in how error-prone, transparent and numerically stable they are along the way, and it is precisely here that the new work makes its mark.

Average Hamiltonian theory has been the workhorse of solid-state NMR since the 1960s. Its logic is intuitive: if the Hamiltonian repeats with a period that is short compared with the timescale on which the interesting dynamics unfold, then over many cycles the system behaves, to a good approximation, as if it evolved under a single constant, effective Hamiltonian. The trick is to compute that effective Hamiltonian as a series of corrections, each involving nested commutators of the Fourier components of the periodic Hamiltonian. The method has an Achilles heel, though. At higher orders it demands explicit time integrations over products of oscillating terms, and those integrations are notorious breeding grounds for algebraic mistakes. Anyone who has carried an average Hamiltonian calculation to third or fourth order knows the tangle of trigonometric factors and commutator chains that must be evaluated without a single slip.

Floquet theory approaches the same physics from a different direction, treating the periodic drive not as a perturbation to be averaged away but as a structural feature of the problem. In the Floquet picture, the Hilbert space of the quantum system is enlarged to include sideband indices that count quanta of the driving frequency, a step closely analogous to how quasi-momentum appears for electrons in a crystal lattice. The time-dependent problem then becomes a time-independent one in the extended space, and standard perturbative machinery can be brought to bear. The authors advocate a specific variant, the Floquet–Van Vleck approach, in which the effective Hamiltonian is accompanied by a second object called the kick operator. The kick operator accounts for the micromotion within each driving cycle, the fast wiggles that the effective Hamiltonian deliberately ignores, and keeping it in the formalism turns out to be essential for a clean and consistent description.

One of the paper’s key technical contributions is a calculation scheme that avoids explicit integration altogether. By working with the Fourier components of the periodic Hamiltonian and separating secular contributions, those terms that survive averaging because they commute with the dominant part of the Hamiltonian, from non-secular ones, the authors obtain a hierarchy of corrections that is far less prone to algebraic error. On this foundation they derive the first four orders of both average Hamiltonian theory and the Floquet expansion, providing the community with explicit analytical expressions at orders where previous treatments often stopped or grew unwieldy. The consistent separation of secular and non-secular pieces, they report, is especially advantageous for numerical robustness, because it keeps the calculation stable even when the perturbative parameters are not particularly small, a situation that arises routinely in real spectrometers where spinning frequencies and radiofrequency amplitudes push against the limits of simple approximations.

The payoff of this rigor is quantified in a striking comparison. Despite the formal equivalence of the two frameworks, the accuracy of the Floquet–Van Vleck approach is about three times better than that of average Hamiltonian theory at the orders studied. The reason is subtle but important: the two methods organize the same physics differently, and the Floquet organization, with its clean split between the effective Hamiltonian and the kick operator, distributes truncation errors more favorably. For experimentalists designing pulse sequences or interpreting high-resolution spectra of solids, a threefold improvement in the accuracy of predicted frequencies and evolution is not an academic nicety. It can determine whether a calculated spectrum matches the measured one well enough to extract structural information about a protein, a catalyst or a battery material, or whether discrepancies are blamed on the theory when they are actually artifacts of the calculational route.

To ground the mathematics in real experiments, the authors examine two canonical test cases. The first is the Bloch–Siegert shift, a small displacement of resonance frequencies caused by the counter-rotating component of an oscillating radiofrequency field, a phenomenon known since the early days of magnetic resonance. Their calculation of this shift out to third order, highlighted in the open discussion accompanying the preprint as particularly revealing, shows how higher-order terms become observable when strong decoupling fields are applied. The second test case involves dipolar coupled spin systems under magic-angle spinning, the technique in which a sample is spun at precisely the angle where the anisotropic dipole-dipole interactions average most effectively, forming the backbone of modern solid-state NMR. Here the periodic drive is mechanical, set by the rotor frequency, and the interplay between spinning, chemical shifts and spin-spin couplings is exactly the kind of problem where the choice of theoretical framework matters most.

The discussion thread attached to the preprint illustrates why this work resonates with practitioners. Commenter Tom Barbara, a veteran of micro-coil NMR, noted that groups working with radiofrequency fields producing nutation frequencies on the order of a megahertz had long wondered why such Bloch–Siegert shifts went unnoticed in spectra. Matthias Ernst’s reply explains the puzzle: all lines in a spectrum shift by the same amount, and since spectroscopists typically measure only relative frequencies, a uniform shift is invisible unless an undecoupled reference is recorded. He points out that recent publications have not only observed but exploited the heteronuclear Bloch–Siegert shift, using it to calibrate the radiofrequency field amplitude of insensitive nuclei, and he recalls personally noticing the significant proton shift that appears when decoupling nitrogen-15 during acquisition. The exchange also touches on a terminological subtlety: the frequency shifts used in MRI to map radiofrequency fields share the Bloch–Siegert name but arise from a different mechanism with a different symmetry under rotations, a confusion that has trailed the field since the early papers by Bodenhausen and Emsley.

What elevates this preprint beyond a methodological housekeeping exercise is its timing. Solid-state NMR is pushing into regimes of ever-faster magic-angle spinning, with rotors now exceeding spinning frequencies of 70 kilohertz in some applications, and into decoupling schemes for nuclei that are difficult to observe directly. In these regimes the perturbative parameters are strained, the boundary between secular and non-secular terms shifts, and the silent assumptions baked into older treatments begin to creak. By proving exactly where Floquet theory and average Hamiltonian theory agree, identifying the Floquet–Magnus expansion as the hinge connecting them, and demonstrating that the Floquet–Van Vleck route with its explicit kick operator delivers superior numerical accuracy, Bock, Ernst and Uhrig have given the magnetic resonance community both a conceptual map and a practical recommendation. The authors’ bottom line is unambiguous: for analyzing magic-angle spinning experiments, Floquet theory is the tool of choice, prized for its numerical robustness, efficiency and accuracy. As the preprint moves through peer review, its four-order analytical results are already positioned to become a reference point for anyone who needs to know precisely what a periodically driven spin system is really doing, cycle after cycle, beneath the measured spectrum.

Subject of Research: Mathematical equivalence of Floquet theory and average Hamiltonian theory and their application to solid-state NMR spectroscopy

Article Title: Floquet Theory and Average Hamiltonian Theory Revisited: Equivalence, Convergence and Applications to NMR

Article References: Floquet Theory and Average Hamiltonian Theory Revisited: Equivalence, Convergence and Applications to NMR. (n.d.). https://doi.org/10.5194/mr-2026-10

Image Credits: AI Generated

DOI: 10.5194/mr-2026-10

Keywords: Floquet theory, average Hamiltonian theory, NMR spectroscopy, magic-angle spinning, Bloch-Siegert shift, spin dynamics, perturbation theory, Floquet-Magnus expansion, kick operator, solid-state NMR, decoupling, effective Hamiltonian