🔢 mathematics

The Fer expansion of a driven spin under $\mathfrak{su}(2)$ operations cannot terminate beyond first order

This paper proves that for any real-analytic drive acting on an $\mathfrak{su}(2)$ system, the Fer expansion of the propagator cannot terminate beyond the first order unless the first exponential factor already provides the exact solution.

Original authors: Lorenzo Bagnasacco, Daniel Burgarth, Vittorio Giovannetti, Kazuya Yuasa

Published 2026-10-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Lorenzo Bagnasacco, Daniel Burgarth, Vittorio Giovannetti, Kazuya Yuasa

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quantum world, the behavior of particles like electrons or atomic nuclei is governed by a mathematical object called a Hamiltonian, which acts as a blueprint for how the system changes over time. When this blueprint is simple and unchanging, scientists can predict the future state of the system with perfect clarity. However, in many real-world scenarios, such as in magnetic resonance imaging or the operation of quantum computers, this blueprint is constantly shifting, driven by external forces that change with time. Predicting the outcome of such a system is notoriously difficult because the standard mathematical tools used to solve these problems often break down. They either produce infinite lists of terms that can never be finished, or they require approximations that lose the fundamental property of the system: that the total probability of all possible outcomes must always add up to one. Physicists have long sought a way to organize these messy, time-dependent problems into a clean, finite sequence of steps that preserves this essential rule, hoping to find a shortcut that turns a complex, unsolvable puzzle into a manageable calculation.

A specific method known as the Fer expansion offers a promising path forward. Instead of trying to force the entire history of a system's evolution into a single, massive calculation, this method breaks the process down into a chain of smaller, distinct steps. Imagine peeling an onion: the first layer is removed, revealing a simpler core underneath. In this mathematical framework, the first step captures the most significant part of the system's movement, leaving behind a smaller, residual problem. This remaining problem is then treated as a new system, and the process is repeated, peeling off another layer, then another, and so on. The hope for researchers has been that for certain types of driving forces, this chain of layers would eventually run out. If the leftover problem vanished completely after a few steps, the entire evolution of the system would be described by a short, exact list of factors, making the solution perfect and finite. This would be a major breakthrough, turning a difficult, continuous problem into a simple, closed-form answer.

A team of researchers has now investigated whether this hope can be realized for a fundamental class of quantum systems known as spins, which include everything from single qubits in a quantum computer to larger atomic magnets. They focused on a specific type of mathematical structure that governs how these spins rotate and interact. By mapping the complex, recursive steps of the Fer expansion onto a geometric space, they were able to trace exactly what happens to the system at every stage of the peeling process. Their work reveals a surprising and rigid limitation: for any smooth, real-world driving force, this chain of layers never naturally stops unless it was already finished at the very beginning.

The researchers proved that the only time the expansion terminates after a finite number of steps is when the first step alone is already the complete, exact answer. In other words, if the driving force is simple enough that the first layer captures everything, the process stops immediately. However, if the driving force is complex enough to require more than one layer, the chain continues indefinitely. The mathematical machinery of the expansion does not have the power to magically simplify a complex, time-varying drive into a short list of factors. Even for drives that are known to be solvable by other means, such as those that rotate at a constant speed, the Fer expansion fails to find a stopping point unless the drive is already trivial.

This finding effectively closes the door on the idea that the Fer expansion can serve as a universal tool for finding exact, finite solutions for complex quantum drives. The study demonstrates that the method cannot generate new, solvable cases out of unsolvable ones. If a system's evolution is not already simple enough to be described by a single exponential term, the recursive process will simply continue forever, producing an infinite sequence of corrections rather than a neat, finite product. The researchers showed that the structure of the equations governing these spins is too rigid to allow a generic, complex drive to suddenly become simple after a few iterations. The only way to get a finite result is to start with a drive that is already finite in its complexity.

The implications of this result are clear for anyone trying to model or control quantum systems. While the Fer expansion remains a powerful tool for understanding the structure of quantum evolution and for creating highly accurate approximations, it cannot be relied upon to produce exact, closed-form solutions for general time-dependent drives. The search for exact solutions must continue to rely on identifying specific, special cases of driving forces that are solvable by construction, rather than hoping that a general mathematical algorithm will reveal them. The work confirms that the complexity of a driven quantum spin is an inherent feature that cannot be peeled away layer by layer; it is either there from the start, or it remains there, no matter how many steps one takes.

Technical Summary: The Fer Expansion of a Driven Spin under su(2) Operations

Problem Statement
The time evolution of a quantum system with a time-dependent Hamiltonian $H(t)$ is formally described by a time-ordered exponential. While the Dyson series provides a perturbative expansion, it truncates unitarity at finite orders. Two primary methods address this: the Magnus expansion, which resums the series into a single exponential (preserving unitarity but often yielding an infinite series of nested integrals), and the Wei–Norman theorem, which factorizes the propagator into an ordered product of exponentials based on Lie algebra generators (yielding a closed system of ODEs that may diverge).

The Fer expansion occupies an intermediate position. It expresses the propagator as an ordered product of exponentials, $U(t) = e^{F_1(t)}e^{F_2(t)}\cdots e^{F_j(t)}U_{j+1}(t)$ , where each factor is generated by an effective Hamiltonian $H_j(t)$ derived recursively from the previous step. A key theoretical question arises: Can this product terminate after a finite number of factors for a generic drive? Termination occurs if the "leftover" evolution $U_{j+1}(t)$ is known in closed form, which happens if the effective Hamiltonian $H_{j+1}(t)$ vanishes (exact termination) or belongs to a class of exactly solvable drives (e.g., commuting with itself or precessing about a fixed axis).

The authors investigate whether, for a spin system governed by the $\mathfrak{su}(2)$ algebra (covering qubits to arbitrary spin sizes), a generic real-analytic drive can be "steered" by the Fer recursion into a solvable class after a finite number of steps, thereby allowing the expansion to stop non-trivially.

Methodology
The paper analyzes the Fer recursion specifically within the $\mathfrak{su}(2)$ algebra. The methodology proceeds as follows:

Algebraic Formulation: The Hamiltonian is expressed as $H(t) = \mathbf{h}(t) \cdot \mathbf{T}$ , where $\mathbf{T}$ are the generators. The recursion is shown to act entirely on the real coefficient vectors $\mathbf{h}(t)$ in $\mathbb{R}^3$ , mapping $\mathbf{h}_j(t) \to \mathbf{h}_{j+1}(t)$ .
Closed-Form Map Derivation: The authors derive a closed-form expression for the Fer map on $\mathfrak{su}(2)$ $su (2)$ . By summing the series of nested commutators (which correspond to cross products in $\mathbb{R}^3$ $R^{3}$ ), they express the next-order drive vector $\mathbf{h}_{j+1}(t)$ $h_{j + 1} (t)$ as a linear combination of two auxiliary vectors, $\mathbf{p}_{j+1}(t)$ $p_{j + 1} (t)$ and $\mathbf{q}_{j+1}(t)$ $q_{j + 1} (t)$ , weighted by scalar functions $\alpha$ $α$ and $\beta$ $β$ dependent on the norm of the integrated drive $\mathbf{f}_j(t) = \int_0^t \mathbf{h}_j(s) ds$ $f_{j} (t) = \int_{0}^{t} h_{j} (s) d s$ .
- The map is geometrically interpreted as projecting the drive into the plane orthogonal to its own time integral.
Rigidity Analysis: The authors investigate the conditions under which the chain of drives $\mathbf{h}_1 \to \mathbf{h}_2 \to \dots$ $h_{1} \to h_{2} \to \dots$ enters a solvable class. They focus on two specific solvable families:
- Family A: Drives where $[H(t), H(t')] = 0$ for all $t, t'$ (the drive vector moves along a fixed line through the origin).
- Family $\tilde{A}$ : Drives corresponding to Rabi's rotating field, where the vector precesses about a fixed axis with constant angular velocity.
Proof Strategy: Assuming the original drive is real-analytic, the authors perform a backward induction. They prove that if a drive at order $j > 1$ belongs to Family A or $\tilde{A}$ , the preceding drive must have been trivial (zero) or already in Family A.

Key Contributions and Results

Closed-Form Fer Map for $\mathfrak{su}(2)$ : The paper provides an explicit, closed-form map (Eq. 4.6) for the Fer recursion in $\mathfrak{su}(2)$ , reducing the operator recursion to a geometric transformation of vectors in $\mathbb{R}^3$ involving elementary trigonometric functions.
Non-Termination Theorem (Family A): Theorem 1 establishes that for a real-analytic drive, if the effective Hamiltonian at any order $j > 1$ commutes with itself at all times (Family A), then that Hamiltonian must be identically zero. Consequently, the original drive must have already belonged to Family A. The recursion does not generate new commuting drives from non-commuting ones.
Non-Termination Theorem (Family $\tilde{A}$ ): Theorem 2 extends this result to Rabi's rotating field (Family $\tilde{A}$ ). It proves that if the effective Hamiltonian at order $j > 1$ describes a uniformly precessing field, it must be identically zero. Thus, the recursion cannot produce a non-trivial rotating field drive from a generic initial drive.
Rigidity of the Chain: The combined results demonstrate a "rigidity" in the Fer expansion for $\mathfrak{su}(2)$ . The expansion cannot terminate after the first factor for a generic drive. It only terminates if the initial drive is already exactly solvable (specifically, if it commutes with itself), in which case the first exponential is already exact and the remainder is zero.

Significance and Claims
The paper claims to settle the question of whether the Fer expansion offers a mechanism to transform generic, non-solvable drives into exactly solvable ones through finite recursion. The authors conclude that for $\mathfrak{su}(2)$ systems with real-analytic drives, the answer is negative.

The significance lies in clarifying the limitations of the Fer expansion compared to other methods. Unlike the Wei–Norman approach, which may require restarting factorizations due to divergence, or the Magnus expansion, which is an infinite series, the Fer expansion is unitary at every truncation. However, this paper shows that this unitarity does not come with the added benefit of "self-termination" for generic drives. The recursion creates no new stopping points; the only drives for which the product terminates finitely are those that were already exactly solvable at the first step.

The authors modestly note that their analysis does not rule out the possibility that the chain could reach other classes of exactly solvable drives (outside Families A and $\tilde{A}$ ) where the expansion could be stopped with a known closed-form remainder, though the product itself would not terminate. However, regarding the specific classes of commuting and uniformly precessing drives, the Fer expansion adds no new exactly solvable cases beyond those already known.

More like this