Surface acoustic waves drive ferromagnetic resonance through magnetoelastic coupling—strain modulates the magnetic anisotropy, generating effective fields that torque the magnetization into precession. This is the textbook explanation, and it works in the transverse geometry.
Park, Lee, and Shuai show that in the longitudinal geometry—magnetization parallel to the acoustic wave vector—the textbook mechanism generates a large effective field but zero transverse torque. The magnetoelastic strain pushes the magnetization along an axis where it cannot precess. The dominant coupling is geometrically impotent.
Note (2026-11-09): I checked the abstract of the Park, Lee and Shuai paper. It presents an extension of the mumax+ micromagnetic framework; the six benchmark simulations validate the implementation (domain-wall motion, switching, Barnett and magneto-rotation fields, nonreciprocal absorption, a standing-wave cavity), and the longitudinal-geometry result (zero transverse torque, a 50 times larger effective field, crossover near 1.1 degrees for YIG) is a simulation result. The strong-coupling cooperativity (C = 257) is for the magneto-rotation constant treated as a tunable parameter (K_mr = 1 MJ/m^3). The essay says the six simulations confirm the result and that magneto-rotation is normally orders of magnitude weaker; neither is in the abstract.
What drives the resonance instead is magneto-rotation coupling, arising from antisymmetric displacement gradients. This is normally treated as a perturbative correction—orders of magnitude weaker than magnetoelastic coupling in standard estimates. But in this geometry, it is the sole operative mechanism. The crossover angle is approximately 1.1 degrees for YIG: below that, magneto-rotation alone does all the work.
Six validation simulations confirm the result and map a cooperativity phase diagram showing that magneto-rotation coupling alone can achieve strong coupling with specific parameters.
This reveals that coupling strength and dynamical relevance are independent properties. The magnetoelastic coupling is larger by standard measures—energy scale, effective field amplitude—but produces no motion. The magneto-rotation coupling is smaller but geometrically aligned to produce torque. Strength of a force and relevance of a force are determined by different things. The largest force in the system can be exactly zero in its effect, leaving what everyone called a perturbation to do the only work that gets done.