Comment on "Spin-orbit coupling induced ultrahigh-harmonic generation from magnetic dynamics" with prescriptions on how to validate scientific software for computational quantum transport

Abstract

In a recent paper [Phys. Rev. B {\bf 105}, L180415 (2022)], Ly and Manchon used open source code {\tt TKWANT} for time-dependent computational quantum transport to predict surprising features in the mature field of current pumping by magnetization dynamics in spintronics -- in the presence of spin-orbit (SO) coupling, the pumped charge current oscillates at both the frequency $\omega_0$ of magnetization precession and high harmonics $N=\omega/\omega_0$, reaching {\em astonishingly high} cutoff \mbox{$N_\mathrm{max} \simeq 1000$} by increasing the SO coupling. However, results in the paper violate two basic theorems of time-dependent quantum transport: ({\em i}) current response to time-periodic external field {\em must be perfectly periodic} itself in the long time limit for a two-terminal device because its active region is attached to two semi-infinite leads bringing continuous energy spectrum; and ({\em ii}) no DC component of charge current is allowed in the left-right symmetric two-terminal devices, or in asymmetric devices its value cannot be changed by simply increasing the SO coupling. We illustrate these two theorems by using completely different calculations applied to one-dimensional two-terminal devices with either ferromagnetic (for which the device is left-right symmetric) and antiferromagnetic (for the device is left-right asymmetric) active region hosting the Rashba SO coupling. We conclude that harmonics in pumped current in the presence of SO coupling do exist, but their ``ultrahigh'' cutoff is an artifact of either ``bugs'' or inadequate algorithms selected within {\tt TKWANT}. Finally, we suggest strategies for {\em validating} time-dependent quantum transport codes, or selection of algorithms by a user within putatively validated (by developers) code, prior to deploying them to produce research papers.