Exchange-correlation torques from gauge symmetries
Jacques K. Desmarais, Kamel Bencheikh, Giovanni Vignale, Stefano Pittalis · 2025
Read Exchange-correlation torques from gauge symmetries by Jacques K. Desmarais, Kamel Bencheikh, Giovanni Vignale, Stefano Pittalis (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-19458.txt.
Introduction
Jacques Desmarais and colleagues resolved a long-standing glitch in spin density theory in a 2025 paper. Simulating magnets whose spins point in varying directions produces a torque from the exchange-correlation term, the theory's catch-all for quantum interactions. That torque should vanish or stay physical, but in plain spin-density theory it comes out spurious, because it secretly depends on the spin current, a quantity the theory does not contain.
Their fix works inside spin-current density theory, whose energy obeys an exact combined gauge symmetry. Nonzero torques on twisted solutions are then fenced by that symmetry and never enter the actual spin dynamics, which involve only physical currents and torques. Two decades of confusion trace to a missing constraint.
Catalog record
| CREATOR | Jacques K. Desmarais, Kamel Bencheikh, Giovanni Vignale, Stefano Pittalis |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-19458
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exchange-correlation torques from gauge symmetries Jacques K. Desmarais, Kamel Bencheikh, Giovanni Vignale, Stefano Pittalis Abstract The problem of predicting accurate exchange-correlation (xc) spin-torques in non-collinear magnetic systems has dominated the scene of spin-density functional theory (SDFT) in the last two decades. Progress has been hindered by the fact that the spin torque is directly connected to the divergence of the spin current, a quantity that is extraneous to SDFT. Furthermore, SDFT does not apply in the presence of vector potentials and spin-orbit couplings. Here, we propose a solution that exploits the physical implications of the U(1)×SU(2) gauge invariance of the xc energy in SpinCurrent-DFT. We derive explicit xc torque expressions based on meta-generalized-gradient approximations in the framework of the generalized Kohn-Sham (GKS) formulation. One key term represents an xc-torque involving the GKS spin-kinetic energy density; and another term resembles the phenomenological spin current of the Landau-Lifshitz equations: both are derived from first-principles. We also show that the functional form ensures that the GKS particle- and spin-currents are identical, in form, to their interacting counterparts. Non-collinear equilibrium conditions and adiabatic dynamics are thus derived that resolve longstanding issues. Introduction. Density functional theory (DFT) can predict properties of materials from solely the knowledge of the atomic species in the materials and an approximation for the exchange-correlation (xc) energy; i.e., a functional of the particle density that maps the relevant interacting many-body problem onto an effective single-particle problem Jones 2015 . Given that the exact xc-energy functional of DFT is not known, empiricism can be reduced only by knowing the exact properties of the xc-energy functional Perdew and Schmidt 2001 . Generalizations of DFT hold the promise of yielding solutions for materials that can be described via Schrödinger-Pauli equations, involving two-component spinors. In Spin-DFT von Barth and Hedin 1972 , the interacting problem is mapped onto an auxiliary non-interacting problem reproducing the interacting particle and spin densities. Thus, one must solve the single-particle equation \displaystyle\left[-\frac{1}{2}\nabla^{2}+v_{s}+\mu_{B}\vec{B}_{s}\cdot\vec{\sigma}\right]\Phi_{k}=\epsilon_{k}\Phi_{k} (1) where \mu_{B}=\frac{1}{2} (in Hartree atomic units) and \Phi_{i} are two-component Pauli spinors. The Kohn-Sham (KS) system reproduces the interacting particle density, via n=\sum_{k}f_{k}\Phi_{k}^{\dagger}\Phi_{k} , and each component of the interacting spin magnetization, via m^{a}=\sum_{k}f_{k}\Phi_{k}^{\dagger}\sigma^{a}\Phi_{k} , where the {f_{k}} are occupation (often, integer) numbers. The effective potentials are v_{s}=v+v_{\rm H}+v_{\rm xc} , {B}^{a}_{s}={B}^{a}+{B}^{a}_{\rm xc} , where v_{\rm H} is the usual Hartree potential. The {\rm xc} -potentials are functional derivatives of the {\rm xc} -energy functional, E_{\rm xc}[n,{\vec{m}}] , with respect to the corresponding conjugate densities. Namely, v_{\rm xc}=\frac{\delta E_{\rm xc}[n,{\vec{m}}]}{\delta n} , and B^{a}_{\rm xc}=\frac{\delta E_{\rm xc}[n,{\vec{m}}]}{\delta m^{a}} . Thus, E_{\rm xc}[n,{\vec{m}}] is the key quantity that needs to be approximated. Although Eq. ( 1 ) reproduces the interacting spin density , it does not reproduce necessarily the (paramagnetic) spin current \vec{J}_{\mu} . The latter quantity enters the exact equation of motion \frac{\partial\vec{{m}}}{\partial t}+\partial_{\mu}\vec{{{J}}}_{\mu}=\vec{B}\times\vec{{m}} (2) which plays a central role in theories of magnetization dynamics and spintronics Bhandary et al. 2023 ; Dewhurst et al. 2020 ; Sharma et al. 2022 ; Géneaux et al. 2024 ; Sanvito 2007 ; Sanvito 2011 ; Sutton and Datta 2015 . In this context, it was noticed long ago Capelle et al. 2001 that Eq. ( 2 ) implies that the xc magnetic field must have a component orthogonal to the magnetization: this creates an “xc spin torque” which accounts for the difference between the exact spin current and the KS spin current \displaystyle{\vec{m}}\times{\vec{B}}_{\rm xc}=\partial_{\mu}\left[\vec{J}_{\mu}-\vec{J}_{s,\mu}\right] (3) where \vec{J}_{s,\mu}=\sum_{k}\frac{f_{k}}{2i}\left\{\left[\partial_{\mu}\Phi_{k}^{\dagger}\right]\vec{\sigma}\Phi_{k}-\Phi_{k}^{\dagger}\vec{\sigma}\left[\partial_{\mu}\Phi_{k}\right]\right\} is the (paramagnetic) spin current of the KS system. Since then, the quest to find xc fields that generate an accurate xc spin torque has dominated the scene of SDFT for non-collinear magnetism Sharma et al. 2007a ; Scalmani and Frisch 2012 ; Bulik et al. 2013a ; Eich and Gross 2013 ; Eich et al. 2013 ; Ullrich 2018 ; Dewhurst et al. 20201825 ; Petrone et al. 2018 ; Goings et al. 2018 ; Komorovsky et al. 2019 ; Desmarais et al. 2021 ; Hill et al. 2023 ; Pu et al. 2023 ; Tancogne-Dejean et al. 2023 ; Hill et al. 2023 ; Moore et al. 2025 . While Eq. ( 3 ) is a promising starting point, the challenge lies in that the difference between the exact spin current and the KS spin current is an uncontrolled quantity, which is beyond the reach of SDFT, and therefore cannot be used to derive local constraints that would reduce the empiricism in the construction of approximate {\vec{B}}_{\rm xc} . Furthermore, systems including extra magnetic interactions (such as vector potentials and spin-orbit couplings) are beyond self-consistent SDFT treatments. The only constraint that could be derived from Eq. ( 3 ) is the zero-torque theorem \displaystyle\int d^{3}r\,{\vec{m}}\times{\vec{B}}_{\rm xc}=\vec{0}\; (4) following from the fact that the electron-electron interaction does not exert a net torque on the magnetization. However, this constraint is too weak to provide useful guidance for improving approximations. We show below that both Eq. ( 3 ) as well as Eq. ( 4 ) have to be further extended to arrive at a solution for a wide domain of applicability. Spin-current density functional theory Vignale and Rasolt 1987 ; Vignale and Rasolt 1988 ; Bencheikh 2003 (SCDFT) provides a resolution, in principle, to the above difficulties. In this theory the spin current (as well as the regular current) is a legitimate basic variable, even when vector potentials and spin-orbit couplings are considered. The latter cannot be ignored in applications for spintronics. Vasili et al. 2025 ; Picozzi 2025 ; Mellnik et al. 2014 ; Fert 2008 A remarkable exact condition — the U(1) \times SU(2) gauge invariance of the xc-energy functional Vignale and Rasolt 1988 ; Bencheikh 2003 imposes, by construction, proper xc-torques. In practice, however, the development of SCDFT has long been hindered by the lack of suitable density functional approximations which could be applied to atomistic systems with reasonable success. This has changed dramatically in recent years with the advent of U(1) \times SU(2) gauge-invariant xc functional of the Meta-GGA (MGGA) class, which use the kinetic energy density and its spin-dependent generalization as building blocks Pittalis et al. 2017 ; Desmarais et al. 2024a ; Desmarais et al. 2025 . Because these building blocks are naturally expressed in terms of single-particle orbitals, they are most efficiently handled in the framework of the generalized Kohn-Sham (GKS) theory Seidl et al. 1996 ; Desmarais et al. 2024b . The GKS theory is a sharpening of the traditional KS theory in which the minimization of the energy with respect to densities is replaced by a minimization with respect to single-particle orbitals, in terms of which the densities are expressed. This framework has been shown to be at least equivalent to, and possibly more general than the traditional KS framework under broad assumptions of noninteracting v -representability Seidl et al. 1996 ; Desmarais et al. 2024b ; Nre . In this work, we show that SpinCurrent-DFT within the GKS-MGGA framework provides a resolution of the xc-spin torque problem not only in principle, but also in practice. In addition, physically sound non-collinear equilibrium conditions as well as adiabatic dynamics can be obtained also when vector potentials and spin-orbit couplings are included. Spin-Current-DFT – the basics . The KS equations in SpinCurrent-DFT have the form