Nonlinear quantum multi-spin dynamics without entanglement

AI summary

Overview

Research area: Quantum physics / condensed-matter theory, specifically the interface between classical atomistic spin dynamics (Landau-Lifshitz and Landau-Lifshitz-Gilbert equations) and quantum spin dynamics for arbitrary spin quantum number s.

Technical level: Advanced. The paper is a mathematical-physics construction: it uses variational principles (Dirac-Frenkel time-dependent variational principle), Rayleigh-Lagrange-d'Alembert dissipation, Onsager gradient flow, and Lie-algebra (su(d)) coherence-vector representations. A reader needs graduate-level quantum mechanics and some familiarity with geometric formulations of dynamics.

Scope in one sentence: The paper defines a product-state (entanglement-free) variational manifold of local quantum density operators and derives, on that manifold, nonlinear quantum Landau-Lifshitz and Landau-Lifshitz-Gilbert equations of motion whose computational cost scales as O(N) for N spins.

What This Paper Is About

Classical atomistic spin dynamics replaces each atomic magnetic moment by a three-component classical vector, which is cheap to simulate but throws away local quantum structure that matters for spin s ≥ 1, where a local density operator carries more information than three spin expectation values. Fully quantum many-body treatments keep that structure but cost exponentially in system size. The authors ask what quantum spin dynamics survives when intersite entanglement is deliberately excluded while each local quantum state is kept in full, and they derive the resulting equations of motion for a product-state ansatz, showing how the quantum Landau-Lifshitz (q-LL) and quantum Landau-Lifshitz-Gilbert (q-LLG) formulations differ and when they coincide.

Key Contributions

1. A product-state variational framework for arbitrary spin. The state is taken as ϱ(t) = ⊗_{j=1}^{N} ϱ_j(t), with each ϱ_j an arbitrary pure or mixed density operator on the local spin-s Hilbert space of dimension d = 2s+1, restricted to an isospectral orbit O_j = {U_j ϱ_j(0) U_j† : U_j ∈ SU(d)}. Intersite entanglement is excluded by construction; local quantum structure is retained.

2. Derivation of a local q-LLG equation. Using the Dirac-Frenkel time-dependent variational principle for the conservative part plus a Rayleigh-Lagrange-d'Alembert construction for dissipation, the authors obtain ϱ̇_j = (i/ℏ)[ϱ_j, h_j] + i κ_j [ϱ_j, ϱ̇_j], with a Rayleigh function ℛ = ½ Σ_j η_j Tr_j ϱ̇_j², damping parameter κ_j ≡ η_j/ℏ > 0, and a local effective mean-field Hamiltonian h_j = Tr_j̄[(1_j ⊗ ϱ_j̄) ℋ].

3. Derivation of a local q-LL equation from a different variational structure. Using an Onsager principle of least dissipation with the functional O[{K_j}] = dE/dt|_D + Φ, with quadratic cost Φ = ½ Σ_j (1/λ̃_j) Tr_j(K_j²), the authors obtain ϱ̇_j = (i/ℏ)[ϱ_j, h_j] − λ̃_j[ϱ_j, [ϱ_j, h_j]], establishing that q-LL and q-LLG follow from inequivalent variational/geometric constructions once local states may be mixed.

4. A unified coherence-vector form and a classical limit check. Both equations are rewritten for generalized coherence vectors x⃗_j in ℝ^(d²−1), with a Lie-algebra cross product (x⃗_j ∧ h⃗_j)a = f_abc x{j,b} h_{j,c}, reducing exactly to the classical LL and LLG equations for d = 2 (s = 1/2). The paper also states two distinct routes back to classical behaviour (s → ∞, and large number of quantum spins N) and a criterion for q-LL/q-LLG equivalence under a common time rescaling.

Main Findings

• Nonlinearity has two separate sources. The equations are nonlinear both because the mean-field Hamiltonians h_j depend on the instantaneous states of all other sites (even without damping) and because the dissipative term i κ_j [ϱ_j, ϱ̇_j] depends nonlinearly on the local state, even for a single non-interacting site.

• The spectrum, not just the vector length, is conserved. Restricting each ϱ_j to an isospectral orbit preserves Tr ϱ_j^m for m = 1, …, d, hence the spectrum, Hermiticity, and positivity of each local state. This is presented as the quantum analogue of conservation of |m_j| in the classical equations.

• Energy is monotonically dissipated in q-LLG. For a time-independent Hamiltonian, dE/dt = Σ_j Tr_j(h_j ϱ̇_j) = −Σ_j ℏ κ_j Tr_j(ϱ̇_j²) ≤ 0.

• Coherence-vector dynamics look like Landau-Lifshitz dynamics with a generalized cross product. The conservative part becomes ẋ⃗_j = ℏ⁻¹ x⃗_j ∧ B⃗_j with B_{j,a} = −∂E/∂x_{j,a} = −2h_{j,a}; q-LL becomes ẋ⃗_j = ℏ⁻¹ x⃗_j ∧ B⃗_j − λ̃′_j x⃗_j ∧ (x⃗_j ∧ B⃗_j) with λ̃′_j = 2λ̃_j, and q-LLG becomes ẋ⃗_j = ℏ⁻¹ x⃗_j ∧ B⃗_j − κ′_j x⃗_j ∧ ẋ⃗_j with κ′_j = 2κ_j.

• The coherence-vector norm is conserved. Total antisymmetry of the structure constants f_abc implies x⃗_j · ẋ⃗_j = 0 and hence d|x⃗_j|²/dt = 0 in both q-LL and q-LLG, mirroring conservation of |m_j| classically. With the normalization Tr(Λ_a Λ_b) = 2δ_ab, one has Tr(ϱ_j²) = 1/d + 2|x⃗_j|².

• q-LL and q-LLG are inequivalent in general. The paper states that equivalence, which holds classically up to a time rescaling, is no longer fundamental once local states can be mixed; the inequivalence is attributed to the different variational and geometric structures (Rayleigh/Gilbert versus Onsager double-bracket) on the quantum state manifold, and a criterion for equivalence under a common time rescaling is derived. The detailed derivation of this section is cut off in the supplied text.

• Two distinct classical limits. Classical behaviour can emerge either from s → ∞ (suppressed relative weight of extra local quantum degrees of freedom) or from large N with finite local spin (collective observables approach classical behaviour while finite local quantum corrections persist at each site). The authors stress these mechanisms need not give identical microscopic dynamics.

• Computational scaling. Because the equations involve only local density operators and self-consistent mean fields, cost scales as O(N) for fixed spin s, placing the framework between classical atomistic spin dynamics and exponentially hard fully quantum many-body dynamics.

• Benchmark results are not reported in the supplied text. Section IV is described in the introduction as containing three benchmarks (first-order anti-coherence generation and sources of nonlinearity; the quantum-versus-classical relation and q-LL/q-LLG inequivalence in the large-spin limit; and the emergence of macroscopic quantum-classical agreement with increasing system size despite finite local discrepancies), but the numerical outcomes, dataset sizes, and any specific error metrics are not present in the content provided.

Methodology in Plain English

Start with the classical LL and LLG equations and notice two things about them: each spin feels a local field produced by all the other spins, and the cross-product structure keeps every spin vector at fixed length. Now replace each classical vector by a full quantum density operator for that site, and replace "fixed length" by "fixed spectrum," meaning the local state is only ever unitarily rotated, never reshaped. Imposing that the total state stays a simple tensor product of these local states gives a much smaller dynamical space than the full many-body Hilbert space, but a much larger one than 3N classical numbers.

To get the reversible part of the motion, the authors apply the Dirac-Frenkel variational principle, which says the true quantum evolution is the one that minimizes the action over the allowed variations, and which produces a commutator with a mean-field Hamiltonian built by tracing out all other sites. To add damping, they borrow a trick from classical mechanics: they write a Rayleigh dissipation function proportional to the square of the local "velocity" of the density operator and add its virtual work to the variational condition, yielding a Gilbert-style term that is implicit in the time derivative. For the alternative q-LL form they instead use Onsager's principle of least dissipation, minimizing the energy decay rate plus a quadratic cost in the generator of the dissipative motion, which yields a double-commutator (double-bracket) damping term. Finally, they re-express everything in generalized Bloch-type coordinates built from the su(d) generators, where the ordinary cross product is replaced by the Lie-algebra structure constants, which makes the correspondence to the classical equations transparent and shows exactly where the two quantum forms part ways.

Why This Matters

Impact on research: The work supplies a middle path between two widely used but incompatible descriptions, classical atomistic spin dynamics (scalable, no local quantum structure) and fully quantum many-body dynamics (microscopically complete, exponentially costly). It also reframes the LL-versus-LLG relationship as a statement about variational geometry on a state manifold rather than a universal equivalence, which is a conceptual point that carries beyond magnetism. It gives a concrete answer to the question of where quantum corrections to spin dynamics can come from when entanglement is negligible: from the dimensionality and geometry of local state space, especially for s ≥ 1 where higher-rank multipolar fluctuations and coherences are not captured by three spin components.

Potential applications (as motivated in the paper):

• Atomistic simulations of magnetic materials at O(N) cost with local quantum corrections retained.
• Magnetization dynamics in localized-spin systems coupled to many environmental and electronic degrees of freedom, where decoherence is argued to suppress intersite quantum correlations on magnetization-relevant time scales.
• Modeling higher-spin magnetic ions, where s ≥ 1 local coherences and multipolar fluctuations are dynamically relevant.
• Interpreting and benchmarking fully quantum simulations of spin systems by providing a controlled entanglement-free reference dynamics.

Industry relevance: The framework targets the same simulation niche as classical atomistic spin dynamics used in magnetic materials and device modeling, but at comparable linear scaling. That makes it a candidate tool for settings where classical LL/LLG simulations may miss local quantum effects, without the exponential cost of a full quantum treatment.

Future Directions

• Quantify where the product-state approximation fails. The authors argue that decoherence suppresses intersite entanglement in real materials, but the error introduced by discarding it, as a function of time, temperature, and coupling, is not established in the supplied text.
• Complete and test the equivalence criterion. The paper promises a criterion for when q-LL and q-LLG produce the same trajectories after a common time rescaling; the derivation and its numerical verification in the large-spin limit are in the truncated part and need to be checked against realistic parameters.
• Characterize the two classical limits quantitatively. Understanding when large N reproduces classical behaviour versus when large s is required, and whether the two limits ever disagree for experimentally relevant observables, is posed as an open structural question.
• Extend the dissipation model. The authors note that state- and time-dependent η_j, and spatially nonlocal Gilbert damping coupling velocities at different sites (as in classical nonlocal LLG models), can be absorbed into the same structure, and that more general anisotropic cost functionals Φ are possible; exploring these is a natural next step.
• Apply to real materials. Because the scheme is O(N), a direct next step is deployment on realistic multi-spin material models to identify observables where local quantum structure changes predictions relative to classical LL/LLG.

Target Audience

Specialists in quantum spin dynamics, micromagnetics, and atomistic spin-dynamics method development; theoretical physicists working on variational and geometric formulations of open quantum dynamics; and computational researchers who need scalable quantum-aware spin models for magnetic materials. The paper assumes comfort with density operators, variational principles, and Lie algebras, so it is not suitable as an introductory text, though the introductory and concluding sections are accessible to a broader physics audience interested in where quantum corrections to Landau-Lifshitz dynamics come from.

Authors’ abstract

We introduce a distinct route for extending classical atomistic spin dynamics into the quantum regime for arbitrary spin $s$, based on a variational product-state \textit{ansatz} that includes both pure and mixed local states. By construction, intersite entanglement -- expected to remain negligible even on very short time scales in many real materials -- is excluded, while local quantum effects are retained through the nonclassical structure of $s\geq 1$ density operators. Nonlinear dynamics arise from both the mean-field interdependence of the local states and the state-dependent dissipative terms. The framework recovers classical spin dynamics for $s=\frac{1}{2}$ while allowing genuinely quantum local dynamics for higher spins. It clarifies the variational and geometric origin of the inequivalence between quantum Landau-Lifshitz and Landau-Lifshitz-Gilbert dynamics, provides a criterion for their equivalence under a common time rescaling, establishes connections with the classical theory through two qualitatively distinct limits ($s\to\infty$ and a large number of quantum spins $N$), and allows realistic $\mathcal{O}(N)$ simulations of quantum spin dynamics in materials.