Large-scale quantum simulations of dissipative spin-1/2 Heisenberg chains

AI summary

Overview

Research area: Quantum simulation of open quantum systems — specifically, using a superconducting quantum processor to compute nonequilibrium steady states (NESS) of the dissipative anisotropic spin-1/2 Heisenberg chain.

Technical level: Advanced. The paper assumes familiarity with Lindblad master equations, Trotterization, Stinespring dilation, static structure factors, and tensor-network methods.

Scope (one sentence): The authors run Trotterized Lindblad dynamics of dissipative Heisenberg chains of up to 50 sites (100 qubits) on the IBM superconducting device ibm_kingston, extract static structure factors from 117 hardware data points, and map the steady-state phase diagram across the J_x–J_y plane.

What This Paper Is About

A quantum many-body system coupled to an environment can settle into a nonequilibrium steady state that has no equilibrium analogue, and the dissipative spin-1/2 Heisenberg chain is the standard benchmark for this physics. Classical methods disagree on its phase diagram, and controlled calculations at large system size have been out of reach because the density-matrix problem squares the Hilbert-space dimension and no free energy selects the steady state. The authors use a noisy quantum computer to simulate the chain's Lindblad dynamics directly and determine the phase diagram at sizes well beyond exact diagonalization.

Key Contributions

1. A large-scale digital Lindblad simulation on hardware. The authors simulate dissipative Heisenberg chains of up to L = 50 sites (100 qubits including ancillae) on the 156-qubit superconducting processor ibm_kingston, at entangling-gate depths of up to 1700. They state that deterministic Lindblad evolution toward a nontrivial ensemble-averaged NESS, scanned across a many-body phase diagram, had not previously been realized on quantum hardware at any scale.

2. A dissipation-based error-resilience argument, demonstrated in data. They articulate and confirm the mechanism by which the engineered dissipation, not the hardware, dominates the long-time state: noise enters as a weak competing dissipator, and increasing the Trotter step Δt strengthens the target Liouvillian relative to the hardware noise floor at the cost of Trotter error.

3. A phase diagram from 117 hardware data points. They measure XX correlation matrices, compute S(q), and classify steady states by peak position — ferromagnetic (q = 0), antiferromagnetic (q = π), spin-density-wave (incommensurate q), and paramagnetic — across the J_x–J_y plane at fixed J_z = γ = 1, for system sizes L = 40, 42, and 50.

4. Identification of a Trotter-induced "incipient" SDW phase. They trace anomalous incommensurate signals in the far corners of the diagram to chiral three-site terms generated by the Trotter decomposition, and propose deliberate Trotterization as a route to engineering magnetic phases in dissipative spin systems.

Main Findings

• Four steady-state phases resolved on hardware: FM, AFM, SDW, and PM are distinguished by the peak structure of the static structure factor S(q). Representative points at L = 40 (80 qubits) agree with small-scale exact diagonalization at L = 12 and with tensor-network calculations at the same size L = 40 (the SDW ordering vectors from the tensor-network runs are shown as dot-dashed lines).

• The paramagnet is much smaller than mean field predicts. The most striking difference from the single-site mean-field phase diagram is the absence of a large central PM region; the paramagnet is confined to the vicinity of the dark-state line J_x = J_y, where the fixed point is |1⟩^⊗L.

• A re-entrant paramagnetic region appears at large couplings. This is attributed to a different mechanism than the dark-state paramagnet: as interactions grow stronger the state becomes increasingly mixed and loses transverse order. The authors note this re-entrance is predicted by cluster mean-field results and is susceptible to finite-momentum instabilities.

• Crossovers, not sharp transitions. Because the model is one dimensional, no spontaneous symmetry breaking is expected; the hardware data show finite-size remnants of mean-field orders separated by crossovers. The authors state their structure-factor diagnosis can locate crossovers but cannot certify a phase boundary.

• Order survives to long times and deep circuits. In the L = 50 periodic chain at (J_x, J_y) = (3, 2) with J_z = γ = 1, the ZZ correlation between nearest neighbors and between the two furthest-apart sites settles at a nonzero value that does not degrade as the circuit deepens (shaded bands are the standard error of the mean). The XX correlation reaches a nonzero steady value only for the nearest-neighbor pair.

• Trotter step size behaves counterintuitively under noise. With Δt = 0.3, the evolution to t = 30 (depth 1700) lands closer to the target NESS than a smaller step would, opposite to the noiseless expectation. A minimal single-spin model gives contrast C ≈ γΔt / (γΔt + ε), increasing monotonically with Δt.

• An incipient SDW phase appears in the corners. Signals in the upper-left and lower-right corners, absent from both mean-field theory and the L = 12 numerics, are attributed to chiral terms Z_{m−1}Y_m X_{m+1} − X_{m−1}Y_m Z_{m+1} and cyclic permutations generated by the Trotter decomposition, confirmed by small exact diagonalization including those terms. This phase is absent in the exact diagonalization runs that use dt = 0.05.

• Step size controls the instability. At (J_x, J_y) = (4, −2) with L = 40, Δt = 0.2 leaves the system largely paramagnetic with a weak residual ferromagnetic component, while Δt = 0.3 triggers the finite-momentum instability and clear incommensurate peaks.

• Cut scans show the crossover sequence. At J_y = 0 (L = 40), the system evolves from ferromagnet through a paramagnet at J_x = 0 to an antiferromagnet. At J_y = 1.5, it starts in an incommensurate q phase, shifts to coexisting FM and AFM peaks with FM losing dominance as J_x decreases, becomes paramagnetic at J_x = −3, and shows an incipient SDW at J_x = −4.

Methodology in Plain English

The team splits the Lindblad time evolution into two pieces — coherent Hamiltonian evolution and incoherent spin decay — and applies them alternately in a first-order Trotter approximation with time step Δt. The Hamiltonian part is a standard quantum circuit. The decay part is implemented by Stinespring dilation: every system qubit is paired with its own ancilla qubit, coupled through a controlled-R_y(θ) rotation and a CNOT with θ = 2 sin⁻¹(√(1 − e^(−γΔt))), and the ancillae are reset after every Trotter step. Because physical and ancilla qubits alternate along the device, the entangling-gate depth is 17 per Trotter step.

The initial state is chosen to match the observable: a fully polarized up state in the measurement basis, i.e. |0⟩^⊗L ⊗ |0⟩^⊗L or |0⟩^⊗L ⊗ |+⟩^⊗L for ZZ and XX correlators respectively. The state is evolved to t = 30 with Δt = 0.3, and the steady state is estimated by averaging the last ten time steps (t ∈ [27.3, 30]). Static structure factors are obtained by Fourier transforming the measured correlation matrix.

Only error suppression is used, both from Qiskit Runtime: Pauli twirling on every two-qubit entangling gate, and XY4 dynamical decoupling on idling qubits. No error mitigation (zero-noise extrapolation, probabilistic error cancellation) is applied; the reported values are raw hardware measurements. Benchmark comparisons come from exact diagonalization at L = 12 and tensor-network simulations (the tensor jump method) at L = 40.

Why This Matters

Impact on research. For roughly a decade, each new classical method redrew the phase diagram of this benchmark model. The authors present their 117 measured points and the underlying S(q) curves at L = 40–50 as a quantitative target for classical contenders — asking at what bond dimension the tensor-jump method (which reached 1000 spins at bond dimension four) reproduces them, and whether variational-perturbative and neural-network ansätze agree. Because hardware systematics (noise, Trotter error) are independent of classical systematics (bond dimension, ansatz bias), agreement between the two carries information about the model rather than about either method.

Potential real-world applications (as motivated by the paper's framing):

• Designing and characterizing driven-dissipative quantum devices whose useful behavior depends on a steady state rather than on coherent unitary evolution.
• Studying magnetic ordering in open quantum materials where coupling to an environment cannot be neglected.
• Engineering synthetic magnetic phases on purpose via controlled Trotterization, instead of treating Trotter error purely as a defect.
• Benchmarking and validating classical simulation algorithms for open quantum systems against hardware ground truth.

Industry relevance. The result argues that decoherence, usually the central obstacle for quantum computing, is partly neutralized by the problem itself: a dissipative steady state is an attractor that erases accumulated errors at a rate set by the Liouvillian spectral gap. For hardware vendors and algorithm developers, that reframes noisy intermediate-scale devices as usable instruments for open-system physics rather than only for closed-system simulation. The paper does not report any application-specific benchmarks, runtimes, or fidelities beyond the 117 data points and the gate depths, so quantitative industrial claims cannot be drawn from it.

Future Directions

• Move beyond one dimension. The authors state this is trivial in their framework because the methods carry over directly, and note that true dissipative transitions are expected in higher dimensions, where classical methods are most strained.
• Extend to driven models. Limit cycles and time crystals are named as additional targets for hardware simulation.
• Deliberate Trotter engineering. Since the stability mechanism is agnostic to which Liouvillian is simulated, Trotter-generated terms could be used intentionally to realize otherwise costly interactions or to tune the universality class of a dissipative process — the quantum contact process is cited as an example.
• Quantitative error mitigation and sharper phase boundaries. The authors note that sharpening the crossover-versus-transition distinction will require quantitative error mitigation beyond the suppression techniques used here, and that the method degrades where large coupling amplitudes amplify Trotter error (the anomalous points at |J_α| ≳ 4).

Target Audience

Researchers in quantum simulation, open quantum systems, and nonequilibrium statistical mechanics; developers of classical tensor-network and neural-network methods for Lindblad dynamics who need a benchmark target; and quantum-hardware practitioners interested in error-resilient algorithm design. The paper is not beginner-friendly — it presumes comfort with Liouvillian spectra, Stinespring dilations, and structure-factor diagnostics.

Authors’ abstract

A quantum many-body system coupled to an environment relaxes to a nonequilibrium steady state that can sustain order with no equilibrium counterpart. Computing such steady states is harder than closed-system dynamics as the density matrix problem squares the Hilbert-space dimension, and no free energy selects the steady state. The dissipative spin-1/2 Heisenberg chain is a benchmark example for nonequilibrium steady state physics; various methods have each calculated its phase diagram but do not agree, and a controlled determination at large system size has remained out of reach. Here we simulate the Lindblad dynamics of chains of up to 50 sites on the superconducting processor ibm_kingston -- 100 simultaneously active qubits at up to 1700 entangling-gate depths -- realizing the dissipation via Stinespring dilation. The system's dissipative evolution is a self-correcting mechanism that effectively erases errors, so hardware noise enters only as a weak competing dissipator. We measure static structure factors and resolve ferromagnetic, antiferromagnetic, spin-density-wave, and paramagnetic steady states, mapping the phase diagram with 117 quantum hardware data points across the $J_x$--$J_y$ plane. We uncover a rich non-equilibrium phase diagram of ordered phases with only remnants of the mean-field order, and where sharp transitions give way to the crossovers expected in one dimension. We also find the existence of an incipient (Trotter-induced) spin density wave phase, highlighting the potential of controlled Trotterization as a tool to engineer various magnetic phases in dissipative spin systems. Our quantum simulations largely settle the lingering uncertainty regarding the correct phase diagram of this benchmark system. Moreover, they show that quantum computers are now a feasible tool for addressing scientific questions involving dissipative quantum systems.