Quantum simulation of field-tunable spin spectroscopy of the quantum magnet Cs2CoCl4 on a trapped-ion quantum computer

AI summary

Overview

Research area: Quantum simulation of strongly correlated quantum matter — specifically, computing the dynamical spin structure factor (DSF) of a quantum magnet on noisy quantum hardware, and benchmarking it against classical tensor-network methods and inelastic neutron scattering (INS).

Technical level: Advanced. The paper assumes familiarity with DMRG, matrix product states, Trotter product formulas, retarded Green's functions, the fluctuation–dissipation relation, and error mitigation on near-term devices.

Scope (one sentence): The authors simulate the field-dependent magnetic excitation spectra and spin dynamics of a 36-site spin-1/2 XXZ chain model of Cs₂CoCl₄ at four transverse fields on IonQ's 36-qubit Forte Enterprise processor, using circuits of only 420 two-qubit gates per time point, and validate the results against DMRG and neutron-scattering data.

What This Paper Is About

Quantum computers are promising tools for predicting collective excitations in quantum magnets, but hardware noise limits how deep a circuit can be, so it is unclear how much reliable magnetic information survives aggressive circuit compression. The authors ask how much of the field evolution of a magnetic spectrum can be recovered on a fixed quantum circuit budget, using the anisotropic XXZ chain that describes Cs₂CoCl₄ as a test case. They reconstruct momentum- and frequency-resolved spin response from hardware measurements and check it against two classical DMRG benchmarks and against experimental inelastic neutron scattering.

Key Contributions

1. An end-to-end compressed-circuit spectroscopy workflow. The authors combine an ancilla-free local kick protocol, DMRG/tensor-network construction of target states, variational "staircase" circuit compression, and error mitigation to reconstruct retarded spin response on a 36-qubit trapped-ion processor. Every nonzero time point is encoded in a circuit with 420 two-qubit gates — a constant gate budget from the first time step to the longest evolution of 20 Trotter steps (Jt = 12).

2. A two-tier validation framework that separates solver error from model error. DMRG-CV (correction-vector DMRG, L = 100, m = 1000 retained states, fixed broadening η = 0.1 J) tests whether the quantum solver reproduces the response of the same Hamiltonian; DMRG+Trotter (L = 36, 20 second-order Trotter steps of Δt = 0.6/J) matches the hardware's system size, discretization, and observation window; and INS comparisons test whether the model describes the actual material. Propagated measurement uncertainty is reported separately from sensitivity to processing choices.

3. A reconstruction-free physical signature of the field-driven magnetic transition. Direct equal-time measurements from the t = 0 circuits reveal a sign reversal of the connected nearest-neighbor zz correlator, a local observable that does not depend on spectral reconstruction or the finite observation window.

4. A test of finite-memory extension (FME) of the measured time records as a way to recover longer-time (lower-frequency) information without additional quantum circuits, compared against longer DMRG+Trotter calculations.

Main Findings

• Field-driven spectral redistribution is recovered on hardware. The reconstructed dynamical structure factors capture the evolution from a broad two-spinon continuum at zero field toward a dispersive branch with magnon character near the expected critical field h_c ≈ 1.604 J, consistent with DMRG-CV references that extend to h_x/J = 2.
• Quantitative agreement with DMRG-CV. At the two fields with a complete total response (h_x/J = 0 and 0.764, corresponding to 0 T and 1 T under the rescaling h_x/J ≈ 0.764 B/T), error mitigation yields Pearson correlations and structural similarity above 0.9 and normalized mean squared errors below 0.15, with fitted amplitudes within about 20% of the reference.
• Quantitative agreement with neutron scattering. Pearson correlations with INS exceed 0.85 (Table 1 of the paper).
• Peak positions agree within resolution. At q = π, total-spectrum peak offsets are below the nominal time-window scale of 0.06 meV for t_T = 12/J, while the taper gives an effective resolution near 0.10 meV. The zero-field continuum bandwidth, however, remains underestimated.
• Compression retains substantial fidelity. Ground-state preparation fidelities are F₀ = 0.905, 0.892, 0.915, and 0.997 at h_x/J = 0, 0.764, 1.146, and 1.604, using (252, 315, 232, 77) two-qubit gates respectively; time-evolved states use 420 two-qubit gates with a minimum channel-averaged compression fidelity of 0.62 (averaged above 0.6 over the spin channels).
• Local spin dynamics survive noise. For the on-site autocorrelation Re C^zz(t) at j_c = 17, error-mitigated hardware curves have Pearson correlations of at least 0.952 with DMRG-CV and MSE values no greater than 0.02891 ± 0.00755, after normalization by the equal-time modulus.
• Sign reversal of short-range correlations. At the three lower fields C(r = 1) is negative and C(r = 2) positive; both signs reverse at h_x = 1.604 J, where hardware gives C(r = 1) = +0.0889 ± 0.0294 and C(r = 2) = −0.0488 ± 0.0197 after error mitigation, the latter following the DMRG-CV value of −0.070. The positive nearest-neighbor value is already present in raw data, so the reversal does not depend on error mitigation.
• The reversal precedes the expected transition. Both DMRG benchmarks place the nearest-neighbor zero crossing at h_x = 1.561 J, below h_c ≈ 1.604 J.
• Errors grow near the critical field. The longitudinal response follows the classical field evolution at all four fields, but intensity and shape errors grow near h_c (Table 2 of the paper).
• Finite observation time is a real limitation. Low-frequency structure and closely spaced excitations are not fully accessible; FME improves spectral reconstruction in a way that depends on field and observable.

Methodology in Plain English

The authors study a one-dimensional chain of 36 spins whose interactions mimic the magnetic behavior of the material Cs₂CoCl₄, using the measured exchange J = 0.23 meV and anisotropy Δ = 0.25, with open boundaries and a transverse magnetic field.

To probe how the magnet responds to a disturbance, they "kick" a single spin near the middle of the chain (site 17) with a ±90-degree rotation. This trick, which requires no extra ancilla qubit, converts the quantity they want — a retarded commutator of spin operators — into ordinary spin expectation values. Taking the difference between the two kick signs isolates the response; symmetry arguments let some spin channels be extracted from a single kick instead.

The hard part is time evolution, which normally needs ever-deeper circuits. Instead, the authors compute each time-evolved state classically as a matrix product state (using DMRG with bond dimension χ = 64 for the ground state, then second-order Trotter steps of size Δt = 0.6/J up to 20 steps), and then train a shallow "staircase" circuit of nearest-neighbor two-qubit blocks (up to D = 4 layers, up to 124 Evenbly–Vidal optimization sweeps) to reproduce each target state. Because every time point becomes its own fixed-cost state-preparation problem, the quantum circuit stays at 420 two-qubit gates regardless of how far in time the simulation reaches.

The 36-qubit IonQ Forte Enterprise processor samples each circuit, with basis rotations (H, HS†, identity) selecting the x, y, or z spin component. The measured correlations are Fourier transformed in space and time — using a half-DPSS taper with time-bandwidth product NW = 1.5 — and converted into the dynamical structure factor via the zero-temperature fluctuation–dissipation relation. Error mitigation combines a mirror-circuit protocol with lineshape/spectral correction strategies (the t = 0 correlation plot uses only the mirror-circuit mitigation). Finally, the reconstructed spectra are scored against DMRG-CV, DMRG+Trotter, and published INS data from 70 mK measurements using Pearson correlation, structural similarity, and normalized mean squared error.

Why This Matters

The paper provides a quantitative, multi-reference benchmark showing that a noisy quantum processor can recover genuine magnetic spectral information — not just qualitative shapes — when paired with classical circuit compression and error mitigation. It also makes a methodological distinction that matters for the field: validating a quantum solver against a classical calculation of the same model is a different question from validating the model against experiment, and the authors report both separately.

• Interpreting neutron-scattering experiments: The workflow connects model parameters to measured magnetic response, offering a way to test candidate spin Hamiltonians against INS data for real materials.
• Materials discovery pipelines: Fixed-depth compressed circuits suggest a route to using near-term quantum hardware as a spectral solver inside materials workflows where entanglement growth limits DMRG at long times or in higher dimensions.
• Benchmarking quantum hardware for science: The reported metric suite (Pearson, structural similarity, NMSE, fitted amplitude) gives hardware vendors and users a reproducible yardstick for "quantum utility" claims in many-body spectroscopy.
• Magnetic sensing and device design: Understanding how field-driven changes reorder local spin correlations and excitation character is relevant to designing and characterizing magnetic materials used in sensors and information storage.

Industry relevance: The work is a collaboration between IonQ (trapped-ion hardware and software), Oak Ridge National Laboratory, and Los Alamos National Laboratory, and it demonstrates a concrete application workflow for a commercially available 36-qubit system, including an explicit accounting of quantum versus classical compute cost.

Future Directions

• Extending to more complex magnets: The authors frame the present study as a basis for extending compressed-circuit spectroscopy to systems beyond a single one-dimensional chain, such as frustrated or higher-dimensional magnets where DMRG and quantum Monte Carlo struggle.
• Recovering low-frequency and closely spaced features: Finite observation windows remain the limiting factor; the paper's finite-memory extension is an initial test, and its benefits depend on field and observable, leaving room for better classical post-processing or longer effective records.
• Reducing classical overhead and improving fidelity near the critical field: The gate count excludes the substantial classical cost of building targets and optimizing circuits, and spectral accuracy degrades near h_c — both are open bottlenecks for scaling the approach.
• Sharpening model-versus-material validation: The INS comparison is currently limited by differences in broadening, background, and statistical uncertainty relative to the DMRG-CV references; a more careful resolution-matching treatment would strengthen claims about how well the XXZ model describes Cs₂CoCl₄.

Target Audience

Quantum-computing researchers working on near-term algorithms for many-body physics and spectroscopy; condensed-matter theorists and neutron-scattering experimentalists interested in what current quantum hardware can and cannot say about real magnetic materials; and quantum-hardware developers looking for realistic, quantitatively benchmarked applications. Readers will need comfort with DMRG/MPS methods and linear-response theory to follow the technical details; the appendices referenced in the text (DMRG convergence, error-mitigation protocol details, and additional spin channels) are not included in the excerpt summarized here.

Authors’ abstract

The collective excitation spectra of quantum magnets provide a direct test of quantum computers as tools for studying strongly correlated matter. Hardware noise limits the circuit depths available for these calculations, leaving the question of how much magnetic information can survive circuit compression. Using IonQ's 36 qubit Forte Enterprise processor, we simulate the magnetic excitation spectra and spin dynamics of a 36 site Heisenberg XXZ chain describing the quantum magnet Cs2CoCl4 in a transverse field using quantum circuits containing just 420 two qubit gates. The workflow combines an ancilla free local kick protocol with classical tensor network compression, encoding both ground state preparation and time evolution. The reconstructed dynamical structure factors capture the evolution from a broad two spinon continuum toward a dispersive branch with magnon character near the expected critical field. We observe an interesting field driven sign reversal of the nearest neighbor spin correlations which shows a complementary signature of this magnetic evolution. Our error mitigation methods bring the reconstructed total spectra into close agreement with density matrix renormalization group calculations, with Pearson correlation and structural similarity above 0.9 and normalized mean squared error below 0.15. We also perform comparisons with experimental inelastic neutron scattering measurements to connect the recovered excitations to the material's magnetic response. Our results demonstrate that circuit compression and error mitigation allow noisy quantum processors to recover both collective excitations and local magnetic correlations providing a quantitative benchmark for quantum spectroscopy and a basis for extending these methods to more complex magnetic systems.