IonQ’s 36-Qubit Spectroscopy Test Tracks a Quantum Magnet—With Classical Guardrails
A compressed 420-gate workflow on Forte Enterprise reconstructed how a model magnetic spectrum changes with field, then checked the result against tensor networks and neutron data.
Quantum simulation is most useful when its output can be checked against other ways of knowing. An IonQ-led team used every on a 36-qubit Forte Enterprise processor to reconstruct the field-dependent excitation spectrum of Cs2CoCl4, a quantum magnet modeled as a 36-site anisotropic spin chain.[1][2]
Compression kept the circuit within reach
The workflow compressed both ground-state preparation and time evolution with classical tensor networks, then used an ancilla-free local-kick protocol. Each hardware circuit contained 420 two-qubit gates. Across four transverse fields, the reconstructed spectrum followed the expected shift from a broad spinon continuum toward a magnon-like branch near the critical field.[1][2]
The team also recovered a field-driven sign reversal in nearest-neighbor spin correlations, giving a local signature alongside the collective spectrum.[1][2]
Two independent yardsticks
After mitigation, the total spectra agreed with density-matrix renormalization-group calculations at Pearson correlation and structural similarity above 0.9, with normalized mean-squared error below 0.15. The paper also compares the recovered excitations with published inelastic-neutron-scattering measurements.[1][2]
Those checks do not prove quantum advantage, but they make the hardware result falsifiable against both classical computation and experiment.[1][2]
The classical scaffolding matters
Classical computation is not merely a rival here; it is built into the method. Tensor networks create the compressed circuits, DMRG provides the main numerical benchmark and mitigation materially changes the hardware output. The experiment studies a one-dimensional 36-site model chosen to fit the processor. It does not show a classically intractable material calculation, and the paper is an unreviewed preprint.[1][2]
A more demanding test would increase system size or dimensionality until the classical compression and benchmark become genuinely strained, while preserving comparisons to measurable spectra. Reporting raw, mitigated and classically reconstructed results together would clarify which information came from the quantum processor and which came from the surrounding classical workflow.[1][2]