Quantinuum H2 Series: 56-Qubit Quantum Simulator

• Quantinuum H2 Series is a 56-qubit trapped-ion quantum computing platform divided into four zones, enabling complex digital simulation of quantum dynamics.
• It achieves high performance with native two-qubit gate fidelities of 99.94(1)% and circuit depths exceeding 2000 two-qubit gates per run for robust simulations.
• The system integrates advanced error mitigation strategies and supports studies of emergent phenomena like Floquet prethermalization and quantum hydrodynamics.

Quantinuum's H2 Series is a 56-qubit, gate-based trapped-ion quantum computing platform architected for large-scale quantum simulations in previously classically inaccessible regimes. Built from four independent zones of 14 ^171Yb^+ ions each, interconnected by ion transport, the H2 supports all-to-all entanglement within each zone via native Mølmer–Sørensen interactions, enabling complex multi-zone quantum circuits. The system is specifically engineered for high-fidelity digital quantum simulation tasks, attaining native two-qubit partial entangler fidelities of 99.94(1)%, circuit volumes exceeding 2000 two-qubit gates per run, and advanced mid-circuit capabilities. This architecture establishes H2 as a practical testbed for studying digitized quantum dynamics, benchmarking classical heuristics, and directly observing emergent phenomena such as Floquet prethermalization and quantum hydrodynamics (Haghshenas et al., 26 Mar 2025).

1. System Architecture and Hardware Capabilities

Quantinuum H2 employs trapped-ion technology using ^171Yb^+ ions, structured into four physically independent “zones” containing 14 ions each. Zones are interconnected via ion transport, enabling distribution and interaction of quantum information across the 56-qubit device. Within a zone, any pair of qubits can be entangled by native, all-to-all Mølmer–Sørensen interactions. The hardware supports single-qubit rotations of the generic form

$$U_{1q}(\theta,\phi) = \exp\left[-i\frac{\theta}{2}(X\cos\phi + Y\sin\phi)\right],$$

and native two-qubit partial entanglers

$$U_{ZZ}(\phi) = \exp\left[-i\frac{\phi}{2} Z\otimes Z\right],$$

as its fundamental gate set. The platform features single-qubit gate infidelity of approximately $3\times10^{-5}$ and average two-qubit fidelity of 99.94(1)% for $U_{ZZ}(0.5)$ gates, verified by sign-averaged cycle benchmarking. Mid-circuit measurement, reset, and qubit reuse further extend the versatility for quantum algorithms and simulation protocols (Haghshenas et al., 26 Mar 2025).

Feature	Implementation Details	Quantitative Metrics
Qubit Type	^171Yb^+ trapped ions	56 qubits (4 zones × 14 qubits)
Native 2Q Entangling Gate	Mølmer–Sørensen	Fidelity: 99.94(1)% (U_ZZ(0.5))
Single-Qubit Gate Infidelity	Arbitrary single-qubit rotations	~3×10⁻⁵
Circuit Depth	Gate-based (all-to-all intra-zone)	>2000 two-qubit gates/run
Mid-Circuit Capabilities	Measure/reset/reuse	Supported

2. Digital Simulation of Quantum Magnetism

enables digital simulation of realistic quantum systems by employing discrete quantum gates to approximate continuous time evolution. A principal demonstration featured the study of digitized dynamics of the two-dimensional transverse-field Ising model on a $7\times8$ torus,

$$H = J\sum_{\langle j,k \rangle} Z_j Z_k + h\sum_j X_j,$$

with $J<0$, $h=2|J|$. The evolution operator $U(t) = e^{-iHt}$ was digitized via a second-order Trotter-Suzuki formula:

$$U(dt) \approx e^{-i(dt/2)H_X} e^{-i dt H_{ZZ}} e^{-i(dt/2)H_X} \equiv \mathcal{U}.$$

Each Trotter step involves four layers of $N/2=28$ $U_{ZZ}$ gates (totaling 112 two-qubit gates per step) plus single-qubit rotations, achieving circuit depths above 2000 two-qubit gates in a single experiment. This scale enables exploration of time and size regimes where classical techniques are not practical or trustworthy (Haghshenas et al., 26 Mar 2025).

3. Floquet Prethermalization and Hydrodynamic Phenomena

Digitized simulation on H2 revealed the emergence of Floquet prethermalization in the quantum Ising model, characterized by the persistence of prethermal plateaus in the disconnected order-parameter fluctuations,

$$\langle Z_{\mathrm{tot}}^2 (s) \rangle = \frac{1}{N^2} \sum_{j,k} \langle Z_j Z_k \rangle_s,$$

up to $s \approx 20$ circuit steps. This duration corresponds to classically inaccessible timescales; classical tensor network methods struggled beyond $s \approx 9$ on comparable instances. The Floquet framework mapped the dynamics to an effective Hamiltonian,

$$\mathcal{D} = i \log(\mathcal{U})/dt = H^{(0)} + H^{(2)} dt^2 + O(dt^4),$$

with explicit forms for $H^{(0)}$ and $H^{(2)}$. Generic Floquet systems heat to infinite temperature, but a large prethermal window

$\tau_H \sim \exp(c/|Jdt|)$

was observed for $dt=0.25/|J|$, supporting long-lived coherent many-body phenomena.

Further, H2 enabled the engineering of an inhomogeneous "stripe" quench on a $14\times4$ torus to probe emergent hydrodynamics. The measurement of the $y$-averaged local ZZ-energy density $E(x)$ and its Fourier modes

$$\widetilde{W}(q,s) = \sum_x e^{iqx} E(x) / \sqrt{L_x}$$

demonstrated decay for non-zero $q$ modes as

$$\widetilde{W}(q,s) \propto e^{-\Gamma_q s}, \quad \Gamma_q \approx \mathcal{D} q^2,$$

with measured diffusion constant $\mathcal{D}=0.38(5)$. This provided direct evidence of late-time (diffusive) energy transport in a strongly interacting digital quantum system (Haghshenas et al., 26 Mar 2025).

4. Error Mitigation Strategies

To maintain fidelity in deep quantum circuits, H2 deployed a synergistic error mitigation protocol combining four primary techniques:

1. Dynamical Decoupling: Phase-alternating $X$ pulses mitigate coherent memory errors.
2. Randomized Compiling: Pauli twirling converts remaining coherent error channels into stochastic Pauli channels.
3. Leakage Detection and Zero-Noise Regression (ZNR): Specialized gadgets suppress leakage errors with manageable overhead instead of prohibitive post-selection.
4. Pauli-Insertion Zero-Noise Extrapolation (ZNE): Experiments run at two amplified noise levels, followed by extrapolation using ZNE designed to minimize estimator variance under an exponential-decay noise model.

These combined approaches enabled statistically consistent estimates of key observables, such as $\langle Z_{\mathrm{tot}}^2 \rangle$ up to $s=20$ time steps in deep circuits with over 2000 two-qubit gates, sustaining fidelity at a level inaccessible to classical heuristics within realistic resource constraints. Cycle benchmarking consistently established the two-qubit infidelity at $6(1)\times10^{-4}$ (Haghshenas et al., 26 Mar 2025).

5. Comparison to State-of-the-Art Classical Methods

enabled digital quantum simulations beyond the reach of leading classical algorithms for the same models and scales. Classically, several advanced tensor network and neural-network-based methods were benchmarked:

• 1D MPS simulations with bond dimensions up to $\chi=4000$ were accurate to $s \lesssim 20$ only at low effective temperatures, but failed beyond $s \lesssim 9$ at intermediate temperatures due to fidelity collapse.
• 2D PEPS Schrödinger-picture (full-update, $D\leq10$) converged for $s \lesssim 3$, and Belief-Propagation PEPS ($D\sim22$) reached $s \lesssim 8$.
• PEPO Heisenberg-picture with BP compression was accurate to $s \lesssim 8$.
• Sparse Pauli Heisenberg methods with $\sim4\times10^9$ terms (1 TB RAM) converged to $s \lesssim 3$.
• Neural-network quantum states (single-layer CNNs, up to 40 features, with t-VMC) offered only qualitatively reasonable results to $s \lesssim 4$.

In contrast, H2 produced statistically robust quantum simulation results for $s=20$ steps and beyond, implementing greater circuit depth and volume at high fidelity. The experiments placed digital quantum hardware in regimes of continuous-time dynamics that are currently classically inaccessible, demonstrating capability for benchmarking and potentially superseding classical heuristics (Haghshenas et al., 26 Mar 2025).

6. Scientific Significance and Application Domains

Quantinuum H2 establishes digital quantum computers as practical tools for studying the dynamics of complex quantum matter, particularly in continuous-time regimes previously inaccessible to classical simulation. The platform enables extraction of transport coefficients, investigation of late-time hydrodynamics, and quantitative benchmarking of emergent many-body phenomena, such as Floquet prethermalization.

H2's architecture and protocols set a standard for digital quantum hardware in algorithmic flexibility, error mitigation, and experimental depth. These attributes underscore the platform’s relevance for accelerating exploration of open quantum system dynamics, condensed matter models, and the frontier benchmarking of classical simulation algorithms, thereby framing an agenda for subsequent hardware and algorithmic advances (Haghshenas et al., 26 Mar 2025).