Enhanced analogue quantum computation with a logically-encoded XX gadget

AI summary

Overview

Research area: Quantum computing — analogue quantum annealing, Hamiltonian simulation, and quantum gadget constructions for near-term hardware.

Technical level: Intermediate. The paper assumes familiarity with Pauli operators, the transverse-field Ising model, quantum annealing schedules, and Trotterization, but its central construction is simple enough to follow with that background.

Scope: The paper introduces and experimentally benchmarks the "truss gadget," an exact encoding that produces effective two-body XX interactions on quantum hardware that natively provides only one-body X and Z fields plus two-body ZZ interactions.

What This Paper Is About

Quantum annealers and many analogue quantum devices can natively implement only a limited set of interactions: local transverse (X) fields, local longitudinal (Z) fields, and two-body ZZ couplings. Two-body XX interactions are known to improve annealing performance and to widen the class of many-body Hamiltonians that can be simulated, but they are not natively available on platforms such as superconducting annealers. The authors construct an exact gadget, built from only the natively available terms, that generates an effective XX interaction, and they test it on two different quantum processors.

Key Contributions

1. The truss gadget: An exact construction that realises two-body XX interactions using only one-body X and Z fields and two-body ZZ interactions. Unlike perturbative gadgets, it does not rely on an asymptotically large energy penalty, so the effective XX coupling is exact by construction and there is no logical subspace to leak out of.

2. A resource analysis: The qubit overhead is at worst quadratic in system size. If 3- and 4-local ZZ interactions are available, the hardware overhead is exactly one additional qubit per chain of XX interactions. The gadget is implementable with the transverse-field Ising model under homogeneous driving.

3. Dynamical validation on trapped ions: A Trotterized truss implementation on the IonQ Forte processor is compared against exact continuous-time evolution, a noiseless digital simulation of the target XX Hamiltonian, and a direct hardware implementation of that Hamiltonian.

4. Annealing demonstration on superconducting hardware: The gadget is applied as an annealing catalyst to weighted maximum-independent-set instances on the D-Wave Advantage2_system2 QPU, with an empirically determined approximate thousand-fold improvement in the probability of finding a state close to the ground state relative to implementation without an XX interaction.

Main Findings

• Faithful reproduction of XX dynamics: On a closed loop of n = 5 physical qubits (mapped to n − 1 = 4 logical degrees of freedom, spanning 2^(n−1) states), the truss implementation on IonQ Forte closely tracked exact continuous-time evolution at the level of individual many-body basis states, including states |00011⟩ and |01010⟩.

• Quantified agreement: The time-averaged total variation distance from exact evolution was D̄_TV ≃ 0.04 (standard deviation ≈ 0.04) for the noiseless second-order Trotter evolution and D̄_TV ≃ 0.095 (standard deviation 0.02) for the truss implementation on hardware. Time-averaged Kullback–Leibler divergences were approximately 0.01 and 0.05, respectively.

• No postselection required: The direct XX implementation on hardware required postselecting onto the encoded even-parity subspace, discarding approximately 20% of the measured population throughout the evolution window (N_steps = 10 at each evolution time). The truss implementation retains all measured outcomes, since every physical bitstring has a well-defined image under the parity map.

• Gap enhancement from exact diagonalisation: For the K₂,₅ MWIS instance with hardness parameter δw = 1.4, adding an XX coupling between qubits A₁ and A₂ raised the minimum spectral gap from ΔE^min₀₁(s) = 0.0094 to 0.0419, and suppressed accumulation of the first-excited-state overlap near the perturbative crossing at s ≈ 0.85.

• Annealer performance gain: An approximate thousand-fold improvement in the probability of finding a state close to the ground state was determined empirically for the studied instances on D-Wave Advantage2_system2.

• Trotterization accuracy: The alternating first-order Suzuki–Trotter scheme, which becomes a symmetric second-order (Strang) decomposition over pairs of steps with local error O(Δt³), reproduced exact dynamics closely in the noiseless case.

Methodology in Plain English

The trick is to stop treating one physical qubit as one logical qubit. Instead, each logical qubit is encoded in the parity of two neighbouring physical qubits, so that the logical Z operator is the product of two physical Z operators, Z̃ᵢ ≡ Ẑᵢ Ẑᵢ₊₁.

For a single pair of logical qubits, three physical qubits are driven with transverse fields of independently tunable strength. Because a single physical spin flip changes the parity of two adjacent pairs, flipping one physical spin flips two logical qubits at once — exactly the action of an XX interaction. The resulting logical Hamiltonian contains single-body X̃ terms and an X̃X̃ term whose coefficient is the field strength on the middle qubit, so the effective XX coupling can be tuned, and its sign set, purely by inhomogeneous driving.

Chaining these three-qubit units into a one-dimensional structure — resembling a Warren truss, hence the name — produces a chain of XX-coupled logical qubits, at the cost of losing single-body X driving on the internal physical qubits. Adding ZZ interactions lets logical Z fields and logical ZZ couplings be implemented with physical two-body terms; a logical ZZ between non-adjacent logical qubits requires a physical four-body ZZZZ term, and coupling the truss to unencoded qubits requires a physical three-body ZZZ term.

For the trapped-ion test, the authors ran a five-qubit closed ring, choosing physical nearest-neighbour ZZ = 0.4, next-nearest-neighbour ZZ = 1.0, and transverse field d = 1.0, corresponding to a target with h = 0.4, J_zz = 1.0 and unit-strength XX. They compared exact evolution, noiseless digital evolution, truss hardware evolution, and direct XX hardware evolution, initialising the physical case in the paramagnetic state with Hadamard gates and the encoded case with Hadamards plus CNOTs computing parity onto a fifth qubit.

For the annealing test, the researchers took a K₂,₅ bipartite maximum weighted independent set instance: the two vertices in sub-graph A each have weight 2.5, the five vertices in sub-graph B each have weight 1 − δw/5, giving a global optimum of weight 5, a local optimum of weight 5 − δw, and a Hamming distance of seven between them. The penalty strength was J_zz = 5, and fields and couplings obtained from the encoding were normalised to unity. Replacing the two A qubits with a three-qubit truss encoded their logical Z operators as parities, and because these qubits couple to the five B qubits, the original two-body ZZ terms became three-body ZZZ terms; additional auxiliary variables C_k were introduced to reduce these back to two-body terms, since the D-Wave annealer cannot implement three-body interactions directly.

Why This Matters

• Removes a hardware bottleneck: The spectral benefits of higher-order transverse interactions can now be pursued on existing devices rather than waiting for new native couplings, and the gadget extends the class of many-body Hamiltonians that analogue devices can simulate.

• Exact rather than perturbative: Because there is no penalty-enforced logical subspace, there is no accuracy-versus-strength trade-off and no subspace for errors to leak into — a structural advantage over earlier perturbative gadgets for XX couplings.

• Real-world applications (as motivated by the paper's problem classes):

  • Combinatorial optimisation mapped to Ising Hamiltonians, including travelling salesperson, maximum satisfiability, minimum graph-colouring, and maximum weighted independent set.
  • Simulation of many-body Hamiltonians that require two-body transverse couplings.
  • Annealing protocol design in which a tailored XX "catalyst" softens perturbative crossings in hard instances.
  • Benchmarking and validation of analogue quantum hardware against known exact dynamics.

• Industry relevance: The construction runs on the native interaction sets of both superconducting annealers (D-Wave) and gate-based trapped-ion processors (IonQ), the two vendors used in the experiments. No new hardware capability is needed — only a change in how the problem is encoded and driven.

Future Directions

1. Scaling beyond short chains: The demonstrated dynamics used five physical qubits and four logical ones; how the construction performs for long chains, and how its at-worst-quadratic qubit overhead behaves in practice, remains to be tested at scale.

2. Exploiting higher-order ZZ availability: Platforms offering 3- and 4-local ZZ interactions would reduce the overhead to one extra qubit per XX chain; whether such devices can be built or accessed is an open hardware question.

3. Noise and open-system behaviour: The authors argue there is no logical subspace to leak out of, but the trapped-ion comparison still showed a larger deviation on hardware than in the noiseless Trotter case (D̄_TV ≃ 0.095 versus ≃ 0.04); characterising the gadget under realistic decoherence is a natural next step.

4. Broader instance surveys and coupling placement: The paper treats an illustrative set of weighted maximum-independent-set instances; systematic rules for where to place XX couplings, and how the hardness parameter δw governs the achievable advantage, remain to be developed beyond the single detailed K₂,₅ example with δw = 1.4.

5. Other platforms: Neutral-atom annealers are cited as an alternative quantum-annealing platform, but were not used here; porting the gadget to them is untested.

Target Audience

Quantum information researchers working on annealing, analogue quantum simulation, and Hamiltonian encodings; experimentalists at superconducting and trapped-ion facilities who need higher-order transverse couplings from limited native interaction sets; graduate students seeking a worked example of a gadget construction validated end-to-end on real hardware; and industry practitioners evaluating whether encoding-level tricks can extract more performance from current annealers.

Authors’ abstract

The introduction of two-body XX interactions has been shown to improve the performance of quantum annealing in solving certain combinatorial optimisation problems and would enable the simulation of a broader class of many-body Hamiltonians. Due to experimental constraints, such interactions are typically not natively available on analogue quantum devices. To overcome this limitation we introduce the truss gadget. This is an exact construction that generates two-body XX interactions using only one-body X and Z fields and two-body ZZ interactions which are readily accessible on, for instance, superconducting quantum annealers. We experimentally validate the construction by reproducing the dynamics of a target XX Hamiltonian on a trapped-ion quantum processor. We then apply the gadget to an illustrative set of instances of the weighted maximum-independent-set problem on a superconducting quantum annealer. We empirically determine an approximate thousand-fold improvement in the probability of finding a state close to the ground state for these instances, over implementation without an XX interaction.