Showcase #12 · Batch #4
Quantum Time-Evolution (Defense: Signal-Processing Dynamics / Quantum Algorithms for Differential Equations)

XY Trotter — Dynamics / Quantum Algorithms for PDEs

Trotter-Suzuki decomposition IS the quantum-computing analog of Strang splitting for classical PDEs — same numerical convergence discipline, now on a quantum simulation kernel.

Smith, Kim, Pollmann, Knolle (2019) — Simulating quantum many-body dynamics on a current digital quantum computerarXiv
XY Trotter — Dynamics / Quantum Algorithms for PDEs

Log-log Trotter step size dt vs L2 statevector error, across k ∈ {1, 2, 4, 8, 16, 32, 64} Trotter steps at fixed total time T = 2/J. Measured convergence order p = 2.00 in the asymptotic regime, matching the analytical second-order Strang prediction. Every factor of 2 reduction in step size yields factor of 4 error reduction — textbook second-order convergence, cleanly demonstrated on a quantum simulation kernel.

p = 2.00
Measured Trotter convergence order — matches analytical Strang O(dt²) prediction exactly

[ overview ]

What this reproduces & why it matters

Trotter-Suzuki decomposition is the quantum-computing analog of Strang splitting for classical PDEs. The same numerical-analysis question that governs classical PDE integrator convergence — how does global error scale with step size Δt? — governs Trotter quantum simulation. Second-order symmetric (Strang) Trotter has provable O(Δt²) global error; first-order Lie-Trotter is only O(Δt). This showcase demonstrates that convergence order directly, in the way a computational-physics group would validate a classical PDE integrator.

Directly aligned to the Naval Research Laboratory Quantum Algorithms for Differential Equations (QADE) program (Demirdjian et al., arXiv:2103.10352), the flagship DoD program on quantum methods for PDE-class problems. Applications span signal-processing dynamics, wave-equation kernels, ocean-acoustic transport, and radar cross-section electromagnetic simulation — all PDE classes that eventually target quantum simulation at scale.

This showcase reproduces Smith, Kim, Pollmann & Knolle (npj QI 5, 2019) — a second-order Strang Trotter implementation on the 1D uniform XX chain — with fidelity ≥ 0.9999 at every snapshot and measured convergence order p = 2.00 exactly matching the analytical Strang O(dt²) prediction.

[ verified results ]

Every number below is [PASS]-checked in source.

verified
Measured convergence order p
matches analytical Strang O(dt²) prediction exactly
2.00
Trotter fidelity vs exact (T = 2/J, k = 64 steps)> 99.999%
Trotter fidelity vs exact (every snapshot in dynamics)≥ 0.9999
Domain-wall imbalance residual at T = 2/J
Trotter I(T) vs exact I(T)
5.9e-3
Circuit gate count per Trotter step
N = 6, PBC, XX + YY layers, standard 2-qubit decomposition
12 CX
Reference benchmark validation
compared against full 2^N matrix exponential (64-dim Hamiltonian)
exact

[ method ]

How it's built

Uniform XX Hamiltonian H = −J Σᵢ (XᵢXᵢ₊₁ + YᵢYᵢ₊₁) on N = 6 sites with periodic boundary conditions. Domain-wall initial state |000111⟩. Second-order Strang Trotter: U(dt) ≈ exp(−iH_XX/2·dt) · exp(−iH_YY·dt) · exp(−iH_XX/2·dt), with each bond term decomposed to standard CX + single-qubit rotations.

Two experiments: (1) domain-wall dynamics evolved to T = 2/J with 20 snapshots, Trotter vs exact matrix exponential compared site-by-site; (2) convergence-order sweep across step counts k ∈ {1, 2, 4, 8, 16, 32, 64} at fixed T, log-log fit of ||ψ_T − ψ_E||₂ vs dt in the asymptotic regime (fidelity ≥ 0.9).

[ figures ]

Physics visuals

Domain-wall dynamics: exact vs Trotter heatmaps + imbalance trajectory
Domain-wall relaxation on N = 6 XX chain. Left: exact ⟨Z_j(t)⟩ heatmap from matrix exponential. Middle: Strang Trotter ⟨Z_j(t)⟩ heatmap — visually indistinguishable from exact. Right: domain-wall imbalance I(t) trajectories overlaid — exact (cyan) and Trotter (purple dashed) sit on top of each other through the full relaxation + recurrence oscillation. Residual at T = 2/J: 5.9e-3.

[ mitigation ]

What Qubital's ZNE buys you here

Baseline reproduction uses noiseless statevector simulation to establish the algorithmic convergence claim against exact matrix exponentiation. Adding Richardson ZNE + PEC to the Trotter loop is scoped as Phase I extension work — the same mitigation layer Qubital ships on chemistry runs. The analytical convergence reference makes mitigation-quality assessment quantitatively precise at every noise level, which is more informative than mitigation on problems without a known exact answer.

[ references ]

Papers & sources

  • Smith, A., Kim, M. S., Pollmann, F., Knolle, J. (2019). "Simulating quantum many-body dynamics on a current digital quantum computer." npj Quantum Information 5, 106.
    arXivThe reproduction target
  • Childs, A. M., Su, Y., Tran, M. C., Wiebe, N., Zhu, S. (2021). "Theory of Trotter error with commutator scaling." Phys. Rev. X 11, 011020.
  • Demirdjian, R. et al. (2021). "Variational quantum solutions to the advection-diffusion equation for applications in fluid dynamics."
    arXivNRL Quantum Algorithms for Differential Equations program — direct Navy alignment
  • Suzuki, M. (1991). "General theory of fractal path integrals with applications to many-body theories and statistical physics." J. Math. Phys. 32, 400.

[ what's next ]

Roadmap for this showcase

roadmap
  • Fourth-order Suzuki decomposition — measure O(dt⁴) convergence, verify against analytical prediction
  • Real IBM Heron hardware run — validate simulator-baseline convergence on physical NISQ hardware with ZNE
  • Extend to disordered XX chain (Anderson-localization dynamics) — direct-relevance to disordered-metamaterial characterization for AFRL
  • Time-dependent Hamiltonians via Magnus expansion — bridges to optimal-control problems
  • Direct partner engagement: NRL Quantum Algorithms for Differential Equations (QADE) program

[ request access ]

Want to run this yourself?

The physics-showcases repo is currently private, protecting IP pre-revenue. Physicists, quantum-industry contacts, and investors: reach out and I'll set up a technical walkthrough, call, or Loom.

Request a walkthrough