Benchmark

Tensor-Network Classical Benchmark of IBM Eagle Kicked Ising

Belief-propagation tensor-network simulation (ITensor) · Simulation · 127 qubits · ITensor

Classical-side benchmark that reproduced — and surpassed the accuracy of — IBM's 127-qubit Eagle kicked-Ising utility-scale experiment using tensor networks tailored to the heavy-hexagon lattice. The authors use belief-propagation-based approximate contraction exploiting the tree-like correlations of the state and demonstrate that the classical method is more accurate and more precise than the quantum processor results at the same Trotter depths, and can be pushed to long times in the thermodynamic limit. This benchmark reframes the 127-qubit IBM claim and is widely cited as the reference classical counter-result.

tensor-networksIBM-Eaglekicked-Isingclassical-benchmarkheavy-hex

Classical-side benchmark that reproduced — and surpassed the accuracy of — IBM's 127-qubit Eagle kicked-Ising utility-scale experiment using tensor networks tailored to the heavy-hexagon lattice. The authors use belief-propagation-based approximate contraction exploiting the tree-like correlations of the state and demonstrate that the classical method is more accurate and more precise than the quantum processor results at the same Trotter depths, and can be pushed to long times in the thermodynamic limit. This benchmark reframes the 127-qubit IBM claim and is widely cited as the reference classical counter-result.

Key Metrics
Qubits simulated
127
Lattice
Heavy-hexagon
Method
Belief-propagation tensor network
Accuracy vs IBM Eagle
Higher
Reach
Thermodynamic limit, long times
Why It Matters

Definitive classical counter-benchmark to the IBM Eagle utility-scale claim — sets the bar that future utility demonstrations must beat on heavy-hex kicked-Ising dynamics.

Hardware

Classical workstation / HPC (no quantum hardware) benchmarking IBM Eagle r3 (127 qubits)

Framework

ITensor