Classical Simulation and Design Frontiers for IBM's Doped Clifford Sampling Experiment
Abstract
We classically simulate the IBM doped Clifford random circuit sampling experiment, comprising $70$ qubits, $70$ entangling layers, and $468$ inserted T gates, as reported in the first version of arXiv:2607.25941. A deterministic temporal-boundary tensor network contraction approach is specifically designed to tackle such open-boundary one-dimensional brickwork circuits with operator-Schmidt-rank-$2$ entangling gates. For an n-qubit circuit of depth d, the resulting unsliced path evaluates an exact amplitude with contraction width ⌈ d/2⌉; Ratcatcher calculations certify that no smaller width is possible for the tested instances. Because one-qubit gates are absorbed without changing the network topology, the width and dense scheduled contraction cost are independent of their values and of the number and placement of T gates. For this initial instance, its largest intermediate tensor contains 235 complex64 entries (256 times smaller than IBM's estimation), corresponding to a tensor payload of $256$ GiB, and is distributed across eight GPUs within a node. Using 32 nodes, with eight NVIDIA H100 GPUs per node, we completed all 2051 amplitude batches corresponding to IBM's published output bitstrings in 37.3 minutes. The resulting probabilities yield a log-XEB estimate of $0.35034$ with a 95% interval of $[0.29763,0.40305]$. Under the Porter--Thomas and scrambled-noise assumptions, this is numerically compatible with IBM's fidelity lower bound; separately, fidelity-weighted resource accounting projects a 583-contraction workload with a 10.6-minute makespan on the same 32 nodes. More broadly, the approach provides a practical diagnostic for experimental outputs and a quantitative tool for designing future doped Clifford sampling experiments.
Related
- Classical Clifford+T Sampling without Computing Marginals
- Polynomial-time classical simulation of boson and fermion sampling under thermal loss
- Classical Simulation of Lossy Gaussian Boson Sampling beyond Square-Root Regime
- Exact Clifford Optimality for Two-Sided Locally Randomized Classical Shadows
- Polynomial-time simulation of non-Clifford quantum error correction
BibTeX
@misc{manabe2026classical,
title = {{Classical Simulation and Design Frontiers for IBM's Doped Clifford Sampling Experiment}},
author = {Hidetaka Manabe and Hanfeng Gu and Feng Pan},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2608.13110},
eprint = {2608.13110},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.13110},
}
RIS
TY - UNPB
TI - Classical Simulation and Design Frontiers for IBM's Doped Clifford Sampling Experiment
AU - Hidetaka Manabe
AU - Hanfeng Gu
AU - Feng Pan
PY - 2026
DA - 2026/10/29
PB - arXiv
DO - 10.48550/arxiv.2608.13110
UR - https://arxiv.org/abs/2608.13110
AB - We classically simulate the IBM doped Clifford random circuit sampling experiment, comprising $70$ qubits, $70$ entangling layers, and $468$ inserted T gates, as reported in the first version of arXiv:2607.25941. A deterministic temporal-boundary tensor network contraction approach is specifically designed to tackle such open-boundary one-dimensional brickwork circuits with operator-Schmidt-rank-$2$ entangling gates. For an n-qubit circuit of depth d, the resulting unsliced path evaluates an exact amplitude with contraction width ⌈ d/2⌉; Ratcatcher calculations certify that no smaller width is possible for the tested instances. Because one-qubit gates are absorbed without changing the network topology, the width and dense scheduled contraction cost are independent of their values and of the number and placement of T gates. For this initial instance, its largest intermediate tensor contains 2^35 complex64 entries (256 times smaller than IBM's estimation), corresponding to a tensor payload of $256$ GiB, and is distributed across eight GPUs within a node. Using 32 nodes, with eight NVIDIA H100 GPUs per node, we completed all 2051 amplitude batches corresponding to IBM's published output bitstrings in 37.3 minutes. The resulting probabilities yield a log-XEB estimate of $0.35034$ with a 95% interval of $[0.29763,0.40305]$. Under the Porter--Thomas and scrambled-noise assumptions, this is numerically compatible with IBM's fidelity lower bound; separately, fidelity-weighted resource accounting projects a 583-contraction workload with a 10.6-minute makespan on the same 32 nodes. More broadly, the approach provides a practical diagnostic for experimental outputs and a quantitative tool for designing future doped Clifford sampling experiments.
ER -