arXiv:2609.21249
The Pinnacle Architecture with fixed connectivity of degree eight
Paul Webster, Tom Peham, Lawrence Z. Cohen
Abstract
We show how the Pinnacle architecture can be adapted to be compatible with superconducting qubits, where the connectivity between qubits must be fixed at fabrication. Specifically, we present a modified instantiation of the architecture that has fixed connectivity degree of eight. With this instantiation, we show that a 2048-bit RSA integer can be factored in one month with approximately 120 000 physical qubits, given a physical error rate of $10^{-3}$, code cycle time of 1 microsecond and reaction time of 10 microseconds.
AI Summary
The paper asks whether the Pinnacle fault-tolerant architecture’s low-overhead factoring estimates survive the requirement that qubit connectivity be fixed at fabrication, as for superconducting hardware. It constructs a “Frozen Pinnacle” architecture with maximum fixed connectivity degree eight, replacing dynamically repositioned measurement gadgets with fixed gadgets and code-block cyclic-shift automorphisms implemented through existing connections. Using the $[[510,16,24]]$ generalised bicycle code, each processing block allocates eight logical qubits to processing and eight to memory. Fixed gadget chains retain arbitrary logical Pauli-product measurements across the processing unit in one logical cycle. Automorphisms increase that cycle from 26 to 30 code cycles. The construction also supplies memory access and a magic engine producing one $|\overline{\mathrm{CCZ}}\rangle$ state every five logical cycles, with estimated infidelity $2\times10^{-10}$ and a 1% rejection rate. Optimising a prior RSA-2048 compilation gives 120,820 physical qubits for factoring within one month, assuming physical error rate $10^{-3}$, a $1\,\mu\mathrm{s}$ code cycle and $10\,\mu\mathrm{s}$ reaction time. Circuit-level depolarising-noise simulations test memory, shifts and bridged measurements at distances 4, 6 and 10. Fitting a subthreshold exponential ansatz and extrapolating to distance 24 gives an estimated per-logical-qubit, per-cycle error rate of $1.4\times10^{-18}$, below the stated requirement of approximately $10^{-14}$. Full-distance error rates are extrapolated because direct sampling is infeasible; the quoted confidence intervals come from the fit. The resource estimate also assumes negligible routing overhead within the magic engine and relies on prior compilation and magic-state-cultivation results.
Central claim may not hold: The reliability required for RSA-2048 factoring is supported by extrapolating distance-4, -6 and -10 simulations to distance 24, with confidence intervals that exclude uncertainty in the extrapolation model. The one-month resource estimate assumes negligible magic-state routing overhead without demonstrating a timed routing schedule compatible with the required production and injection throughput.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 8 pages, 2 figures