Coogee '26 Talks - Paul Webster (Iceberg Quantum)

Coogee '26 Talks - Paul Webster (Iceberg Quantum)

by Paul Webster

Formal & Physical Sciences Computing & Cybersecurity UYComputer scienceUYXNatural computing
🎙 Paul Webster 👥 137 📅 February 19, 2026 ⏱ 63 min 👁 365 📄 expert opinion 🧭 2026-08-16
Available in: English (current) Français

Keywords

Pinnacle Architecturequantum LDPC codesRSA-2048fault-tolerant quantum computingoverhead reduction

Summary

Paul Webster presents the Pinnacle Architecture, a fault-tolerant quantum computing architecture that uses quantum low-density parity-check (QLDPC) codes to reduce the spacetime overhead of universal quantum computation. The architecture is modular, consisting of processing units, magic engines, and memory, and it achieves a significant reduction in physical qubits for factoring RSA-2048 integers: less than 100,000 physical qubits at a physical error rate of 10^-3, compared to nearly a million qubits in previous work. The talk covers the motivation, the compilation approach based on pi-product measurements, the design of processing blocks with generalized bicycle codes, and the performance simulations using circuit-level noise and most-likely-error decoding. The speaker emphasizes that the architecture maintains low time overhead while achieving space savings, and it is designed to be scalable and compatible with various hardware platforms. The results are presented as a preview, with the paper to be released soon.

145 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into a novel quantum computing architecture, with a clear argument for the advantages of QLDPC codes over surface codes. The speaker justifies the choice of RSA-2048 as a benchmark and explains the compilation method in detail. The argumentation is solid, supported by simulations and extrapolations, though the lack of published paper limits immediate verification.

Scientific Rigor, Source Quality, Title Accuracy

The talk demonstrates scientific rigor by referencing prior work (e.g., Gidney’s result, IBM’s work) and discussing limitations, such as the need for real-time decoding. The sources cited are primarily from the speaker’s own group and other researchers in the field. The title accurately reflects the content, and the talk is well-structured.

125 words

Title / Content Match

The title accurately reflects the content: a talk on the Pinnacle Architecture and its application to RSA-2048 factoring.

Quality & Reliability

8/10

The talk presents a novel architecture with detailed technical content, but the paper is not yet on arXiv, and the results rely on extrapolations and simulations. The speaker is transparent about limitations, and the work is presented as a preview.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • arXiv preprint (to be released) — The paper describing the Pinnacle Architecture is not yet on arXiv but will be available soon.
  • Gidney's work on factoring RSA-2048 — Previous best result for factoring RSA-2048 with ~1 million physical qubits.
  • IBM's work on LDPC codes — Recent work on low-space-overhead LDPC architectures with time trade-offs.

Concurring Sources

  • Gidney's work on factoring RSA-2048 — Provides a benchmark for comparison, showing the improvement.

Contribution & Novelties

The Pinnacle Architecture introduces a modular fault-tolerant quantum computing design that leverages QLDPC codes to achieve an order-of-magnitude reduction in physical qubits for RSA-2048 factoring compared to previous work. It addresses the time-overhead trade-off seen in earlier LDPC architectures and provides a scalable framework. The talk also introduces the concept of ‘magic engines’ for efficient magic state distillation and injection.

Pour aller plus loin :

89 words

Radar Profile

The radar profile shows high scores in quantity and technical level, indicating a dense, technical talk. Quality and reliability are slightly lower due to the unpublished nature and reliance on extrapolations.

Reliability 7/10

💬 No comments were provided for analysis.