Antonio Corcoles | Three Layers of Qiskit | QDC 2025

Antonio Corcoles | Three Layers of Qiskit | QDC 2025

by Antonio Corcoles

Formal & Physical Sciences Computing & Cybersecurity UYComputer scienceUYXNatural computing
🎙 Antonio Corcoles 👥 203K 📅 December 1, 2025 ⏱ 36 min 👁 702 📄 expert opinion 🧭 2026-08-16
Available in: English (current) Français

Keywords

Qiskitquantum computingexecution modelerror mitigationHPC integration

Summary

Antonio Corcoles presents IBM’s evolving quantum software stack and the new execution model designed to help users move toward quantum-classical advantage. He explains the distinction between a QPU and a quantum computer, and introduces the concept of a quantum ISA (instruction set architecture) as the core of quantum computing. The talk is structured in three layers: quantum information science, algorithms, and HPC integration. In the first layer, he introduces new tools: Samplatic, which generates circuit templates and sampling graphs; the Executor primitive for running full quantum programs; and Noise Learner v3 for box-based noise modeling. These enable advanced error mitigation techniques like shaded light cones and propagated noise absorption. In the algorithms layer, he highlights add-ons for probabilistic error cancellation, operator back propagation, and poly propagation. In the HPC integration layer, he discusses Slurm integration, Prefect-based hybrid workflows, and the Qiskit C API. Jim Garrison then demonstrates a fully compiled SQD workflow in C++ using RIKEN’s SPD solver, showcasing Qiskit’s readiness for high-performance quantum-classical workflows.

165 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into IBM’s quantum software strategy and introduces concrete tools that are being released. The argumentation is coherent, building from foundational concepts to specific implementations. The speaker demonstrates a deep understanding of the technical challenges and presents solutions that are practical for developers. However, the talk is promotional in nature, and the claims about quantum advantage are forward-looking without direct evidence.

Scientific Rigor, Source Quality, Title Accuracy

The presentation is scientifically rigorous in its technical descriptions, but it relies heavily on IBM’s own tools and publications. The speaker references several papers and open-source projects, but does not provide external validation. The title accurately reflects the content, which is well-structured around three layers. The talk includes a demo by Jim Garrison that adds credibility to the claims about HPC integration.

142 words

Title / Content Match

The title accurately reflects the content, which is structured around three layers of Qiskit: quantum information science, algorithms, and HPC integration.

Quality & Reliability

8/10

Presentation by an IBM Quantum expert detailing the Qiskit software stack and new execution model. The content is technical and specific, referencing internal tools and collaborations. However, it is a promotional talk with limited external validation.

Key Moments

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

Cited Sources

  • Benchpress paper — Referenced as a set of tests for benchmarking quantum software.
  • RustworkX — Mentioned as a tool for graph and network performance in the transpiler.
  • Qiskit Addon SQD HPC — Mentioned as a new HPC-ready add-on for SQD.
  • SPD solver from RIKEN — Mentioned as an open-source solver used in the demo.
  • Qiskit C API — Mentioned as part of the Qiskit 2.2 release.

Concurring Sources

Contribution & Novelties

The talk introduces a new execution model for Qiskit that promises to simplify advanced error mitigation and enable HPC integration. The main novelty is the Executor primitive and the Samplatic library, which abstract the complexity of error mitigation techniques. The talk also highlights the C API and C++ add-ons, which are significant for HPC users.

Pour aller plus loin :

84 words

Radar Profile

The radar profile shows high scores in information quantity and technical level, indicating a dense and technical presentation. The quality and reliability scores are slightly lower, reflecting the promotional nature and lack of external validation.

Reliability 7/10

💬 No comments were provided for analysis.