End-to-End Analysis of Charge Stability Diagrams with Transformers

2025-08-21

AI Summary

To address low accuracy, poor generalizability, and high adaptation costs in identifying triple points and analyzing connectivity within charge stability diagrams of semiconductor quantum dot arrays, this work proposes a Transformer-based end-to-end object detection framework. By explicitly modeling global image structure, the method directly localizes triple points and infers their topological connectivity without manual feature engineering or architecture retraining. Experiments across three heterogeneous quantum dot platforms demonstrate significantly higher triple-point detection and connection recognition accuracy than CNN baselines, with over 40% improvement in inference speed and seamless cross-platform plug-and-play capability. This is the first device-agnostic, scalable automated stability diagram parser, providing a unified, robust, AI-driven tool for critical spin-qubit control tasks—including virtual gate calibration, charge-state initialization, drift compensation, and pulse sequence generation.

Technology Category

• Machine Learning: Quantum Machine Learning
• Knowledge Representation and Reasoning: Qualitative Reasoning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Security and Privacy: Data transparency and provenance
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Transformer models and end-to-end learning frameworks are rapidly revolutionizing the field of artificial intelligence. In this work, we apply object detection transformers to analyze charge stability diagrams in semiconductor quantum dot arrays, a key task for achieving scalability with spin-based quantum computing. Specifically, our model identifies triple points and their connectivity, which is crucial for virtual gate calibration, charge state initialization, drift correction, and pulse sequencing. We show that it surpasses convolutional neural networks in performance on three different spin qubit architectures, all without the need for retraining. In contrast to existing approaches, our method significantly reduces complexity and runtime, while enhancing generalizability. The results highlight the potential of transformer-based end-to-end learning frameworks as a foundation for a scalable, device- and architecture-agnostic tool for control and tuning of quantum dot devices.

Problem

• Identifying triple points and connectivity in quantum dot arrays
• Reducing complexity and runtime for device calibration tasks
• Enhancing generalizability across different spin qubit architectures

Innovation

• Transformer-based object detection for quantum diagrams
• End-to-end learning without retraining across architectures
• Reduced complexity with enhanced generalizability in quantum control