Nvidia has introduced Ising, a suite of open source quantum AI models designed to enhance the development and practical application of quantum processors. This marks the first family of open source AI models specifically targeting the quantum computing sector. The Ising models aim to address significant challenges such as processor calibration and quantum error correction, which are critical for the scalability and reliability of quantum computing systems.
The Ising suite is positioned as a transformative tool for current quantum processors, enabling them to run useful applications by accelerating calibration and error correction processes. Nvidia claims that the Ising models deliver up to 2.5 times faster performance and three times higher accuracy for quantum decoding tasks compared to existing standards like pyMatching. The suite includes Ising Calibration, a vision-language model that automates the interpretation of quantum processor measurements, potentially reducing calibration time from days to hours.
Ising Decoding features two variants of a 3D convolutional neural network, optimized for either speed or accuracy, to perform real-time decoding for quantum error correction. These models are already in use by a variety of organizations, including quantum computing companies, academic institutions, and national laboratories. The Ising suite allows users to run models locally, providing full control over data and reducing reliance on external infrastructure.
The adoption of Ising's calibration and decoding models spans institutions across North America, Europe, and Asia. Nvidia supports this ecosystem with additional resources such as workflow documentation, training data, and NIM microservices, enabling organizations to tailor the Ising models to specific hardware and use cases. The models and data are accessible on GitHub, promoting ongoing open development.
Ising complements Nvidia's CUDA-Q software platform for hybrid quantum-classical computing, facilitating direct integration with Nvidia's NVQLink QPU-GPU hardware interconnect for real-time control and quantum error correction. This comprehensive workflow from calibration to operational use in advanced computing systems is part of Nvidia's broader range of open model offerings, which are available through GitHub for the research and developer communities.
Source: https://www.techmonitor.ai/news/nvidia-unveils-ising-open-source-ai-suite-for-quantum-calibration