内容来源:IBM——Quantum System Two模块化量子计算平台
Text丨Peixian/Langweixian Typesetting丨Peixian
In-depth good article: 1200 words丨6 minutes to read
Abstract: IBM is leveraging its Watsonx platform and Granite AI model to combine AI technology with quantum computing to enhance quantum computing capabilities and accelerate its adoption.
Over the past year, there has been a growing focus on how quantum computers can converge and connect with classical computing architectures. Quantum computers can be used as accelerators to perform certain complex computational tasks that are beyond the capabilities of even classical supercomputers.
Classical computers or servers are used for pre-processing for quantum algorithms and circuit development, as well as post-processing for error management, result optimization, and task completion. It's clear from the growing number of AI use cases that AI can augment classical computing. In the same way, AI should enhance the power of quantum computing, and many companies are working to achieve this.
Although many people and companies are beginning to talk about quantum and artificial intelligence together, they are technically different. Artificial intelligence is the training and application of neural network models on a classical computing platform driven by traditional binary processing logic elements such as CPU, GPU, NPU, DSP, and FPGA.
Quantum computers, on the other hand, use a new computing architecture to construct quantum physical systems such as superconducting qubits, etc., and complete calculations through the evolution of quantum systems to solve complex problems. Although there are significant differences in hardware, software, and supporting systems, researchers are already trying to integrate the two, which is also helping to advance quantum computing. IBM is one of the pioneers in this development.
IBM is considered a giant in quantum computing, making progress in hardware, software and system technology, and the quantum computers they develop are beginning to be deployed around the world. On the other hand, IBM is also a leader in AI technology with its WatsonX platform, which has made a lot of progress since its victory in the Jeopardy game show in 2011. Since then, Watsonx has evolved into a scalable, enterprise-grade platform with AI studios, data, management solutions, and assistant solutions.
Now, IBM is bringing these two technologies together to enhance quantum computing and accelerate its adoption. Recently, IBM outlined how AI technology can be integrated into its Qiskit software for quantum computing to improve the ease of use of SDK tools and OpenQASM3 (Open Quantum Assembly Language). IBM leverages its watsonx generative AI platform to provide developer support through Granite AI models to assist in the generation of quantum code. In addition, IBM is researching and developing new AI models to improve other key problems in quantum computing, such as optimizing superconducting quantum circuits, managing resources; and suppress, mitigate, and correct errors in quantum computing.
IBM is also launching the Qiskit Code Assistant service with the Visual Studio extension, and plans to offer two quantum chatbots — one for developers and one for general users of IBM Quantum services.
When it comes to circuit optimization, AI models can embed the transpiler service as a plug-in into the Qiskit SDK or in combination with heuristics. According to IBM, the transpiler service can better map abstract circuits to quantum ISA circuits, resulting in 40% larger circuits, better quality, and 2 to 5x faster processing.
For resource management, IBM is developing AI solutions to better estimate quantum runtimes, flag workloads that are likely to fail, and differentiate circuits for parallel processing, making better use of classical and quantum resources for parallel computing, including AI supercomputing clusters.
The heterogeneous data center of the future will include QPU IBM
Combined with IBM's ambitious goal of 100 million quantum gates for quantum circuits by 2029 and 1 billion quantum gates by 2033, quantum computing will accelerate into the deployment of quantum real-world applications in the coming years. So in 2029, we could see heterogeneous data centers that combine state-of-the-art CPUs, GPUs, and QPUs (quantum processing units).
IBM Quantum Development & Innovation Roadmap IBM
On the other hand, large model training and deep learning of artificial intelligence are also the most computing power-consuming fields at present, is it possible to use the super computing power of quantum computing to accelerate it? At present, IBM has not disclosed the progress in this regard, but Tsinghua University and Bose Quantum Company have disclosed that they have completed a research entitled "Optical Quantum Machine Training of Multi-layer Neural Networks", and proposed the Ising training algorithm of multi-layer neural networks, which is one of the few cases in the world of using dedicated quantum computers to accelerate neural network training, representing an important breakthrough in the application of quantum computing technology to AI. Taking advantage of quantum parallelism, Bose Quantum's coherent optical quantum computer can solve large-scale binary optimization problems in milliseconds, providing another possible path for AI training in the era of large models.
It is believed that in the near future, with the unremitting efforts of Chinese and foreign scholars, quantum computing and artificial intelligence will eventually combine to bring a better life to mankind with stronger computing power.
Sources:
https://www.forbes.com/sites/tiriasresearch/2024/06/24/ibm-develops-the-ai-quantum-link/
https://mp.weixin.qq.com/s/apkUsmQ5fuC1t7loiB00ZQ
It is hereby noted that the translation of this article by Quantum Outpost is for information transmission and reference only, and does not mean that it agrees with the views and data in this article.
——end——