A partnership between Quantum Machines and Nvidia is taking the industry one step closer to achieving error-corrected quantum computers. The collaboration has led to a significant improvement in calibration techniques for quantum processors, with the potential to significantly enhance performance and fidelity.
The partnership focused on using reinforcement learning models running on Nvidia’s DGX platform to optimize control over qubits in a Rigetti quantum chip. By constantly adjusting pulses in near real-time, the team was able to improve performance and maintain high fidelity over time.
This breakthrough is crucial for unlocking fault-tolerant quantum computing, which requires error correction. The current approach relies heavily on classical computer engines, but this collaboration has shown that using general-purpose compute engines can be effective.
Nvidia’s DGX platform proved essential in this process, providing the necessary power and scalability to tackle complex control problems. The partnership aims to create open-source libraries to simplify access to these tools for researchers.
While the individual result is a small step towards solving key quantum computing challenges, it marks an important milestone. As Nvidia plans to release more powerful Blackwell chips next year, this collaboration will likely drive further innovation in the field.
Source: https://techcrunch.com/2024/11/02/quantum-machines-and-nvidia-use-machine-learning-to-get-closer-to-an-error-corrected-quantum-computer