AI Helps Quantum Computers with NVIDIA Ising Models
The world of quantum computers is still in its infancy, requiring significant developments to reach a level where devices will be usable for industrial and scientific research calculations. There are two obstacles that need to be overcome to achieve this goal: error correction and initial calibration of the devices.
NVIDIA has announced a new family of AI models, called Ising, which promises significantly superior performance in both fields compared to the solutions used so far.
NVIDIA Ising: AI to Improve Quantum Computers
The Ising model, invented by German physicists Wilhelm Lenz and Ernst Ising, is used to describe ferromagnetic materials and has represented a significant advancement in this area, capable of simplifying the description of highly complex systems.
For this reason, NVIDIA chose Ising as the name of its new family of open AI models designed to aid in the development of quantum computers. Specifically, the Ising model family is divided into three: an Ising Calibration model, which interprets measurements from a quantum computer and continuously recalibrates the system in real time, and two Ising Decoding models, optimized to be either fast or accurate, which perform real-time decoding of the system's state to enable error correction. According to NVIDIA, the Ising Decoding models are up to 2.5 times more accurate and 3 times faster than traditional methods based on pyMatching.
The models are designed to run on NVIDIA GPUs alongside NVIDIA CUDA-Q, the software platform to enable collaboration between GPUs and QPUs, and NVIDIA NVQLINK, a technology to interconnect classical and quantum computers.
NVIDIA has reported that Ising Calibration is already in use at Atom Computing, Academia Sinica, EeroQ, Conductor Quantum, Fermi National Accelerator Laboratory, Harvard John A. Paulson School of Engineering and Applied Sciences, Infleqtion, IonQ, IQM Quantum Computers, Advanced Quantum Testbed at Lawrence Berkeley National Laboratory, Q-CTRL, and the UK National Physical Laboratory (NPL). Ising Decoding, on the other hand, is in use at Cornell University, EdenCode, Infleqtion, IQM Quantum Computers, Quantum Elements, Sandia National Laboratories, SEEQC, University of California San Diego, UC Santa Barbara, University of Chicago, University of Southern California, and Yonsei University.
NVIDIA's approach is thus significantly different from that adopted by other players in the sector: for example, IBM announced a few months ago that it achieved stunning results in error correction using AMD/Xilinx FPGAs, which proved to be ten times faster than expected and necessary in intervening to correct errors. Just as in the construction of processors, different approaches are also emerging for error correction, and it will likely take time for a clear winner to emerge.