JUPITER supercomputer breaks world record with 50-qubit quantum simulation
Summary
Researchers at the Jülich Supercomputing Centre and NVIDIA have successfully simulated a universal quantum computer with 50 qubits, setting a new world record. This achievement, surpassing their previous 48-qubit record from 2019, was made possible by JUPITER, Europe's first exascale supercomputer, launched in September. Quantum simulations are crucial for testing algorithms like Variational Quantum Eigensolver (VQE) and Quantum Approximate Optimisation Algorithm (QAOA), and for validating experimental findings before real quantum hardware can handle such tasks. Simulating 50 qubits demands approximately 2 petabytes of memory and extensive computing power, which JUPITER's NVIDIA GH200 Superchips provided. The Jülich Universal Quantum Computer Simulator (JUQCS) software was upgraded to JUQCS-50, incorporating byte-encoding compression and dynamic optimization across 16,000+ GH200 Superchips.
Key takeaway
For AI Scientists and Research Scientists developing quantum algorithms, this 50-qubit simulation milestone on JUPITER demonstrates the current frontier of quantum emulation. You should consider utilizing JUNIQ to access JUQCS-50 for testing and validating your quantum algorithms, as it offers capabilities beyond existing quantum processors and serves as a benchmark for future supercomputers.
Key insights
JUPITER supercomputer enabled a record 50-qubit quantum simulation, advancing quantum algorithm development.
Principles
- Quantum simulation complexity doubles with each qubit.
- High-performance computing is critical for quantum research.
Method
The JUQCS-50 software, optimized for NVIDIA GH200 Superchips, uses byte-encoding compression and dynamic data exchange to simulate 50 qubits by efficiently managing memory across CPU and GPU.
In practice
- Use JUQCS-50 for high-fidelity quantum computer emulation.
- Access JUQCS-50 via JUNIQ for external research.
Topics
- JUPITER Supercomputer
- Quantum Simulation
- 50-qubit Quantum Computing
- NVIDIA GH200 Superchips
- Exascale Computing
Best for: AI Scientist, Research Scientist, AI Hardware Engineer
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Robotics Research News -- ScienceDaily.