Researchers Map Noise on Up to 92 Qubits Using New Learning Method, Improving Quantum Computing Reliability
Researchers from IBM Quantum, the University of Chicago, and École Polytechnique Fédérale de Lausanne have successfully mapped noise characteristics on up to 92 qubits, with validation on up to 21 qubits, using a new learning method. This breakthrough addresses a fundamental challenge in quantum computing: the impact of noise on the reliability of quantum computations. The team demonstrated that previously considered limitations, such as 'unlearnable (gauge) degrees of freedom,' do not hinder accurate predictions of noisy quantum dynamics or error mitigation when using learnable parameters. Their gate set Pauli noise learning framework efficiently characterizes and mitigates noise across a complete set of operations, including state preparation, measurements, and both single- and multi-qubit gates. This method ensures that even without uniquely identifying the noise, accurate results are attainable, and a consistent set of parameters reduces sampling overhead. 
