Pavithran Iyer, Université de Sherbrooke
Arbitrary precision quantum control of qubit systems appears to be unobtainable due to environmental influences that manifest themselves as errors in a quantum algorithm. Errors modelled by the probabilistic application of Pauli operators during the computation are convenient for analytical proofs and classical simulation but the level of accuracy of such a model depends on the quantumness of the error source. Many fault-tolerant schemes have been designed for the probabilistic Pauli error model and guarantee reliable computations when the error probability is below a threshold value. However, many quantum processes are poorly approximated by the Pauli error model and the notion of an error rate is less clear for such processes making fault tolerance proofs troublesome; an extreme example being coherent errors. We therefore need appropriate metrics for characterizing the strength of general quantum noise that integrates well into fault-tolerance proofs. There are many candidate metrics, for instance, experimentalists prefer to report the fidelity of a physical qubit while theorists on the other hand, fancy the Diamond norm. My talk will be motivated by the following question: “Which of these definitions can help us best predict the response of a fault-tolerant scheme to the underlying noise process?” To answer this, I will show some results of our numerical studies with the concatenated Steane code. Our findings suggest that popular measures of the noise strength can only give coarse information on the performance of an error correcting code. I will also outline some of the challenges in simulating quantum error correction with generic single-qubit noise models. Lastly, I will discuss applications of machine learning techniques to find new measures of the noise strength that offer a better predictive power on the error correction performance.