Accuracy of dynamical-decoupling-based spectroscopy of Gaussian noise

TL;DR

This paper examines the accuracy of dynamical decoupling-based noise spectroscopy for Gaussian and Lorentzian noise spectra, identifying conditions for precise spectral reconstruction and proposing improvements to avoid artifacts.

Contribution

It provides a detailed analysis of the conditions under which dynamical decoupling spectroscopy accurately reconstructs environmental noise spectra, especially for finite-range Gaussian spectra.

Findings

Standard formulas are accurate for Lorentzian spectra with negligible corrections.

Finite-range Gaussian spectra pose challenges, leading to potential misinterpretation of long-tail behavior.

An extended reconstruction method can eliminate artifacts and improve accuracy.

Abstract

The fundamental assumption of dynamical decoupling based noise spectroscopy is that the coherence decay rate of qubit (or qubits) driven with a sequence of many pulses, is well approximated by the environmental noise spectrum spanned on frequency comb defined by the sequence. Here we investigate the precise conditions under which this commonly used spectroscopic approach is quantitatively correct. To this end we focus on two representative examples of spectral densities: the long-tailed Lorentzian, and finite-ranged Gaussian---both expected to be encountered when using the qubit for nano-scale nuclear resonance imaging. We have found that, in contrast to Lorentz spectrum, for which the corrections to the standard spectroscopic formulas can easily be made negligible, the spectra with finite range are more challenging to reconstruct accurately. For Gaussian line-shape of environmental spectral density, direct application of the standard dynamical decoupling based spectroscopy leads to erroneous attribution of long-tail behavior to the reconstructed spectrum. Fortunately, artifacts such as this, can be completely avoided with the simple extension to standard reconstruction method.