Qubit-channel metrology with highly noisy initial states and additional undesirable noisy evolution

David Collins, Taylor Larrechea · 2024

Read Qubit-channel metrology with highly noisy initial states and additional undesirable noisy evolution by David Collins, Taylor Larrechea (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-12085.txt.

Introduction

David Collins and Taylor Larrechea ask whether teamwork between quantum bits still pays when everything starts out noisy. The task is metrology: estimating a hidden setting of a quantum channel as precisely as possible. One plan prepares many qubits in a linked state and sends one through the channel. The other uses a single qubit. Earlier work compared the two when the spare qubits sat idle and clean.

Real devices are not so kind: the spare qubits suffer unwanted noisy evolution. Extending that math, they derive plain formulas deciding which plan wins under a given noise type. For some noise the many-qubit plan collapses below the single-qubit one, but for other noise it keeps its edge. They give a trick that cancels certain noise on the waiting qubits. The paper serves experimenters who must choose a strategy for their hardware.

Catalog record

CREATORDavid Collins, Taylor Larrechea
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE107 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-12085

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Qubit-channel metrology with highly noisy initial states and additional undesirable noisy evolution David Collins, Taylor Larrechea Abstract We consider protocols for estimating the parameter in a single-parameter unital qubit channel, assuming that the available initial states are highly mixed with very low purity. We compare two protocols, each invoking the channel once, via their quantum Fisher informations. One uses n n qubits prepared in a particular correlated input state and subsequently queries the channel on one qubit. The other uses a single qubit. We extend the results of Collins [ 1 ] by allowing for additional noisy evolution on the spectator qubits in the n n -qubit protocol. We provide simple algebraic expressions that will determine when the n n -qubit protocol is superior. We provide a technique that can alleviate certain types of noise. We show that for certain types of noisy evolution the n n -qubit protocol will be inferior but for others it will be superior. I Introduction Quantum metrology considers using quantum systems to determine parameters that describe physical processes or states. Metrology takes the view that parameter determination must be described in terms of particular physical systems, specific operations and measurements on those systems and subsequent processing of measurement outcomes. All but the outcome processing are subject to the physical laws that govern the particular systems and these will constrain possible parameter determination procedures and partly dictate their effectiveness. Quantum metrology provides a framework for analyzing any scheme for determining a physical parameter. This combines classical statistics and quantum theory to quantify the effectiveness of schemes and to decide which may be preferable [ 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ] . This has frequently been applied to estimate parameters that govern particular qubit evolutions. In many cases the effectiveness of an estimation process using manifestly quantum resources such as entangled states can be greater than that attained by comparable “classical” processes [ 9 , 18 , 19 , 20 , 21 , 22 , 23 ] . Studies in quantum metrology initially considered absolutely optimal schemes, using pure initial quantum states. However, some systems, such as solution-state nuclear magnetic resonance (NMR), cannot access pure states and this raises the question of what advantages exist when only mixed states are available [ 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 ] . Results here were unified for a large class of qubit evolution processes, described by a single parameter to be estimated, on systems with highly mixed initial states [ 1 ] . A collection of n qubits, each in the same highly mixed state, are initially subjected to a parameter-independent preparatory unitary process and subsequently only one of the qubits is subjected to the parameter-dependent evolution process; the rest act as spectators. For very highly mixed states, this yields a roughly n -fold estimation accuracy gain versus a single qubit process [ 1 ] . In [ 1 ] , the spectator qubits do not evolve after the preparatory unitary. The present article addresses how the presence of additional, typically noisy, evolution on the spectators affects estimation. For example, consider estimating the dephasing time for a single nuclear spin within a multi-spin molecule [ 32 ] . The additional nuclear spins could be used as spectators to enhance the estimation accuracy according to the protocol of [ 1 ] . However, they would surely suffer their own dephasing while that of the spin of interest is being probed. How might this affect the enhancement promised in [ 1 ] ? Are there ways to reduce the effects of spectator dephasing? Noisy initial state metrology has also been investigated in optical systems [ 30 , 33 ] . This article covers all physical implementations. We will provide a method for quantifying the effects of a large class of spectator noise on the estimation accuracy previously obtained [ 1 ] . We describe when the additional spectator noise is catastrophic and when it can be tolerated. It will also provide a process of manipulating the spectator qubits so as to reduce the adverse effects of some types of spectator noise. Although the techniques offered here are similar those of [ 1 ] , the inclusion of additional spectator noise results in significant modifications to the existing results on noisy parameter estimation and approaches much more realistic scenarios than those analyzed previously [ 26 , 28 , 29 , 1 ] . This article is organized as follows. Section II describes the framework for qubit channel metrology. Section III provides the general description of qubit channels used in the later analysis. Sec. IV describes metrology with highly mixed initial states. Sections V and VI describe the two competing estimation protocols considered here and contain the main results that drive the rest of the article. Section VII describes a technique that can alleviate the detrimental effects of some noise. Section VIII describes the final measurements that various protocols could use. Section IX summarizes the estimation procedures and Sec. X applies this to various examples. II Qubit channel metrology Suppose that a channel of known type \hat{\Gamma}(\lambda) , but dependent on an unknown parameter, \lambda , acts on a single qubit. Metrology considers how to determine the parameter as accurately as possible via physical use of the channel. A protocol to determine \lambda specifies the number of qubits used, their initial states, possible supplementary evolutions, eventual measurements and processing of measurement outcomes. Figure 1 illustrates examples of typical protocols [ 34 , 9 , 22 , 12 , 26 , 13 , 35 , 36 , 37 , 38 , 39 , 40 , 41 ] . Figure 1: Possible metrology protocols. The protocol in a) uses \hat{\Gamma}(\lambda) independently on each of q qubits. The protocol in b) uses n qubits with \hat{\Gamma}(\lambda) acting on each of q channel qubits. There are additional spectator qubits, on which the channel does not act. All are subjected to a parameter-independent unitary operation, \hat{U}_{\mathrm{prep}} , prior to querying the channel and a measurement, possibly global, after channel queries. Quantum metrology assesses the accuracy of protocols in terms of the statistical fluctuations in parameter estimates and compares competing protocols, aiming for the most accurate. A well-established measure of accuracy is the quantum Cramér-Rao bound, which bounds the variance in the estimate [ 10 ] . Let m_{1},m_{2},\ldots , be the measurement outcomes of the protocol and let p(m_{1},m_{2}\ldots|\lambda) be the associated probability. An estimate for the parameter is generated via \lambda_{\mathrm{est}}=\lambda_{\mathrm{est}}(m_{1},m_{2},\ldots) , where \lambda_{\mathrm{est}}(\ldots) is an estimator function. The requirements are that the estimator is unbiased, i.e. \langle\lambda_{\mathrm{est}}\rangle=\lambda where \langle\lambda_{\mathrm{est}}\rangle=\sum_{m_{1},m_{2},\ldots}\lambda_{\mathrm{est}}(m_{1},m_{2},\ldots)p(m_{1},m_{2}\ldots|\lambda) and that the variance \variance{(\lambda_{\mathrm{est}})}:=\left<\left(\lambda_{\mathrm{est}}-\langle\lambda_{\mathrm{est}}\rangle\right)^{2}\right> is minimized. The variance depends on the choice of estimator but, regardless of this choice, is bounded by classical Cramér-Rao bound (CRB) [ 42 , 10 ] , \variance{(\lambda_{\mathrm{est}})}\geqslant 1/F(\lambda) where the Fisher information is F(\lambda)=\sum_{m_{1},m_{2},\ldots}\left[\frac{\partial\ln{p(m_{1},m_{2}\ldots|\lambda)}}{\partial\lambda}\right]^{2}p(m_{1},m_{2}\ldots|\lambda). The maximum likelihood estimator, determined by solving \frac{\partial p}{\partial\lambda}=0 for \lambda , saturates the bound [ 42 ] as n\rightarrow\infty. For measurements performed on quantum systems, the system state immediately

Read the complete text