Lennart Bittel,Francesco Anna Mele,Antonio Anna Mele,Salvatore Tirone,Ludovico Lami
Abstract:Gaussian states of bosonic quantum systems enjoy numerous technological applications and are ubiquitous in nature. Their significance lies in their simplicity, which in turn rests on the fact that they are uniquely determined by two experimentally accessible quantities, their first and second moments. But what if these moments are only known approximately, as is inevitable in any realistic experiment? What is the resulting error on the Gaussian state itself, as measured by the most operationally meaningful metric for distinguishing quantum states, namely, the trace distance? In this work, we fully resolve this question by demonstrating that if the first and second moments are known up to an error $\varepsilon$, the trace distance error on the state also scales as $\varepsilon$, and this functional dependence is optimal. To prove this, we establish tight bounds on the trace distance between two Gaussian states in terms of the norm distance of their first and second moments. As an application, we improve existing bounds on the sample complexity of tomography of Gaussian states.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: when there are errors in the first moment and covariance matrix of a Gaussian state, how to estimate the resulting error of the Gaussian state itself, especially the specific manifestation of this error under the trace distance metric. Specifically, if the estimation errors of the first moment and covariance matrix are \(\varepsilon\), how does this error affect the overall estimation error of the Gaussian state? The goal of the paper is to determine the specific dependence of this error on the trace distance and provide strict upper and lower bounds.
### Problem Background
Gaussian states have important applications in continuous - variable quantum systems because they are not only ubiquitous in nature but also relatively simple mathematically and can be uniquely determined by the experimentally measurable first moment and covariance matrix. However, in actual experiments, there are always certain errors in the measurement of these quantities. Therefore, understanding how these errors propagate to the overall error of the Gaussian state is crucial for the accurate estimation of quantum states.
### Specific Problem
The following problems are defined in the paper:
**Problem 1 (Informal Version)**
Let \(\rho(V, m)\) be a Gaussian state with the first moment \(m\) and covariance matrix \(V\). Suppose \(\tilde{V}\) and \(\tilde{m}\) are \(\varepsilon\)-approximations of \(V\) and \(m\) respectively:
\[
\tilde{m}\approx_{\varepsilon}m,\quad\tilde{V}\approx_{\varepsilon}V.
\]
Then \(\rho(\tilde{V},\tilde{m})\) is an approximation of \(\rho(V, m)\):
\[
\rho(\tilde{V},\tilde{m})\approx\rho(V, m).
\]
The problem is how to quantify the error in this approximation, that is, the trace - distance error between \(\rho(\tilde{V},\tilde{m})\) and \(\rho(V, m)\), expressed in terms of \(\varepsilon\).
### Trace Distance
The trace distance is the most commonly used operationally - well - defined distance measure between quantum states, established by the famous Holevo - Helstrom theorem. The trace distance is defined as:
\[
\frac{1}{2}\|\rho-\sigma\|_1,
\]
where \(\|\Theta\|_1 = \text{Tr}\sqrt{\Theta^{\dagger}\Theta}\) represents the trace norm of the linear operator \(\Theta\).
### Formal Problem
**Problem 2 (Formal Version)**
Let \(\rho(V, m)\) be a Gaussian state with the first moment \(m\) and covariance matrix \(V\). Similarly, let \(\rho(\tilde{V},\tilde{m})\) be a Gaussian state with the first moment \(\tilde{m}\) and covariance matrix \(\tilde{V}\). The problem is to determine strict upper and lower bounds for the trace distance between \(\rho(V, m)\) and \(\rho(\tilde{V},\tilde{m})\):
\[
\frac{1}{2}\|\rho(V, m)-\rho(\tilde{V},\tilde{m})\|_1,
\]
in terms of the norm distance \(\|m - \tilde{m}\|\) of their first moments and the norm distance \(\|V-\tilde{V}\|\) of their covariance matrices.
### Main Results
The main result of the paper is to provide strict upper and lower bounds for the trace distance between Gaussian states. These bounds show that if the estimation accuracy of the first moment and covariance matrix is \(\varepsilon\), then the trace - distance error also grows linearly with \(\varepsilon\). Specifically, the main theorem in the paper is as follows:
**Theorem 3 (Strict Upper Bound for the Trace Distance between Gaussian States)**