Kostant relation in filtered randomized benchmarking for passive bosonic devices

David Amaro-Alcalá · 2025

Read Kostant relation in filtered randomized benchmarking for passive bosonic devices by David Amaro-Alcalá (2025) 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-2511-00842.txt.

Introduction

David Amaro-Alcala works on cheaper ways to benchmark passive bosonic devices, hardware such as optical systems whose behavior is tested by randomized benchmarking. The existing proposal for such devices is costly, partly because it relies on Clebsch-Gordan coefficients from group theory.

He introduces two filter functions that use the same measurement data but avoid those coefficients. One filter is built from immanants, the other from characters of the special unitary group, and both yield simple formulas for the variance of the estimate. The character filter is cheap to compute and keeps a constant, low variance. A test case with photon loss and gain suggests estimates stay close to the ideal values, which could simplify device characterization.

Catalog record

CREATORDavid Amaro-Alcalá
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-00842

Browse the full library · About the Tower of Babel

Full text

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

Kostant relation in filtered randomized benchmarking for passive bosonic devices Preprint: APS/123-QED David Amaro-Alcalá Abstract We reduce the cost of the current bosonic randomized benchmarking proposal. First, we introduce a filter function using immanants. With this filter, we avoid the need to compute Clebsch–Gordan coefficients. Our filter uses the same data as the original, although we propose a distinct data collection process that requires a single type of measurement. Furthermore, we argue that weak coherent states and intensity measurements are sufficient to proceed with the characterization. Our work could then allow simpler platforms to be characterized and simplify the data analysis process. I Introduction Characterizing passive bosonic devices is an important step in the development of a continuous-variable quantum computer Masada et al. 2015 ; Takeda and Furusawa 2017 ; Yonezu et al. 2023 ; Fukui and Takeda 2022 . A recent extension of the randomised benchmarking scheme Emerson et al. 2005 ; Magesan et al. 2011 ; Knill et al. 2008 ; Amaro-Alcalá et al. 2024 ; Jafarzadeh et al. 2020 ; Helsen et al. 2022 , one of the most successful methods for characterising finite-dimensional quantum gates, extends the framework to bosonic passive devices Arienzo et al. 2025 ; Wilkens et al. 2024 . This scheme estimates a fidelity-like figure of merit of the noise associated with such devices. It inherits many of the desirable features of standard randomised benchmarking, including robustness to state preparation and measurement (SPAM) errors and a well-developed theoretical foundation. However, the original proposal suffers from two important drawbacks. First, it requires the evaluation of matrix permanents, which are computationally hard to calculate Valiant 1979 ; Bürgisser 2000 . Moreover, the required permanents must be determined on a case-by-case basis due to their dependence on complicated decompositions involving Clebsch–Gordan coefficients. Second, the experimental design is challenging for most laboratories, as it requires the preparation of Fock states and the use of photon-number-resolving detectors. In this work, we build upon the existing randomized benchmarking framework for bosonic devices Arienzo et al. 2025 by addressing both practical and computational limitations. We propose an improved protocol that maintains the simplicity of the original method. Our method eliminates the need for photon-number-resolving or homodyne detection and removes the dependence on permanents and Clebsch–Gordan coefficients in the data analysis. Moreover, the filter expression can be known beforehand (it does not depend on the initial state or measurement used), greatly simplifying the characterization scheme. Our approach yields a more experimentally accessible characterization scheme that significantly reduces both computational and experimental costs. We show that experimental data can be collected using weak coherent states and intensity measurements, thereby greatly expanding the practical applicability of the protocol. Furthermore, we simplify the data analysis by leveraging a relation due to the late Bertram Kostant Kostant 1995 , which allows us to express the filter function in terms of a smaller set of less computationally expensive immanants, thereby avoiding the computation of Clebsch-Gordan coefficients. This paper is organised as follows. In Sec. II , we introduce the tools and the problem we aim to solve, and in particular, we recall the relation between immanants and irreducible representations (irreps) of the unitary group. In Sec. III , we introduce our new filter function and demonstrate that it yields a single exponential decay of the parameters making up the figure of merit. We also present an alternative to the original data-gathering process, which is simpler and requires less complex experimental arrangements. In Sec. IV , we describe the new filtering process in detail. Finally, in the Discussion and conclusion section (Sec. V )—we summarize our scheme in relation to the original and highlight its advantages. II Background II.1 States and channels In this subsection, our goal is to introduce the representation of unitary operators as interferometers. We start by introducing the Hilbert space and the states. We then proceed to discuss the action of unitary gates on the states, which allows us to recall the relevant unitary representations. Throughout this work, we consider the Hilbert space of n indistinguishable photons with m modes, which we denote by \mathcal{H}^{n}_{m} . The states living in \mathcal{H}^{n}_{m} are labeled by the occupation numbers \mathbf{n}=(n_{0},\ldots,n_{m-1}) ; n_{i} denotes how many photons are in the i -th mode. The corresponding state is written as \ket{\mathbf{n}}\coloneqq\left|n_{1},\ldots,n_{m}\right\rangle\coloneqq\left(\prod_{k=1}^{m}\frac{1}{\sqrt{n_{k}!}}(a_{k}^{\dagger})^{n_{k}}\right)\ket{\mathbf{0}}, (1) where a_{k}^{\dagger} is the creation operator in the k -th mode. We denote by SU( m ) the group of m\times m unitary matrices with unit determinant. The action of a unitary matrix U on a creation operator is \mathcal{U}(a^{\dagger}_{i})\coloneqq\sum_{j}U_{ji}a^{\dagger}_{j}. (2) Therefore, the action on state \ket{\mathbf{n}} is \mathcal{U}(\ket{\mathbf{n}})=\left(\prod_{k=1}^{m}\frac{1}{\sqrt{n_{k}!}}\left(\mathcal{U}(a_{k})^{\dagger}\right)^{n_{k}}\right)\ket{\mathbf{0}} (3) with the product running over all the input ports. Building on the action on \ket{\mathbf{n}} , we define the action of a unitary on states \ket{\mathbf{n}}\bra{\mathbf{m}} \mathcal{U}(\ket{\mathbf{n}})\otimes\mathcal{U}(\bra{\mathbf{m}}), (4) which we denote by \Gamma\colon U\mapsto\mathcal{U}\otimes\overline{\mathcal{U}}, (5) with \overline{\mathcal{U}} denoting the complex conjugate of \mathcal{U} . The main ingredient of the characterization is the gates. For an arbitrary linear map on \mathcal{H}^{n}_{m} , we consider the set of completely positive trace-preserving (CPTP) mappings, and we denote its elements by \mathcal{E} . In this article, we use the terms “noise” and “operator” interchangeably to refer to a channel. Thus, we use \mathcal{E} to represent both. For clarity and simplicity, we assume a gate- and time-independent, Markovian noise model of the form \tilde{\mathcal{U}}=\mathcal{E}\circ\mathcal{U} (6) and \tilde{\varrho}=\mathcal{E}(\varrho) . The tilde notation indicates objects affected by noise, thus \tilde{\varrho} denotes \mathcal{E}^{\prime}(\varrho) for some channel \mathcal{E}^{\prime} not necessarily equal to \mathcal{E} . Lastly, in this subsection, we recall that the vectorization of a state \varrho and measurement E is represented by the vectors \ket{\varrho} and \ket{E} that satisfy \operatorname{tr}[E\mathcal{E}(\varrho)]=\bra{E}\Gamma(\mathcal{E})\ket{\varrho}. (7) Both sides in Eq. ( 7 ) correspond to the probability to measure E after the channel \mathcal{E} is applied to \varrho . In finite dimensional systems Lin et al. 2019 , the vectorization operation compatible with the transformation \Gamma is the tensor product \operatorname{vec}(\varrho)\otimes\overline{\operatorname{vec}(\varrho)} , where \operatorname{vec} corresponds to stacking the rows of the matrix representation of \varrho into a single vector. II.2 Representation theory detour Irreps of the special unitary group play a fundamental role in both bosonic randomised benchmarking Arienzo et al. 2025 and in this work. In this subsection, we introduce standard language in representation theory, as well as the relevant representation we use. We also describe the decomposition of such a representation. Ultimately, we introduce the fidelity-like figure of merit that we use for characterization. We now discuss the decomposition of Eq. ( 5 ) into irreducible representations. The irreps of the special unitary group are labeled by integer partitions \lambda=(\lambda_{0},\ldots,\lambda_{m-1}) of the number of photons n .For a more extensive discussion, refer to the appendices in Ref. Arienzo et al. 2025 . The action of \mathcal{U} is discussed in Eq. ( 3 ); for states with m photons it corresponds to the action of the irreducible representation (irrep) of the special unitary group SU( m ), labeled by the partition \lambda\coloneqq(n,\underbrace{0,\ldots,0}_{m-1\text{ times}}), (8) with \lambda^{*} the dual irrep of \lambda ; an example is provided in Appendix B . Thus, using Pieri’s formula Fulton 1996 , the representation \Gamma introduced in Eq. ( 5 ) is reducible and decomposes into a finite list of representations (with

Read the complete text