Approximate pushforward designs and image bounds on approximations
Jakub Czartowski, Adam Sawicki, Karol Życzkowski · 2025
Read Approximate pushforward designs and image bounds on approximations by Jakub Czartowski, Adam Sawicki, Karol Życzkowski (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-2512-01858.txt.
Introduction
Jakub Czartowski, Adam Sawicki, and Karol Zyczkowski extend quantum pushforward designs to approximate forms. Pushforward designs average quantum states or channels in a structured way. Exact averaging is often out of reach. Approximate forms settle for finite precision.
Using Schatten norms and Lipschitz arguments, they bound the approximation traits. Builds come from complex projective spaces, simplices, mixed states, and channels. Bounds tell users how close an easy build sits to the ideal. That helps in tomography and benchmarking of quantum devices. Projective builds are explicit, so a user can construct the design rather than merely know that one exists somewhere in theory.
Catalog record
| CREATOR | Jakub Czartowski, Adam Sawicki, Karol Życzkowski |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-01858
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Approximate pushforward designs and image bounds on approximations Jakub Czartowski, Adam Sawicki, Karol Życzkowski Abstract We extend the framework of quantum pushforward designs to the approximate setting, where averaging is achieved only up to finite precision. Using Schatten p p -norms and Lipschitz continuity arguments, we derive bounds on the approximation parameters of pushforward designs obtained from complex projective spaces, including simplices, mixed states, and quantum channels. In the mixed-state case, we refine the bounds by exploiting the symmetric subspace structure, leading to asymptotically tighter estimates. Numerical simulations support our theoretical results, showing near-optimality in low-dimensional scenarios. 1 Introduction Designs, also known as averaging sets or quadratures, provide powerful methods for evaluating polynomial functions over compact spaces with respect to a given measure. They have found diverse applications, from Gauss quadratures in numerical integration [ 10 ] to spherical designs [ 9 , 29 ] , which are used in fields such as electrostatics [ 30 ] . More recently, complex projective designs and unitary designs have been applied to quantum state tomography and quantum channel tomography, respectively [ 27 , 28 ] . In recent years, these concepts have been extended to continuous-variable systems [ 18 ] , noncompact spaces [ 22 ] , and connections between different spaces have been realized. Notably, projective toric designs combined with simplex designs have been used to generate complex projective designs of product form [ 19 ] . Furthermore, recent work has shown that mapping designs from one space to another provides new methods for constructing designs, leading to the introduction of mixed-state designs and channel designs, collectively referred to as pushforward designs [ 5 , 6 ] . In parallel, the concept of approximate designs has gained traction, with notable examples such as approximate complex projective designs, which are used in derandomization of random measurements [ 1 ] , and approximate unitary designs, which demonstrate the generic nature of quantum speedups [ 12 ] . A lot of attention has been put recently to understanding how good approximate t -designs can be obtained from random quantum circuits [ 3 , 20 , 21 ] . Moreover, recent studies have shown that a quantum gate set must provide at least an approximate 4-design to be considered universal [ 25 , 26 ] . In this work, we address a more general question: Given a \delta -approximate t -design in an original space X , what is the approximation parameter \delta^{\prime} for a \delta^{\prime} -approximate pushforward design in the image space Y ? 2 Preliminaries Before moving to the proper content let us set the notation for the remainder of the text. \ket{\psi}\in\mathbb{C}P^{d-1}\equiv\mathcal{H}_{d} will denote elements of complex projective space of dimension d-1 , representing pure quantum states in Hilbert space of dimension d subject to normalisation \absolutevalue{\ip{\psi}}^{2}=1 and equivalence \ket{\psi}\sim e^{i\phi}\ket{\psi} for \phi\in\mathbb{R} ; capital letter subscript, like \mathcal{H}_{A} , will denote the corresponding subspace of dimension d_{A} . The set of density operators \mathcal{B}(\mathcal{H})\equiv\quantity{\rho\in\mathcal{L}( \mathcal{H}):\Tr(\rho) = 1,\,\rho\geq 0,\,\rho= \rho^\dagger} is defined as the subset of linear operators \mathcal{L}(\mathcal{H}) over the Hilbert space positive semidefinite operators, representing mixed quantum states. \mathcal{U}(d) is the unitary group of dimension d acting on the Hilbert space \mathcal{H}_{d} . Finally, \mathcal{C}(\mathcal{H},\mathcal{H}^{\prime})\equiv\quantity{\Phi:\mathcal{L}(\mathcal{H})\ni Q\mapsto\Phi(Q)\in\mathcal{L}(\mathcal{H}')} denotes the set of quantum channels, which are additionally Completely Positive and Trace Preserving (CPTP), thus ensuring \mathcal{B}(\mathcal{H})\ni\rho\mapsto\Phi(\rho)\in\mathcal{B}(\mathcal{H}^{\prime}) . In particular, any quantum channel admits Kraus representation, \Phi(\rho)=\sum_{i=1}^{d_{E}}K_{i}\rho K_{i}^{\dagger}\quad\mbox{with}\quad\sum_{i}K_{i}^{\dagger}K_{i}=\mathbb{I} where d_{E}\leq\operatorname{dim(\mathcal{H})^{2}} is the number of Kraus operators K_{i} . Additionally, \operatorname{Sym}_{t}(X) will denote the subspace of X^{\otimes t} symmetric with respect to permutation of the elements of the tensor product. \Pi_{X,t} will represent the canonical projection from X^{\otimes t} to \operatorname{Sym}_{t}(X) . We begin our considerations by recalling the definition of an exact t -design in the context of approximating averages of polynomial functions with respect to a given measure \mu_{X} on a given space X\subset\mathbb{C}^{d} using another measure \nu_{X} . In what follows we will consider all measures to be normalized such that \int\differential{\mu}=1 . Definition 1 (Exact t -design) . Let us consider a space X and a natural measure \mu_{X} . We call a (continuous or discrete) measure \nu_{X} a t -design if averages of a balanced polynomial g_{t} of order t in both components of x and their conjugates, with respect to both measures are equal, \expectationvalue{g_t}_{\mu_{X}}\equiv\int_{X}g_{t}(x)\differential{\mu_X}=\int_{X}g_{t}(x)\differential{\nu_X}\equiv\expectationvalue{g_t}_{\nu_{X}}. (1) for any polynomial g_{t} . For a discrete measure \nu_{X} , we define S_{\nu_{X}}:=\quantity{x\in X\mid\nu_X(x)\neq 0} and call it the support of \nu_{X} . For x\in\nu_{X} the value w_{x}=\nu_{X}(x) will be called the weight of x . When w_{x}=w_{y} for all x,y\in S_{\nu_{X}} we say that \nu_{X} is uniform discrete measure. This and similar definitions have appeared in the literature, in the context of different spaces under different names, including quadratures in the context of integration using finite sets of points [ 10 ] , spherical designs [ 9 ] , averaging sets [ 29 ] , complex projective designs [ 23 , 17 , 14 , 2 , 15 , 16 ] and unitary designs [ 7 ] . The notion of t -designs can be encapsulated more easily by introducing moment operator T_{\mu_{X},t}, with respect to measure \mu T_{\mu_{X},t}=\int_{X}(x\otimes x^{*})^{\otimes t}\differential{\mu}, (2) where x^{*} is complex conjugate of the object under consideration. To obtain an average of a specific polynomial g_{t} , one has to take the corresponding coefficient matrix G and calculate its scalar product with the moment operator, \expectationvalue{g_t}_{\mu}=\innerproduct{G_t}{T_{\mu_X,t}}, (3) where the scalar product depends on the space under consideration. As we will be dealing mostly with matrices, the expression can be actually reduced to \Tr(G_tT_{\mu_X,t}) . Additionally, the balanced property of the polynomial g_{t} is encoded in the coefficient matrix G_{t} being hermitian. This allows us to reformulate the definition of the exact t -designs in terms of the moment operators. Definition 2 (Exact t -design in terms of moment operator) . Let us consider a space X and a natural measure \mu_{X} . We call a (continuous or discrete) measure \nu_{X} a t -design if the respective moment operators are equal, T_{\mu_{X},t}=T_{\nu_{X},t}. (4) In order to extend the above definition to approximate designs and evaluate the approximation precision \delta_{p} , we will use Schatten p -norms, with definition restricted to finite-dimensional Hilbert spaces for the sake of clarity. Definition 3 . Consider a linear operator X\in\mathcal{L}\quantity(\mathcal{H}^1,\mathcal{H}^2) , with \mathcal{H}^{1},\,\mathcal{H}^{2} being finite-dimensional Hilbert spaces. Schatten p -norm is defined as \norm{X}_{p}\equiv\quantity(\sum_{i=1}^{\min(d_1,d_2)}\qty[s_i(X)]^p)^{1/p} (5) where s_{i}(X)\geq 0 denotes i -th singular value of the operator X . Note that Schatten p -norms for p=1,2,\infty are given by trace, Frobenius and operator norms, respectively, and have been already used in the context of approximate designs. In particular, trace and operator norms have direct operational interpretation for state-discrimination tasks [ 13 ] and worst-case scenario considerations [ 4 ] With this, a definition of approximate designs is straightforward consequence. Definition 4 ( \delta_{p} -approximate t -design) . Let us consider a space X and a natural measure \mu_{X} . We call a (continuous or discrete) measure \nu_{X} a \delta_{p} -approximate t -design if the Schatten p -norm of the difference of the respective moment operators is, \norm{T_{\mu_X,t} - T_{\nu_X,t}}_{p}=\delta_{p}. (6) The main subject of this study is concerned with image spaces