Continuous-variable approximate unitary 2-design, with applications to unclonable encryption
Arpan Akash Ray and Boris Škorić · 2026
Read Continuous-variable approximate unitary 2-design, with applications to unclonable encryption by Arpan Akash Ray and Boris Škorić (2026) 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-2603-06393.txt.
Introduction
Ray and Skoric set out an epsilon approximate unitary 2-design for p and q quadratures in continuous variable systems, alternating q hat and p hat based unitaries. The design acts live on a finite discretisation of that space and operate on the full space, a first for such systems. Epsilon equals 1 over d to the power ell, with d the truncated Hilbert space dimension. An Unclonable Encryption scheme uses the design acts. Security follows from recent decoupling results, giving unclonable hidden security for such encryption.
Catalog record
| CREATOR | Arpan Akash Ray and Boris Škorić |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62262 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.06393
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Continuous-variable approximate unitary 2-design, with applications to unclonable encryption Arpan Akash Ray and Boris Škorić Abstract We introduce an ε \varepsilon -approximate unitary 2-design that is compatible with the structure of p- and q-quadratures in continuous-variable (CV) quantum systems. The design unitaries are defined on a finite-dimensional discretisation of the CV space and can be physically implemented as operations on the full CV space. This establishes the first approximate unitary design for CV systems. The design alternatingly acts with unitaries based on the quadrature operators q ^ \hat{q} and p ^ \hat{p} . We prove that the parameter ε \varepsilon is given by 1 / d ℓ 1/d^{\ell} , where d d is the dimension of the truncated Hilbert space and ℓ \ell is the number of iterations. We propose an Unclonable Encryption scheme in which the encryption operators are given by the unitaries which constitute the approximate unitary design. We prove its security using recent results on decoupling. This establishes unclonable-indistinguishable security for a CV encryption for the first time. 1 Introduction 1.1 Unitary two-designs Random unitaries drawn from the Haar measure on the unitary group play a central role in quantum information theory, because they provide a mathematically clean notion of “generic” dynamics and enable uniform averages that underlie decoupling, benchmarking, tomography, and cryptographic reductions. In practice, however, Haar-random unitaries are rarely available as a physical primitive, and even sampling them exactly is computationally expensive in high dimensions [ 28 ] . A standard remedy is to replace Haar randomness by unitary designs : structured ensembles of unitaries that reproduce Haar statistics up to some fixed polynomial degree. For many protocols, matching only the first few moments already suffices, so low-order designs provide a bridge between idealized Haar arguments and implementable randomization schemes [ 12 ] . One convenient way to define unitary designs is via the t -fold twirling channel {\cal G}^{(t)}_{\mu} induced by a probability measure \mu on \mathrm{U}(d) , the space of unitary operators in d dimensions, \mathcal{G}^{(t)}_{\mu}(X)=\int_{\mathrm{U}(d)}U^{\otimes t}\,X\,(U^{\dagger})^{\otimes t}\,{\rm d}\mu(U),\qquad X\in\mathcal{L}\!\bigl((\mathbb{C}^{d})^{\otimes t}\bigr). (1) Here {\cal L} stands for the space of linear operators. The measure \mu is called an (exact) unitary t -design if \mathcal{G}^{(t)}_{\mu} coincides with the Haar twirl \mathcal{G}^{(t)}_{\mathrm{Haar}} (equivalently, if all balanced polynomials of degree at most t in the matrix entries of U have the same average under \mu as under Haar) [ 12 , 9 ] . The case t=2 is especially prominent: it captures second-moment statistics and is therefore the minimal notion of pseudorandomness that already supports a wide range of ‘average-case’ tasks. It also underlies optimal strategies for state and process tomography and related estimation problems [ 10 , 5 , 12 ] . Unitary designs exist for finite dimension d . A natural question is whether analogous objects exist for Continuous-Variable (CV) quantum systems. Here one immediately encounters conceptual and technical obstacles: CV systems are modeled on separable infinite-dimensional Hilbert spaces (e.g. L^{2}(\mathbb{R}) for a single bosonic mode), and there is no direct counterpart of the finite-dimensional Haar measure that yields finite, well-behaved moment operators on the full unitary group. Motivated by physical implementability, a particularly important candidate class is the set of Gaussian states/unitaries, often viewed as the CV analogue of the Clifford group. However, Blume-Kohout and Turner [ 5 ] showed that Gaussian states cannot form a CV 2-design on L^{2}(\mathbb{R}) , and moreover that no true Gaussian ensemble on the entire infinite CV space is even “close” to a 2-design, because the relevant invariant averaging fails to produce a normalizable density operator. This already rules out the most accessible CV candidate family for reproducing second-moment Haar statistics. Another perspective on CV ‘design-like’ randomness was developed by Zhuang et al. [ 34 ] in the study of scrambling in phase space. Because phase space has infinite volume, they emphasize that even CV analogues of uniform ensembles (e.g. displacements, which play the role of Pauli operators) are not normalizable, so some form of regularization is unavoidable. They proposed several methods to analyse higher-moment pseudorandomness on random Gaussian circuits, finding that large squeezing can produce signatures reminiscent of higher-design behavior while still falling short of an exact CV 2-design. This shows the need for explicit energy or discretisation constraints in any CV notion of (approximate) unitary 2-design, and connects with later general non-existence results for exact CV designs of order t\geq 2 which we describe below. Finally, Iosue et al. [ 21 ] recently published a systematic study of designs in infinite-dimensional settings and proved a much stronger no-go theorem: under the natural CV extensions of finite-dimensional design definitions (including those used in [ 5 ] ), no CV state t -designs exist for any t\geq 2 , and likewise no CV unitary t -designs exist for any t\geq 2 on separable infinite-dimensional Hilbert spaces. They introduced an alternative framework for state designs based on rigged Hilbert spaces (“rigged t -designs”) and gave explicit constructions for t=2 , together with regularization procedures that turn these into approximate, physically normalizable ensembles under energy cutoffs. For unitary designs, they also discuss regularized notions of approximation, but explicit constructions beyond t=1 remain an open problem. Given these no-go results, it is natural to relax the exact equalities and consider approximate designs. In finite dimension, an \varepsilon -approximate unitary 2-design can be defined by requiring the induced 2-fold twirl to be \varepsilon -close to the Haar twirl in a suitable norm, with several equivalent or comparable formulations in terms of moment operators, induced Schatten norms, or frame potentials [ 12 , 18 ] . Approximate 2-designs are abundant in discrete-variable systems and admit efficient constructions; in fact, much of the usefulness of designs in applications such as fidelity estimation persists under quantitatively controlled approximation [ 12 ] . A conceptually simple discrete-variable construction relevant for the current work is due to Nakata et al. [ 30 ] , who showed that alternating random unitaries diagonal in the Pauli- Z basis with random unitaries diagonal in the Pauli- X basis yield efficient \varepsilon -approximate unitary 2-designs, with the approximation error decaying as \varepsilon=\Theta(d^{-\ell}) after \ell repetitions. In contrast to the maturity of the finite-dimensional case, there is currently no explicit construction of an \varepsilon -approximate CV unitary 2-design. 1.2 Decoupling and Unclonable Encryption Unclonable Encryption (UE) is a form of encryption of classical messages in which the ciphertext is quantum and cannot be split into two parts that can both be decrypted successfully once the decryption key is revealed. Broadbent and Lord [ 8 ] formalized two security notions: unclonable security and unclonable indistinguishability . Security is defined via a cloning-style experiment: an adversary A receives a challenge ciphertext, produces two quantum registers intended for two non-communicating parties B and C, and after the key is revealed, both parties attempt to recover the correct plaintext. Unclonable security demands that the probability that both parties output the correct message is small for any attack. 1 1 1 An earlier work by Gottesman [ 16 ] introduced an ‘unclonable encryption’ notion somewhat similar to unclonable security. Unclonable indistinguishability strengthens this notion by having only two possible plaintexts and requiring that it is infeasible for both B and C to correctly identify the plaintext