Abstract
Quantum cloning is a fundamental protocol of quantum information theory. Perfect universal quantum cloning is prohibited by the laws of quantum mechanics, only imperfect copies being reachable. Symmetric quantum cloning is concerned with case when the quality of the clones is identical. In this work, we study the general case of \(1 \rightarrow N\) asymmetric cloning, where one asks for arbitrary qualities of the clones. We characterize, for all Hilbert space dimensions and number of clones, the set of all possible clone qualities. This set is realized as the nonnegative part of the unit ball of a newly introduced norm, which we call the \(\mathcal Q\)-norm. We also provide a closed-form expression for the quantum cloner achieving a given clone quality vector. Our analysis relies on the Schur–Weyl duality and on the study of the spectral properties of partially transposed permutation operators.
Similar content being viewed by others
Data Availability
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
Notes
In \(\mathcal {V}^\lambda \), the vectors \(\left| \textbf{v}^{\lambda }_j\right\rangle \) are defined on different tensor products, depending on the value of i in their corresponding \(\left| \Omega \right\rangle _{(0,i)}\).
References
Bennett, C.H., Brassard, G.: Quantum cryptography: Public key distribution and coin tossing. arXiv preprint arXiv:2003.06557 (2020)
Bridgeman, J.C., Chubb, C.T.: Hand-waving and interpretive dance: an introductory course on tensor networks. J. Phys. A: Math. Theor. 50(22), 223001 (2017)
Bruß, D., Cinchetti, M., D’Ariano, G.M., Macchiavello, C.: Phase-covariant quantum cloning. Phys. Rev. A 62(1), 012302 (2000)
Barnum, H., Caves, C.M., Fuchs, C.A., Jozsa, R., Schumacher, B.: Noncommuting mixed states cannot be broadcast. Phys. Rev. Lett. 76(15), 2818 (1996)
Bruß, D., DiVincenzo, D.P., Ekert, A., Fuchs, C.A., Macchiavello, C., Smolin, J.A.: Optimal universal and state-dependent quantum cloning. Phys. Rev. A 57(4), 2368 (1998)
Bruss, D., D’Ariano, G.M., Macchiavello, C., Sacchi, M.F.: Approximate quantum cloning and the impossibility of superluminal information transfer. Phys. Rev. A 62(6), 062302 (2000)
Bužek, V., Hillery, M.: Quantum copying: Beyond the no-cloning theorem. Phys. Rev. A 54(3), 1844 (1996)
BuŽek, V., Hillery, M.: Universal optimal cloning of qubits and quantum registers. In: NASA International Conference on Quantum Computing and Quantum Communications, pp. 235–246. Springer, (1998)
Bhatia, R., Kittaneh, F.: Norm inequalities for partitioned operators and an application. Math. Ann. 287, 719–726 (1990)
Coyle, B., Doosti, M., Kashefi, E., Kumar, N.: Variational quantum cloning: improving practicality for quantum cryptanalysis. arXiv preprint arXiv:2012.11424 (2020)
Cerf, N.J.: Asymmetric quantum cloning in any dimension. J. Mod. Opt. 47(2–3), 187–209 (2000)
Ćwikliński, P., Horodecki, M., Studziński, M.: Region of fidelities for a 1 \(\rightarrow \) n universal qubit quantum cloner. Phys. Lett. A 376(32), 2178–2187 (2012)
Coecke, B., Kissinger, A.: Picturing quantum processes. Cambridge University Press, Cambridge (2017)
Coecke, B., Kissinger, A.: Picturing quantum processes. In: International Conference on Theory and Application of Diagrams, pp. 28–31. Springer (2018)
Christandl, M., Leditzky, F., Majenz, C., Smith, G., Speelman, F., Walter, M.: Asymptotic performance of port-based teleportation. Commun. Math. Phys. 381, 379–451 (2021)
Du, J., Durt, T., Zou, P., Li, H., Kwek, L.C., Lai, C.H., Oh, C.H., Ekert, A.: Experimental quantum cloning with prior partial information. Phys. Rev. Lett. 94(4), 040505 (2005)
Duan, L.-M., Guo, G.-C.: Probabilistic cloning and identification of linearly independent quantum states. Phys. Rev. Lett. 80(22), 4999 (1998)
Dieks, D.G.B.J.: Communication by epr devices. Phys. Lett. A 92(6), 271–272 (1982)
Duncan, R., Kissinger, A., Perdrix, S., Van De Wetering, J.: Graph-theoretic simplification of quantum circuits with the zx-calculus. Quantum 4, 279 (2020)
Fiurasek, J., Filip, R., Cerf, N.J.: Highly asymmetric quantum cloning in arbitrary dimension. arXiv preprint arXiv:quant-ph/0505212 (2005)
Fulton, W., Harris, J.: Representation Theory: A First Course, vol. 129. Springer, Berlin (2013)
Fan, H., Wang, Y.-N., Jing, L., Yue, J.-D., Shi, H.-D., Zhang, Y.-L., Liang-Zhu, M.: Quantum cloning machines and the applications. Phys. Rep. 544(3), 241–322 (2014)
Gisin, N., Massar, S.: Optimal quantum cloning machines. Phys. Rev. Lett. 79(11), 2153 (1997)
Grinko, D., Ozols, M.: Linear programming with unitary-equivariant constraints. arXiv preprint arXiv:2207.05713 (2022)
Hashagen, A.K.: Universal Asymmetric Quantum Cloning Revisited. arXiv preprint arXiv:1607.03723 (2016)
Kay, A.: Optimal Universal Quantum Cloning: Asymmetries and Fidelity Measures. arXiv preprint arXiv:1407.4951 (2014)
Kay, A., Kaszlikowski, D., Ramanathan, R.: Optimal cloning and singlet monogamy. Phys. Rev. Lett. 103(5), 050501 (2009)
Kay, A., Ramanathan, R., Kaszlikowski, D.: Optimal asymmetric quantum cloning. arXiv preprint arXiv:1208.5574 (2012)
Keyl, M., Werner, R.F.: Optimal cloning of pure states, testing single clones. J. Math. Phys. 40(7), 3283–3299 (1999)
Leditzky, F.: Optimality of the pretty good measurement for port-based teleportation. Lett. Math. Phys. 112(5), 98 (2022)
Lostaglio, M., Senno, G.: Contextual advantage for state-dependent cloning. Quantum 4, 258 (2020)
Mozrzymas, M., Horodecki, M., Studziński, M.: Structure and properties of the algebra of partially transposed permutation operators. J. Math. Phys. 55(3), 032202 (2014)
Mozrzymas, M., Studziński, M., Horodecki, M.: A simplified formalism of the algebra of partially transposed permutation operators with applications. J. Phys. A: Math. Theor. 51(12), 125202 (2018)
Mozrzymas, M., Studziński, M., Strelchuk, S., Horodecki, M.: Optimal port-based teleportation. New J. Phys. 20(5), 053006 (2018)
Nielsen, M.A., Chuang, I.L.: Quantum Computation and Quantum Information. Cambridge University Press, Cambridge (2010)
Nechita, I., Pellegrini, C., Rochette, D.: A geometrical description of the universal \(1 \rightarrow 2\) asymmetric quantum cloning region. Quantum Inf. Process. 20(10), 1–18 (2021)
Penrose, R.: Applications of negative dimensional tensors. Comb. Math. Its Appl. 1, 221–244 (1971)
Studziński, M., Ćwikliński, P., Horodecki, M., Mozrzymas, M.: Group-representation approach to 1\(\rightarrow \) n universal quantum cloning machines. Phys. Rev. A 89(5), 052322 (2014)
Studziński, M., Horodecki, M., Mozrzymas, M.: Commutant structure of \(u^{\otimes (n - 1)} \otimes u^*\) transformations. J. Phys. A: Math. Theor. 46(39), 395303 (2013)
Scarani, V., Iblisdir, S., Gisin, N., Acin, A.: Quantum cloning. Rev. Mod. Phys. 77(4), 1225 (2005)
Studziński, M., Strelchuk, S., Mozrzymas, M., Horodecki, M.: Port-based teleportation in arbitrary dimension. Sci. Rep. 7(1), 10871 (2017)
Watrous, J.: The Theory of Quantum Information. Cambridge University Press, Cambridge (2018)
Wood, C.J., Biamonte, J.D., Cory, D.G.: Tensor networks and graphical calculus for open quantum systems. arXiv preprint arXiv:1111.6950 (2011)
Wood, C.J., Biamonte, J.D., Cory, D.G.: Tensor networks and graphical calculus for open quantum systems. Quantum Inf. Comput. 15(9–10), 759–811 (2015)
Werner, R.F.: Optimal cloning of pure states. Phys. Rev. A 58(3), 1827 (1998)
Wootters, W.K., Zurek, W.H.: A single quantum cannot be cloned. Nature 299(5886), 802–803 (1982)
Acknowledgements
We thank the reviewers for the careful reading of our manuscript and for their useful and constructive remarks on our work. The authors were supported by the ANR project ESQuisses, grant number ANR-20-CE47-0014-01 as well as by the PHC programs Sakura (Random Matrices and Tensors for Quantum Information and Machine Learning) and Procope (Entanglement Preservation in Quantum Information Theory). I.N. has also received support from the ANR project STARS, grant number ANR-20-CE40-0008. C.P is also supported by the ANR projects Q-COAST ANR- 19-CE48-0003, “Quantum Trajectories” ANR- 20-CE40-0024-01, and “Investissements d’Avenir” ANR-11-LABX-0040 of the French National Research Agency.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: Structure of permutation operators
This appendix is devoted to the presentation of technical results on permutations operator. Numerous results or remarks are illustrated by graphical calculus.
1.1 graphical calculus
To illustrate the technical results of this section, we employ a graphical method for representing tensors, which is based on Penrose’s graphical notation [37]. Comparable approaches have been established in the context of tensor network states and categorical quantum information theory, as detailed in [2, 13, 44].
In this diagrammatic notation, tensors are depicted as boxes and wires.
The wires are labeled with indices, and a box represents the tensor’s value corresponding to those specific indices.
Specifically, a vector is represented by a box with a single wire extending to the left, labeled by the coordinate index. Conversely, a dual vector has its wire directed to the right.
Note that the left and right directions follow the convention that corresponds to the standard right-to-left composition in linear algebra.
Tensor diagrams can be combined using two methods: tensor product and tensor contraction. The tensor product combines two diagrams either vertically or horizontally.
Tensor contraction combines two diagrams by taking the sum across all common indices.
Specifically, we recover the scalar product.
The matrix product.
And the matrix trace.
Scalars multiply the diagrams and are displayed adjacent to the diagram.
In conclusion, the transposition can be visually illustrated by exchanging the “input” and “output” wires of a box representing a matrix. The partial transposition involves only the permutation of wires corresponding to specific spaces.
1.2 Contributions of permutations
In Sect. 3, we only focus on permutations where 0 is not a fixed point. This is motivated by the following remarks
Let \(T: \mathcal {M}_d \rightarrow {(\mathcal {M}_d)}^{\otimes N}\) be a \(\mathcal {U}(d)\)-covariant quantum channel, and \(C_T\) its Choi matrix, then from Proposition 3.2 we have for all \(\rho \in \mathcal {D}^1_d\)
for some \(\beta _\sigma \in \mathbb {C}\) (see Example A.3 for the case \(N=2\)).
Let us justify that \(\sigma \in \mathfrak {S}_{N + 1}\) such that \(\sigma (0) = 0\) do not contribute to the performance of the cloning map, whereas the \((N + 1)\) cycles contribute the most to the copies. Indeed, for a permutation \(\sigma \in \mathfrak {S}_{N + 1}\) such that \(\sigma (0) = 0\) we have \(\Pi _\sigma ^{\mathsf {\Gamma }} = \Pi _\sigma \) and for all \(\rho \in \mathcal {D}^1_d\)
whose contributions consist in deteriorating the quality of the cloning, since the quantity \(\hbox {Tr}_1 \sigma \) does not depend on \(\rho \). Now, if \(\sigma \in \mathfrak {S}_{N + 1}\) is a \((N + 1)\) cycle, the marginals of
are all equals to identity, i.e., \({(T_{\scriptscriptstyle \text {perfect}})}_i(\rho ) = \rho \) for all \(\rho \in \mathcal {M}_d\). This would be the case of a perfect quantum cloning map, but the operator \(T_{\scriptscriptstyle \text {perfect}}\) are of course not completely positive operator due to the no-cloning theorem (see Example A.4 for computations with graphical calculus).
More generally, let \(T: \mathcal {M}_d \rightarrow {(\mathcal {M}_d)}^{\otimes N}\) be a \(\mathcal {U}(d)\)-covariant quantum channel. From Proposition 3.2, the marginals \(T_i\) are for all \(\rho \in \mathcal {D}^1_d\)
where the \(p_i\) are the sums of the \(\frac{N!}{2}\) coefficients \(\beta _\sigma \) of the permutations contributing for this copy, that is, the \(\sigma \) such that i is in the orbit of 0 under the action of \(\sigma \). The following examples illustrate some results in terms of graphical calculus.
Example A.1
(Partition of \(\Sigma _{a,b}, N = 3\)) The permutation set \(\big \{ \Pi ^{\mathsf {\Gamma }}_\sigma \; \big | \;\sigma \in \mathfrak {S}_4, \quad \sigma (0) \ne 0 \big \}\) is partitioned in
Example A.2
(Partially transposed permutations, \(N = 3\)) On \(\mathfrak {S}_{4}\), the \((0 \, i)\)-partially transposed permutations are
Example A.3
(Choi Matrix \(N = 2\)) Let \(T: \mathcal {M}_d \rightarrow \mathcal {M}_d \otimes \mathcal {M}_d\) be a \(\mathcal {U}(d)\)-covariant quantum channel, then its Choi matrix \(C_T\) is:
Example A.4
(Contributions of quantum cloning) Let \(T: \mathcal {M}_d \rightarrow \mathcal {M}_d \otimes \mathcal {M}_d\) be a \(\mathcal {U}(d)\)-covariant quantum channel, such that its Choi matrix \(C_T\) is the partially transposed cycle \((1 \, 2 \, 3)\):
Then, the two marginals are
and
1.3 Proof of Lemma lem:struct
In this subsection, we prove the key Lemma 3.4, which gives the structure of partially permutations operators and action on vectors. We illustrate as well with graphical calculus.
Lemma A.5
Let \(1 \le a,b \le N\), for any \(\sigma \in \Sigma _{a,b}\), there exists a unique \(\hat{\sigma } \in \mathfrak {S}_{N - 1}\) such that the partially transposed permutation operator \(\sigma ^{\mathsf {\Gamma }}\) is
Proof
Let \(1 \le a,b,c,d \le N\), and \(\sigma \in \Sigma _{a,b}\), then \(\Pi _{(c \, a)} \; \Pi _\sigma ^{\mathsf {\Gamma }} \; \Pi _{(d \, b)}\) is a partially transposed permutation operator from \(\Sigma _{c,d}\). Let \(\sigma ' \in \Sigma _{1,1}\) be the permutation such that
There is a unique \(\hat{\sigma } \in \mathfrak {S}_{N - 1}\), such that
and finally
where the permutation operator \(\Pi _{\hat{\sigma }}\) is
\(\square \)
Example A.6
Let \(\sigma = (0 \, 3 \, 2) \in \mathfrak {S}_4\). Its partial transposition operator
can be decomposed into
Recall that for two integers i, j, we denote by \((i \,: \, j)\) the permutation \(\mathfrak {S}_n\) defined by
That is, the \(|i - j|\) cycle of \(\mathfrak {S}_n\) between i and j.
Example A.7
The non-trivial \((i \,: \, j)\)-cycles of \(\mathfrak {S}_3\) are:
Lemma A.8
Let \(1 \le a,b,c \le N\), and let \(\sigma \in \Sigma _{a,b}\). Let \(\phi \in \mathcal {H}^{\otimes (N - 1)}\), then
Proof
From Lemma A.5, we can write \(\Pi _\sigma ^{\mathsf {\Gamma }} = \Pi _{(1 \, a)} \, (\omega _{(0,1)} \otimes \Pi _{\hat{\sigma }}) \; \Pi _{(1 \, b)}\), or equivalently
Together with \(\omega \left| \Omega \right\rangle = d \cdot \left| \Omega \right\rangle \), we have the following relations:
and
If \(b = c\), then
such that finally \(\Pi _\sigma ^{\mathsf {\Gamma }} \big ( \left| \Omega \right\rangle _{(0,c)} \otimes \left| \phi \right\rangle \big ) = d \cdot \Pi _{(1 \, a)} \Big [ \left| \Omega \right\rangle _{(0,1)} \otimes \big ( \Pi _{\hat{\sigma }} \circ \Pi _{((b - 1) \,: \, 0)} \circ \big ) \left| \phi \right\rangle \Big ]\). If \(b \ne c\), we have the following relation:
for some \(\sigma \in \mathfrak {S}_{N-1}\). Note that \(\left| \phi \right\rangle \) is permuted according to \(\sigma \) which is a cycle whose order depends on whether \(c<b\) or \(b<c\). That is,
Then,
such that finally
To conclude, from the relation
we obtain
\(\square \)
Example A.9
Let \(\sigma = (0 \, 3 \, 1 \, 2) \in \mathfrak {S}_4\), and let \(\phi \in \mathcal {H}^{\otimes 2}\), then
In order to simplify the next equations, we use the notations \(\Pi _{\hat{\sigma }(a,b,c)}\) and \(\Pi _{b,c}\), defined by:
and
The following lemma concerns the image and the kernel of operators \(\Pi _\sigma ^{\mathsf {\Gamma }}\).
Lemma A.10
For any \(1 \le a,b \le N\) and any \(\sigma \in \Sigma _{a,b}\),
Proof
Let \(1 \le a,b \le N\), let \(\sigma \in \Sigma _{a,b}\), and let \(v \in {{\,\textrm{Im}\,}}\Pi _\sigma ^{\mathsf {\Gamma }}\), then \(v = \left| \Omega \right\rangle _{(0,a)} \otimes \left| \phi \right\rangle \) with \(\phi \in \mathcal {H}^{\otimes (N - 1)}\), from Lemma A.5, and
Let \(\phi \in \mathcal {H}^{\otimes (N - 1)}_d\), and define . Then from Lemma A.8, \(\Pi _\sigma ^{\mathsf {\Gamma }}(u) = \left| \Omega \right\rangle _{(0,a)} \otimes \left| \phi \right\rangle \), and
Since on finite-dimensional spaces, the kernel of a linear map M is equal to the orthogonal complement of the image of adjoint \(M^*\), i.e., \({{\,\textrm{Ker}\,}}M = {\big ( {{\,\textrm{Im}\,}}M^* \big )}^\perp \), and because the adjoint of \(\Pi _\sigma ^{\mathsf {\Gamma }}\) is a partially transposed permutation operator \(\Pi _{\sigma ^*}^{\mathsf {\Gamma }}\) with \(\sigma ^* \in \Sigma _{b,a}\), we have
\(\square \)
1.4 Partial trace of partially transposed permutation operators
Lemma A.11
Let \(1 \le a,b \le N\), then
where the orthogonal projector onto the trivial representation subspace of \(\mathfrak {S}_{N - 1}\) is defined by \(P^+_{\mathfrak {S}_{N - 1}} = \frac{1}{(N - 1)!} \sum _{\sigma \in \mathfrak {S}_{N - 1}} \Pi _\sigma \).
Proof
For the first equation, we have from Lemma A.5
where \(\Pi _{(0 \, (a - 1))}\) is a permutation operator of \(\mathfrak {S}_N\). For the second equation, since the partial transposition is a linear operator, we have the equality
For any \(\sigma \in \Sigma _{a,b}\), the relation
holds for all \(U \in \mathcal {U}(d)\), such that \(\hbox {Tr}_{\scriptscriptstyle [\![ 1,N ]\!]} \big [ \sum _{\sigma \in \Sigma _{a,b}} \Pi _\sigma ^{\mathsf {\Gamma }} \Big ]\) is a multiple of the identity, and
with \(c = \frac{1}{d} \hbox {Tr}\big [ \sum _{\sigma \in \Sigma _{a,b}} \Pi _\sigma \Big ]\). For a permutation \(\sigma \in \mathfrak {S}_n\), we write \(\# \sigma \) the number of disjoint cycles of \(\sigma \). Then,
Let \(\sigma \in \Sigma _{a,a}\) and write \(\sigma = c_1 \circ \cdots \circ c_k\) the decomposition of \(\sigma \) in disjoint cycles \(c_i = (s^i_1 \, \cdots \, s^i_{l_i})\). If it exists \(i \in [k]\) such that \(a,b \in c_i\), then \(c_i \circ (a \, b)\) can be decomposed in two disjoint cycles. Otherwise, if there exist \(i,j \in [k]\) such that \(a \in c_i\) and \(b \in c_j\), then \(c_i \circ c_j \circ (a \, b)\) can be decomposed in one disjoint cycle. Finally,
But since \(\sigma \in \Sigma _{a,a}\) and \(b \ne a\), in the decomposition of \(\sigma \) in disjoint cycles, a is in the cycle \((0 \, a)\), and \(\# \big [ \sigma \circ (a \, b) \Big ] = \# \sigma - 1\). That is
\(\square \)
1.5 Scalar product of vectors in \(\mathcal {V}^\lambda \)
Lemma A.12
Let \(\lambda _1, \lambda _2\) be two irreducible representations of \(\mathfrak {S}_{N - 1}\), let \(\textbf{v}_1, \textbf{v}_2\) be two normalized vectors in the irreducible subspace \(\lambda _1\) and \(\lambda _2\) of \(\mathfrak {S}_{N - 1}\), and let \(1 \le k,l \le N\). Then,
As a consequence,
and if \(\lambda _1\) and \(\lambda _2\) are distinct,
In particular, the vector spaces \({{\,\textrm{Span}\,}}\mathcal {V}^\lambda \), defined in Eq. (3.3), are in orthogonal direct sum.
Proof
Since both \(\textbf{v}_1\) and \(\textbf{v}_2\) are normalized, we have \(\langle \textbf{v}_i|\textbf{v}_i\rangle = 1\) for all \(i \in \{1, 2\}\). Then, from \(\langle \Omega |\Omega \rangle = d\) we have
When \(k \ne l\), from the equation \(\left\langle \Omega _{(0, k)} \otimes \phi \right| \left| \Omega _{(0, l)} \otimes \psi \right\rangle = \left\langle \phi \right| \Pi _{((l - 1) \,: \, (k - 1))} \left| \psi \right\rangle \) for any \(\phi , \psi \in \mathcal {H}^{\otimes (N - 1)}\), we have
When we look at \(\lambda _1, \lambda _2\), two distinct irreducible representations of \(\mathfrak {S}_{N - 1}\), the scalar product becomes
Note that \(\Pi _{((l - 1) \,: \, (k - 1))} \left| \textbf{v}_2\right\rangle \) lives in the irreducible subspace of \(\mathfrak {S}_{N - 1}\) associated with \(\lambda _2\), since the irreducible representations are stable by any representation of permutations. Then, both scalar products are null because the irreducible spaces are orthogonal each other.\(\square \)
Remark A.13
In the particular case of the trivial representation, we have for all \(1 \le k,l \le N\) and all normalized \(\textbf{v}\) in the irreducible subspace \(\vee ^{(N - 1)}(\mathcal {H})\),
From Lemma A.8, the action of the permutation operators \(\Pi ^{\mathsf {\Gamma }}_{\sigma _{a,b}}\) on the vectors \(\left| \Omega \right\rangle _{(0,k)} \otimes \left| \textbf{v}\right\rangle \), for all \(1 \le k \le N\), is
with \(\Pi ^\lambda _\sigma = P^\lambda \, \Pi _\sigma \, P^\lambda \), and \(P^\lambda \) the projector onto the irreducible subspace \(\lambda \).
For any permutation operator \(\Pi ^{\mathsf {\Gamma }}_{\sigma _{a,b}}\), we know from Lemma A.10 that the \(V^{\lambda }\)’s form a generators of the orthogonal complement of the kernel of \(\Pi ^{\mathsf {\Gamma }}_{\sigma _{a,b}}\). However, in general they do not define a basis. We can extract a subset of linearly independent vectors to form a basis. In this basis, \(\Pi ^{\mathsf {\Gamma }}_{\sigma _{a,b}}\) can be block diagonalized due to Eq. (31), since for all irreducible representation \(\lambda \), and any vector v in \(\mathcal {V}^{\lambda }\), we have
Example A.14
(\(N = 3\)) Let \(x_1, x_2, x_3 \in \mathbb {R}_+\), and define
that is
Then, an eigenvector for the largest eigenvalue of R is
that is,
for some coefficients \(\beta _i\), and \(\textbf{v} = \displaystyle \sum _{1 \le i \le j < d} \left| ij\right\rangle \).
Appendix B: Some results in linear algebra
In the proofs, we use simple linear algebra results that are not mainstream. We gather them in this appendix for the convenience of the reader.
The first one is a consequence of [9], for which we provide a simple proof.
Lemma B.1
For all \(1 \le i, j \le n\), let \(M_{ij}\) be some matrices in \(\mathcal {M}_m\), and define the block matrix
and the matrix . We then have
Proof
Consider v and w two unit vectors such that \(\Vert M\Vert = \left\langle v\right| M \left| w\right\rangle \), we can decompose \(v=\displaystyle \sum \nolimits _{i=1}^n \vert v_i\rangle \otimes \vert i\rangle \) with \(v_i\in \mathbb C^m\) and we define \(v'=\displaystyle \sum \nolimits ^n_{i=1} \Vert v_i\Vert \cdot \vert i\rangle \). Since we consider the Euclidian norm, we have obviously \(\Vert v'\Vert =\Vert v\Vert =1\). We define \(w'\) similarly and then have
which ends the proof.\(\square \)
The following lemma is useful in the paper when we relate the spectrum of \(S_x\) with the one of \(\tilde{S}\)
Lemma B.2
Let \(A\in \mathcal M\) be an operator, and let \(n\ge m\) let \(\{ f_1, \ldots , f_n \}\) be a set of vectors such that \(\mathbb C^m = {{\,\textrm{Span}\,}}\{ f_1, \ldots , f_n \}\). We suppose that for all \(i=1,\ldots ,n\) there exists \(a_{ij}\), \(j=1,\ldots ,n\) such that
Denote \(\tilde{A}\) the \(n\times n\) matrix with coefficients \((a_{ij})\). Then,
Proof
Let F be the matrix composed of the vectors \(f_i\), \(i=1,\ldots ,n\) written in column. Then, F is a \(m\times n\) matrix of rank m since \(\mathbb C^m=Vect\{ f_1, \ldots , f_n \}\). By definition of \(\tilde{A}\), we have
Furthermore, for all scalar \(\alpha \)
Now using the fact that \(rank(AB)\le min(rank(A),rank(B))\) for all matrices A and B, if \(\alpha \in Spec A\) since \(\hbox {rank}(F)=m\)
Now, since \(n\ge m\), this necessarily implies that \(\hbox {rank}(\tilde{A}-\alpha I_n)<n\) which says that \(\alpha \in Spec \tilde{A}\) \(\square \)
Remark B.3
Note that the matrix \(\tilde{A}\) is not uniquely defined and depends on a choice of the way of writing \(Af_i=\sum _{i,j=1}^n a_{ij}f_j\), but the results concerning the inclusion are true for any form of the matrix \(\tilde{A}\). Of course the other inclusion is not true in general unless \(n=m\) and \(\{ f_1, \ldots , f_n \}\) is linearly independent (but this is trivial in this case).
Remark B.4
In the lemma, the condition \(Af_i=\sum _{i,j=1}^n a_{ij}f_j\) means that A leaves \(V= {{\,\textrm{Span}\,}}\{ f_1, \ldots , f_n \}\) invariant and we can rephrase this lemma by considering the restriction to A on V and then saying that \({{\,\textrm{Spec}\,}}{A_{\vert V}} \subseteq {{\,\textrm{Spec}\,}}{\tilde{A}}.\) which is what we use in the body of the paper.
Appendix C: The \(\mathcal Q\)-norm
Lemma C.1
For all \(x\in \mathbb R^N\), we have
In particular, the \(\Vert \cdot \Vert _{\mathcal Q}\) quantity is nonnegative. The first inequality is saturated if and only if the matrix \(S_x\) is a (nonnegative) multiple of the identity. The second inequality is saturated if and only if x has at most one nonzero entry.
Proof
For the first inequality, note that
The inequality above is saturated if and only if the eigenvalues of \(S_x\) are identical.
For the second inequality, we use the subadditivity of the \(\lambda _{\max }\) functional:
The inequality above is saturated if and only if the matrices \(|x_i| \cdot \omega _{(0,i)} \otimes I\) have a common largest eigenvector, which can happen only if the support of x has size 0 or 1.\(\square \)
Lemma C.2
For all \(x, y \in \mathbb {R}^N_+\), \(\Vert x + y\Vert _{\mathcal {Q}} \le \Vert x\Vert _{\mathcal {Q}} + \Vert y\Vert _{\mathcal {Q}}\).
Proof
This follows from the subadditivity of the \(\lambda _{\max }\) functional: For \(x, y \in \mathbb {R}^N_+\), we have
\(\square \)
Lemma C.3
For all \(t \in [0,1]^N\) and \(x \in \mathbb {R}^N_+\), \(\Vert t \cdot x\Vert _{\mathcal {Q}} \le \Vert x\Vert _{\mathcal {Q}}\).
Proof
Let us show that for \(0 \le x \le y\) (meaning that \(x_i\le y_i\), \(i=1,\ldots ,N\)), we have \(\Vert x\Vert _{\mathcal {Q}} \le \Vert y\Vert _{\mathcal {Q}}\). Let \(\chi = \sum _k \beta _k \left| \Omega _{(0,k)}\right\rangle \otimes \left| \textbf{v}\right\rangle \) be a normalized eigenvector corresponding to the largest eigenvalue \(\lambda _{\max }(S_x)\), from Theorem 3.13, that is
Note that due to the Perron–Frobenius theorem, \(\beta _i > 0\) for all \(i=1,\ldots ,N\). Let us point out two facts. First using Lemma 3.6
Second
Now using that \(\left( \sum _{i=1}^N\beta _i\right) ^2\ge \sum _{i=1}^N\beta _i^2\) we get that \(1\le d\left( \sum _{i=1}^N\beta _i\right) ^2\) then since \(\beta _i\ge 0\)
This way since \(y\ge x\), we have that \(\lambda _{max}(S_y)\ge \langle \chi ,S_y\chi \rangle \) then
Then,
which was the desired result.\(\square \)
Appendix D: Optimal cloning
Lemma D.1
Let some positive real numbers \({(\beta _i)}_{1 \le i \le N}\). The operator
is an orthogonal projection if and only if the condition
is satisfied. In this case, the operator
is a positive operator, such that \(\hbox {Tr}_{\scriptscriptstyle [\![ 1,N]\!]} C_{T_\beta } = I\)
Proof
Let some positive real numbers \({(\beta _i)}_{1 \le i \le N}\), then
The projection condition \({\big ( \widetilde{C}_{T_\beta } \big )}^2 = \widetilde{C}_{T_\beta }\) is equivalent to
Since (17) is assumed to be satisfied, then \(\widetilde{C}_{T_\beta }\) is an orthogonal projection and hence a positive operator. This way \(C_{T_\beta }\) is positive. We have now from Lemma A.11
Then,
where the last equation comes from \(\hbox {Tr}\big [ P^+_{\mathfrak {S}_{N - 1}} \Big ] = \frac{N}{N + d - 1} \hbox {Tr}\big [ P^+_{\mathfrak {S}_{N}} \Big ]\).\(\square \)
Lemma D.2
For any \(1 \le a,b \le N\), and any \(\sigma \in \Sigma _{a,b}\), the permutation operator \(\sigma ^{\mathsf {\Gamma }}\) is the Choi matrix a linear map \(T_{\mu ,\nu }: \mathcal {M}_d \rightarrow {(\mathcal {M}_d)}^{\otimes N}\) defined by
for some permutations \(\mu \) and \(\nu \) in \(\mathfrak {S}_N\) such that \(\mu (0) = a - 1\) and \(\nu (b - 1) = 0\).
As a consequence for some fixed \(1 \le a,b \le N\),
where
Proof
Let \(1 \le a,b \le N\), and \(\sigma \in \Sigma _{a,b}\). Then from Lemma A.5, we have
such that on \(X \in \mathcal {M}_d\), the partial trace yields
And the result holds for \(\Pi _\mu = \Pi _{(0 \, (a - 1))}\) and \(\Pi _\nu = (I \otimes \Pi _{\hat{\sigma }}) \circ \Pi _{(0 \, (b - 1))}\). \(\square \)
Lemma D.3
Let an orthogonal projector \(\widetilde{C}_{T_\beta }\) for some positive real number \({(\beta _i)}_{1 \le i \le N}\) satisfying Eq. (17). Then for any \(1 \le i \le N\), we have
Let Choi matrix \(C_{T_\beta }\) for some positive real numbers \({(\beta _i)}_{1 \le i \le N}\) satisfying Eq. (17). Then for any \(1 \le i \le N\), we have
Proof
Since we are summing on all permutations \(\Pi ^{\mathsf {\Gamma }}_{\sigma _{a,b}}\), a direct calculation gives us
And then we look at the action of the Choi matrix \(C_{T_\beta }\) onto some \(\omega _{(0,i)}\) and take the trace.
Lemma A.11 yields
\(\square \)
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Nechita, I., Pellegrini, C. & Rochette, D. The asymmetric quantum cloning region. Lett Math Phys 113, 74 (2023). https://doi.org/10.1007/s11005-023-01694-8
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11005-023-01694-8