ENTANGLEMENT OF GAUSSIAN STATES Gerardo Adesso
ENTANGLEMENT OF GAUSSIAN STATES Gerardo Adesso ENTANGLEMENT OF GAUSSIAN STATES Gerardo Adesso
184 11. Efficient production of pure N-mode Gaussian states such a procedure consists in considering the reverse of the phase space 1 × (N − 1) Schmidt decomposition, as introduced in Sec. 2.4.2.1. Namely, a completely general (not accounting for the local invariances) state engineering prescription for pure Gaussian states can be cast in two main steps: (i) create a two-mode squeezed state of modes 1 and 2, which corresponds to the multimode state in its Schmidt form; (ii) operate with the most general (N − 1)-mode symplectic transformation S −1 on the block of modes {2, 3, . . . , N} (with modes i = 3, . . . , N initially in the vacuum state) to redistribute entanglement among all modes. The operation S −1 is the inverse of the transformation S which brings the reduced CM of modes {2, 3, . . . , N} in its Williamson diagonal form, see Sec. 2.2.2.1. It is also known that any such symplectic transformation S −1 (unitary on the Hilbert space) can be decomposed in a network of optical elements [193]. The number of elements required to accomplish this network, however, will in general greatly exceed the minimal number of parameters on which the entanglement between any two subsystems depends. Shifting the local-unitary optimization from the final CM, back to the engineering symplectic network, is in principle an extremely involved and nontrivial task. This problem has been solved in Ref. [GA14] for a special subclass of Gaussian states, which is of null measure but still of central importance for practical implementations. It is constituted by those pure N-mode Gaussian states which can be locally put in a standard form with all diagonal 2 × 2 submatrices in Eq. (2.20) (i.e. with null σqp in the notation of Appendix A). This class encompasses generalized GHZ-type Gaussian states, useful for CV quantum teleportation networks [236] (see Sec. 12.2), Gaussian cluster states [280, 242] employed in CV implementations of one-way quantum computation [155], and states of four or more modes with an unlimited promiscuous entanglement sharing (see Chapter 8). It also comprises (as proven in Sec. 7.1.2) all three-mode pure Gaussian states, whose usefulness for CV quantum communication purposes has been thoroughly investigated in this Dissertation (see Chapter 7, Chapter 10, and Sec. 12.3). In the physics of many-body systems, those states are quite ubiquitous as they are ground states of harmonic Hamiltonians with spring-like interactions [11]. As such, they admit an efficient “valence bond” description, as discussed in Chapter 13. For these Gaussian states, which we will call here block-diagonal — with respect to the canonical operators reordered as (ˆq1, ˆq2, . . . , ˆqN , ˆp1, ˆp2, . . . , ˆpN ) — the minimal number of local-unitarily-invariant parameters reduces to N(N − 1)/2 for any N. 28 Accordingly, one can show that an efficient scheme can be devised to produce block-diagonal pure Gaussian states, involving exactly N(N − 1)/2 optical elements which in this case are only constituted by single-mode squeezers and beam-splitters, in a given sequence [GA14]. We will now detail the derivation of these results explicitly, as it will lead to an important physical insight into the entanglement structure (which we define “generic”) of such block-diagonal Gaussian states. The latter, we recall, are basically all the resources currently produced and employed in optical realizations of CV quantum information and communication processing. 28 This number is easily derived from the general framework developed in Appendix A.2: for σqp = 0, Eqs. (A.5) and (A.6) reduce to σq = σ −1 p . The only further condition to impose after the local reduction is then diag(σq) = diag(σ −1 q ), which brings the number of free parameters of the symmetric σq from (N + 1)N/2 down to N(N − 1)/2.
11.2. Generic entanglement and state engineering of block-diagonal pure states 185 11.2. Generic entanglement, standard form and state engineering of block-diagonal pure Gaussian states 11.2.1. Generic entanglement of Gaussian states In this Section, based on Ref. [GA14] we address the question of how many physical resources are really needed to engineer and characterize entanglement in pure Gaussian states of an arbitrary number of modes, up to local unitary operations. Let us recall again (see Sec. 2.4.2) that for states of N ≤ 3 modes, it has been shown that such a number of minimal degrees of freedom scales as N(N − 1)/2 . For a higher number of modes, however, a richer structure is achievable by pure Gaussian states, as from symplectic arguments like those of Appendix A.2.1 a minimal number of parameters given by N(N − 2) can be inferred. A random state of N ≥ 4 modes, selected according to the uniform distribution over pure Gaussian states, will be thus reducible to a form characterized by such a number of independent quantities. However, in practical realizations of CV quantum information one is interested in states which, once prepared with efficient resources, still achieve an almost complete structural variety in their multipartite entanglement properties. Such states will be said to possess generic entanglement [166], where generic means practically equivalent to that of random states, but engineered (and described) with a considerably smaller number of degrees of freedom. Precisely, we define as “generic-entangled” those Gaussian states whose local entropies of entanglement in any single mode are independent, and bipartite entanglements between any pair of modes are unconstrained. Having a standard form for such N-mode Gaussian states, may be in fact extremely helpful in understanding and quantifying multipartite CV entanglement, in particular from the theoretical point of view of entanglement sharing and monogamy constraints (see Chapter 6), and from a more pragmatical approach centered on using entanglement as a resource. We show that, to achieve generic entanglement, for the global pure N-mode Gaussian state it is enough to be described by a minimal number of parameters (corresponding to the local-unitarily invariant degrees of freedom) equal to N(N − 1)/2 for any N, and thus much smaller than the 2N(2N + 1)/2 of a completely general CM. Therefore, generic entanglement appears in states which are highly not ‘generic’ in the sense usually attributed to the term, i.e. randomly picked. Crucially, we demonstrate that “generic-entangled” Gaussian states coincide with the above defined “block-diagonal” Gaussian states, i.e. with the resources typically employed in experimental realizations of CV quantum information [40]. Accordingly, we provide an optimal and practical scheme for their state engineering. 11.2.2. Minimal number of parameters Adopting the above definition of generic entanglement, we prove now the main Proposition 1. A generic-entangled N-mode pure Gaussian state is described, up to local symplectic (unitary) operations, by N(N − 1)/2 independent parameters. Proof. Let us start with a N-mode pure state, described by a CM σ p ≡ σ as in Eq. (2.20), with all single-mode blocks σi (i = 1 . . . N) in diagonal form: we can always achieve this by local single-mode Williamson diagonalizations in each
- Page 148 and 149: 134 7. Tripartite entanglement in t
- Page 150 and 151: 136 7. Tripartite entanglement in t
- Page 152 and 153: 138 7. Tripartite entanglement in t
- Page 154 and 155: 140 7. Tripartite entanglement in t
- Page 156 and 157: 142 7. Tripartite entanglement in t
- Page 158 and 159: 144 7. Tripartite entanglement in t
- Page 160 and 161: 146 7. Tripartite entanglement in t
- Page 162 and 163: 148 8. Unlimited promiscuity of mul
- Page 164 and 165: 150 8. Unlimited promiscuity of mul
- Page 166 and 167: 152 8. Unlimited promiscuity of mul
- Page 168 and 169: 154 8. Unlimited promiscuity of mul
- Page 171: Part IV Quantum state engineering o
- Page 174 and 175: 160 9. Two-mode Gaussian states in
- Page 176 and 177: 162 9. Two-mode Gaussian states in
- Page 178 and 179: 164 9. Two-mode Gaussian states in
- Page 180 and 181: 166 9. Two-mode Gaussian states in
- Page 182 and 183: 168 9. Two-mode Gaussian states in
- Page 184 and 185: 170 9. Two-mode Gaussian states in
- Page 186 and 187: 172 9. Two-mode Gaussian states in
- Page 189 and 190: CHAPTER 10 Tripartite and four-part
- Page 191 and 192: 10.1. Optical production of three-m
- Page 193 and 194: (out) (in) TRITTER 10.1. Optical pr
- Page 195 and 196: 10.2. How to produce and exploit un
- Page 197: CHAPTER 11 Efficient production of
- Page 201 and 202: 11.2. Generic entanglement and stat
- Page 203 and 204: 11.2. Generic entanglement and stat
- Page 205 and 206: 11.3. Economical state engineering
- Page 207: Part V Operational interpretation a
- Page 210 and 211: 196 12. Multiparty quantum communic
- Page 212 and 213: 198 12. Multiparty quantum communic
- Page 214 and 215: 200 12. Multiparty quantum communic
- Page 216 and 217: 202 12. Multiparty quantum communic
- Page 218 and 219: 204 12. Multiparty quantum communic
- Page 220 and 221: 206 12. Multiparty quantum communic
- Page 222 and 223: 208 12. Multiparty quantum communic
- Page 224 and 225: 210 12. Multiparty quantum communic
- Page 226 and 227: 212 12. Multiparty quantum communic
- Page 228 and 229: 214 12. Multiparty quantum communic
- Page 230 and 231: 216 12. Multiparty quantum communic
- Page 233 and 234: CHAPTER 13 Entanglement in Gaussian
- Page 235 and 236: 13.1. Gaussian valence bond states
- Page 237 and 238: Eq. (13.3) thus takes the form 13.1
- Page 239 and 240: 13.2. Entanglement distribution in
- Page 241 and 242: s 20 17.5 15 12.5 10 7.5 5 2.5 13.2
- Page 243 and 244: 13.3. Optical implementation of Gau
- Page 245 and 246: of the form Eq. (13.1), with 13.3.
- Page 247 and 248: 13.4. Telecloning with Gaussian val
11.2. Generic entanglement and state engineering of block-diagonal pure states 185<br />
11.2. Generic entanglement, standard form and state engineering of<br />
block-diagonal pure Gaussian states<br />
11.2.1. Generic entanglement of Gaussian states<br />
In this Section, based on Ref. [GA14] we address the question of how many physical<br />
resources are really needed to engineer and characterize entanglement in pure<br />
Gaussian states of an arbitrary number of modes, up to local unitary operations.<br />
Let us recall again (see Sec. 2.4.2) that for states of N ≤ 3 modes, it has been shown<br />
that such a number of minimal degrees of freedom scales as N(N − 1)/2 . For a<br />
higher number of modes, however, a richer structure is achievable by pure Gaussian<br />
states, as from symplectic arguments like those of Appendix A.2.1 a minimal number<br />
of parameters given by N(N − 2) can be inferred. A random state of N ≥ 4<br />
modes, selected according to the uniform distribution over pure Gaussian states,<br />
will be thus reducible to a form characterized by such a number of independent<br />
quantities.<br />
However, in practical realizations of CV quantum information one is interested<br />
in states which, once prepared with efficient resources, still achieve an almost complete<br />
structural variety in their multipartite entanglement properties. Such states<br />
will be said to possess generic entanglement [166], where generic means practically<br />
equivalent to that of random states, but engineered (and described) with a<br />
considerably smaller number of degrees of freedom.<br />
Precisely, we define as “generic-entangled” those Gaussian states whose local<br />
entropies of entanglement in any single mode are independent, and bipartite entanglements<br />
between any pair of modes are unconstrained. Having a standard form for<br />
such N-mode Gaussian states, may be in fact extremely helpful in understanding<br />
and quantifying multipartite CV entanglement, in particular from the theoretical<br />
point of view of entanglement sharing and monogamy constraints (see Chapter<br />
6), and from a more pragmatical approach centered on using entanglement as a<br />
resource.<br />
We show that, to achieve generic entanglement, for the global pure N-mode<br />
Gaussian state it is enough to be described by a minimal number of parameters<br />
(corresponding to the local-unitarily invariant degrees of freedom) equal to N(N −<br />
1)/2 for any N, and thus much smaller than the 2N(2N + 1)/2 of a completely<br />
general CM. Therefore, generic entanglement appears in states which are highly not<br />
‘generic’ in the sense usually attributed to the term, i.e. randomly picked. Crucially,<br />
we demonstrate that “generic-entangled” Gaussian states coincide with the above<br />
defined “block-diagonal” Gaussian states, i.e. with the resources typically employed<br />
in experimental realizations of CV quantum information [40]. Accordingly, we<br />
provide an optimal and practical scheme for their state engineering.<br />
11.2.2. Minimal number of parameters<br />
Adopting the above definition of generic entanglement, we prove now the main<br />
Proposition 1. A generic-entangled N-mode pure Gaussian state is described, up to<br />
local symplectic (unitary) operations, by N(N − 1)/2 independent parameters.<br />
Proof. Let us start with a N-mode pure state, described by a CM σ p ≡ σ as<br />
in Eq. (2.20), with all single-mode blocks σi (i = 1 . . . N) in diagonal form: we<br />
can always achieve this by local single-mode Williamson diagonalizations in each