Arxiv:1101.2944V2 [Quant-Ph] 30 Jul 2021 Entropic Uncertainty Relations

arXiv:1101.2944v2 [quant-ph] 30 Jul 2021 rporpy[–1,adsga rcsig[2 13]. [12, wid processing applied signal entanglemen and are as such [9–11], They mechanics, cryptography quantum simultaneously. to unrelated precision and obs high related incompatible arbitrarily that inequalities to rigorous mos sured by the reveal of betw They differences one theories. conceptual are many relations underlying uncertainty mechanics, quantum analog, classical a Without Introduction 1 ascuet 23,USA. 02139, Massachusetts ievle of eigenvalues dniy htw get we that (density) | sdfie ste(ieeta)Sannetoyof entropy Shannon (differential) the as defined is o 1]frobservables: for [14] son nrp of entropy Ψ i ∗ † h aineof variance The . eascaearno variable random a associate We h aoscmuao-ae esnegucranypicpeis principle uncertainty Heisenberg commutator-based famous The rsn drs:Cne o hoeia hsc,Massachuse Physics, Theoretical for Center address: Present [email protected] nrpcucranyrltosi multidimensional in relations uncertainty Entropic eateto hsc,Uiest fClfri,Berkeley California, of University Physics, of Department iesoa ainebsducranypicpe h ar bou The conjecture. lower principle. optimal uncertainty provide variance-based They dimensional states. previou one-mode generalizing for twofold lations derived, are spaces momentum omttrbsdetoi netit eain nmulti in relations uncertainty entropic Commutator-based A ˆ denoted , A ˆ n h rbblt dniy that (density) probability the and , oiinadmmnu spaces momentum and position A a H ˆ eoe ∆ denoted , hnw esr h operator the measure we when ( A, ˆ | Ψ i ihnHag( Huang Yichen ,or ), ∆ aiona970 USA 94720, California A ˆ ∆ A A H ˆ uut2 2021 2, August B ˆ stevrac of variance the is , iha operator an with ( A ≥ ˆ Abstract ,or ),

h Ψ 1 | H [ A, ˆ ( P B 黄 ˆ A ( ] a | . Ψ 溢 ) where )), A i

辰 ae h value the takes A 2 A ˆ ˆ t nttt fTcnlg,Cambridge, Technology, of Institute tts / A h osbevle of values possible The . 4 ) ihrsett unu state quantum a to respect with il ocue iha open an with concludes ticle n h dffrnil Shannon (differential) the and , . ∗† nrpcucranyre- uncertainty entropic s d n ml h multi- the imply and nds P iesoa oiinand position dimensional ( a eeto 18,quantum [1–8], detection t e lsia n quantum and classical een stedsrbto of distribution the is ) ralscno emea- be cannot ervables udmna da of ideas fundamental t omltdb Robert- by formulated a Berkeley, , steprobability the is l naesboth areas in ely A r the are (1) A , We set ~ = 1 throughout this article. Denote the n-dimensional position and momentum space Hn. For the position and the momentum operatorsx, ˆ pˆ on H1, (1) reduces to ∆ˆx∆ˆp ≥ 1/4. (2) (2) is generalized to multidimensional spaces. The n-dimensional position and momentum space Hn is described by 2n operators Rˆ = (ˆx1, pˆ1, xˆ2,..., xˆn, pˆn), which satisfy the canonical commutation relations [Rˆj, Rˆk]= iΩjk, for j, k =1, 2,..., 2n, where n 0 1 Ω= . (3) −1 0 j=1 M For an n-mode density operator ρ, define the covariance matrix γ as

γjk = 2tr(ρ(Rj − tr(ρRj))(Rk − tr(ρRk))) − iΩjk. (4) γ is real and symmetric. The multidimensional variance-based uncertainty relation [15] is given by γ + iΩ ≥ 0. (5) Reference [16] provides much more detailed backgrounds. There are some other types of uncertainty relations for multimode states (e.g., Ref. [6]). A diﬀerent approach is to formulate uncertainty relations based on the Shannon entropy, rather than the variance. In the continuous variable case, Refs. [17, 18] prove inf{H(ˆx)+ H(ˆp)} = 1 + ln π (6) |Ψi

for H1. Equation (6) implies (2) [17, 18], showing the advantages of entropic uncertainty relations. Lots of entropic uncertainty relations in the discrete variable case are proposed (e.g., Ref. [19]). Reference [10] is a recent survey on this topic. The main contribution of the present work is to twofold generalize Eq. (6). The commutator-based entropic uncertainty relation Eq. (7) holds for more general operators on multidimensional position and momentum spaces. Equation (7) implies the multidimensional variance-based uncertainty relation (5), so every time we use (5) in applications, we might think of using Eq. (7) instead to produce better results.

Theorem 1 (entropic uncertainty relation in Hn). inf{H(Aˆ)+ H(Bˆ)} = 1 + ln π + ln |[A,ˆ Bˆ]|, (7) |Ψi where n n ˆ ′ ˆ ′ A = (aixî + aipî), B = (bixî + bipî) (8) i=1 i=1 X X ˆ ′ ′ are linear combinations of the components of R (ai, ai, bi, bi are real coefficients). Equivalently and more precisely,

n n ′ ′ inf H (aixî + aipî) + H (bixî + bipî) |Ψi ( i=1 ! i=1 !) X n Xn n ′ ′ ′ ′ = 1 + ln π + ln (aixî + aipî), (bixî + bipî) = 1 + ln π + ln (aibi − biai) . (9) " # i=1 i=1 i=1 X X X

2 ′ ′ Letting n = a1 = b1 =1, a1 = b1 = 0, Eq. (7) obviously reduces to Eq. (6). Equation (7) strengthens the importance of commutation relations and supports the intuitive idea that commutators quantify the extent of incompatibility of two operators. The paper is organized as follows. Section 2 provides the detailed proof of Theorem 1, which is arranged in lemmas to help you get the whole picture. Section 3 proves that Eq. (7) implies the variance-based uncertainty principle (5). Section 4 concludes with an open conjecture.

2 Proof of Theorem 1

We begin our discussion in H1. The fractional Fourier transform [20] Φ(ω) = Fˆ(θ)Ψ(x) plays an important role in the proof. It is deﬁned as

iθ−πi/2 iω2 ∞ iωx ix2 e − + Φ(ω)= e 2 tan θ e sin θ 2 tan θ Ψ(x) dx. (10) 2π sin θ r Z−∞ Naturally, Fˆ(0) is the identity map Iˆ, and

Fˆ(π/2) = Fˆ, Fˆ(−π/2) = Fˆ−1, (11)

where Fˆ and Fˆ−1 are the Fourier transform and the inverse Fourier transform, respectively. Fˆ satisﬁes [20] Fˆ(θ1 + θ2)= Fˆ(θ1) ◦ Fˆ(θ2)= Fˆ(θ2) ◦ Fˆ(θ1). (12) The eigenvector of the operator d xˆ cos θ +ˆp sin θ =x ˆ cos θ − i(sin θ) (13) dx corresponding to the eigenvalue ω is

πi/2−iθ iω2 iωx ix2 e − + − e 2 tan θ sin θ 2 tan θ . (14) r2π sin θ Let Ψ(x) be the position wave function of a quantum state |Ψi. Following from the deﬁnition of the fractional Fourier transform, the wave function in thex ˆ cos θi +ˆp sin θi representation is Ψi = Fˆ(θi)Ψ(x) for i = 1, 2, which implies Ψ2 = Fˆ(θ2 − θ1)Ψ1 from Eq. (12). Therefore, wave functions of the same quantum state in diﬀerent representations are related by the fractional Fourier transform. Lemma 1 ([21, 22]). For c ∈ R,

H(cAˆ)= H(Aˆ) + ln |c|. (15)

Lemma 2 ([13]).

inf{H(ˆx cos θ1 +ˆp sin θ1)+ H(ˆx cos θ2 +ˆp sin θ2)} = 1 + ln π + ln | sin(θ2 − θ1)|, (16) |Ψi

3 which can be rephrased as inf {H(|Ψ(x)|2)+ H(|Φ(ω)|2)} = 1 + ln π + ln | sin θ|, (17) Ψ(x)

where Ψ(x) runs through all legitimate wave functions and Φ(ω)= Fˆ(θ)Ψ(x).

Theorem 2 (Theorem 1 in H1).

inf{H(a1xˆ + a2pˆ)+ H(b1xˆ + b2pˆ)} = 1 + ln π + ln |[a1xˆ + a2p,ˆ b1xˆ + b2pˆ]| |Ψi

= 1 + ln π + ln |a1b2 − a2b1|. (18) Proof. Using Lemma 1 and then Lemma 2, we have

inf{H(a1xˆ + a2pˆ)+ H(b1xˆ + b2pˆ)} |Ψi

We have completed our discussion in H1. Let us move on to H2. Deﬁne cos θ − sin θ R = . (20) θ sin θ cos θ Lemma 3 (invariance of inﬁmum under local rotations).

xˆ1 xˆ2 xˆ1 xˆ2 inf H (a1 a2) +(a3 a4) , |Ψi + H (b1 b2) +(b3 b4) , |Ψi |Ψi pˆ pˆ pˆ pˆ 1 2 1 2 xˆ1 xˆ2 = inf H (a1 a2)Rθ1 +(a3 a4)Rθ2 , |Ψi |Ψi pˆ pˆ 1 2 xˆ xˆ + H (b b )R 1 +(b b )R 2 , |Ψi . (21) 1 2 θ1 pˆ 3 4 θ2 pˆ 1 2 T T T T Proof. We apply local rotations (ˆx1 pˆ1) → Rθ1 (ˆx1 pˆ1) and (ˆx2 pˆ2) → Rθ2 (ˆx2 pˆ2) . Under this transform, a state |Ψi, whose position wave function is Ψ(x1, x2), should become a new state denoted as Fˆ(θ1)⊗Fˆ(θ2)|Ψi, whose position wave function is (Fˆ(θ1)⊗Fˆ(θ2))Ψ(x1, x2). Thus, the left-hand side of Eq. (21) is equal to

xˆ1 xˆ2 ˆ ˆ inf H (a1 a2)Rθ1 +(a3 a4)Rθ2 , F (θ1) ⊗ F (θ2)|Ψi |Ψi pˆ pˆ 1 2 xˆ xˆ + H (b b )R 1 +(b b )R 2 , Fˆ(θ ) ⊗ Fˆ(θ )|Ψi . (22) 1 2 θ1 pˆ 3 4 θ2 pˆ 1 2 1 2

This is equal to the right-hand side of Eq. (21) as Fˆ(θ1) ⊗ Fˆ(θ2)|Ψi is a legitimate quantum state if and only if |Ψi is a legitimate quantum state.

4 The commutator is preserved under local rotations, which can be veriﬁed by direct computation:

xˆ xˆ xˆ xˆ (a a ) 1 +(a a ) 2 , (b b ) 1 +(b b ) 2 1 2 pˆ 3 4 pˆ 1 2 pˆ 3 4 pˆ 1 2 1 2 xˆ xˆ xˆ xˆ = (a a )R 1 +(a a )R 2 , (b b )R 1 +(b b )R 2 . (23) 1 2 θ1 pˆ 3 4 θ2 pˆ 1 2 θ1 pˆ 3 4 θ2 pˆ 1 2 1 2 Lemma 4 (invariance of inﬁmum under global rotations).

xˆ1 pˆ1 xˆ1 pˆ1 inf H (a1 a2) +(a3 a4) , |Ψi + H (b1 b2) +(b3 b4) , |Ψi |Ψi xˆ pˆ xˆ pˆ 2 2 2 2 xˆ1 pˆ1 = inf H (a1 a2)Rθ +(a3 a4)Rθ , |Ψi |Ψi xˆ pˆ 2 2 xˆ pˆ + H (b b )R 1 +(b b )R 1 , |Ψi . (24) 1 2 θ xˆ 3 4 θ pˆ 2 2 T Proof. For simplicity, the position wave function of |Ψi is denoted Ψ((x1 x2) ). By change T T T T of variables: (x1 x2) → Rθ(x1 x2) , which naturally yields (ˆx1 xˆ2) → Rθ(ˆx1 xˆ2) and T T (ˆp1 pˆ2) → Rθ(ˆp1 pˆ2) , we see that the left-hand side of Eq. (24) is equal to

xˆ1 pˆ1 x1 inf H (a1 a2)Rθ +(a3 a4)Rθ , Ψ Rθ |Ψi xˆ pˆ x 2 2 2 xˆ pˆ x + H (b b )R 1 +(b b )R 1 , Ψ R 1 . (25) 1 2 θ xˆ 3 4 θ pˆ θ x 2 2 2 T This is equal to the right-hand side of Eq. (24) as Ψ((x1 x2) ) is a legitimate wave function T if and only if Ψ(Rθ(x1 x2) ) is a legitimate wave function. The commutator is preserved under global rotations, which can be veriﬁed by direct computation:

xˆ pˆ xˆ pˆ (a a ) 1 +(a a ) 1 , (b b ) 1 +(b b ) 1 1 2 xˆ 3 4 pˆ 1 2 xˆ 3 4 pˆ 2 2 2 2 xˆ pˆ xˆ pˆ = (a a )R 1 +(a a )R 1 , (b b )R 1 +(b b )R 1 . (26) 1 2 θ xˆ 3 4 θ pˆ 1 2 θ xˆ 3 4 θ pˆ 2 2 2 2 Both local and global rotations are symplectic transformations, which preserve commutation relations (Ref. [16] provides detailed relevant backgrounds). This is an alternative argument for the validity of Eqs. (23) and (26). Lemma 5. inf{H(ˆx1)+ H(ˆx1 cos θ +ˆp1 sin θ)} = 1 + ln π + ln | sin θ|. (27) |Ψi

5 Proof. We ﬁrst show that the right-hand side is a valid lower bound and then prove its optimality. Let Ψ(x1, x2) be the position wave function of the quantum state |Ψi and Φ(ω, x2) be the wave function of the same state in the representation (ˆx1 cos θ +p ˆ1 sin θ, xˆ2). Thus, Φ(ω, x2)=(Fˆ(θ) ⊗ Iˆ)Ψ(x1, x2). Deﬁne ∞ 2 P (x2)= |Ψ(x1, x2)| dx1, (28) Z−∞

which satisﬁes ∞ P (x2) dx2 =1. (29) Z−∞ According to the deﬁnition of H(ˆx1) and due to the concavity of the Shannon entropy, ∞ ∞ ∞ 2 2 H(ˆx1)= − |Ψ(x1, x2)| dx2 ln |Ψ(x1, x2)| dx2 dx1 Z−∞ Z−∞ Z−∞ ∞ ∞ |Ψ(x , x )|2 ∞ |Ψ(x , x )|2 = − P (x ) 1 2 dx ln P (x ) 1 2 dx dx 2 P (x ) 2 2 P (x ) 2 1 Z−∞ Z−∞ 2 Z−∞ 2 ∞ ∞ |Ψ(x , x )|2 |Ψ(x , x )|2 ≥ − P (x ) 1 2 ln 1 2 dx dx . (30) 2 P (x ) P (x ) 1 2 Z−∞ Z−∞ 2 2 Similarly, ∞ ∞ |Φ(ω, x )|2 |Φ(ω, x )|2 H(ˆx cos θ +ˆp sin θ) ≥ − P (x ) 2 ln 2 dω dx , (31) 1 1 2 P (x ) P (x ) 2 Z−∞ Z−∞ 2 2 where Φ(ω, x2) Ψ(x1, x2) Φ(ω, x2)=(Fˆ(θ) ⊗ Iˆ)Ψ(x1, x2) =⇒ = Fˆ(θ) . (32) P (x2) P (x2) In the last equation, Ψ and Φ are regarded as functionsp only of xp1 and ω, respectively. Finally, applying Lemma 2,

Ψ(x1, x2)= ψ(x1)ϕ(x2), Φ(ω, x2)= φ(ω)ϕ(x2), (34) where ϕ is an arbitrary one-dimensional legitimate wave function. In this case, it is easy to verify that the lower bound in Eq. (27) is attained, proving its optimality. Lemma 6.

inf{H(ˆx1 + axˆ2)+ H(ˆx1 cos θ +ˆp1 sin θ)} = 1 + ln π + ln | sin θ|. (35) |Ψi

6 Proof. Let Ψ(x1, x2) be the position wave function of the quantum state |Ψi and Φ(ω, x2) be the wave function of the same state in the representation (ˆx1 cos θ +ˆp1 sin θ, xˆ2):

iθ−πi/2 iω2 ∞ iωx ix2 e − 1 + 1 Φ(ω, x )=(Fˆ(θ) ⊗ Iˆ)Ψ(x , x )= e 2 tan θ e sin θ 2 tan θ Ψ(x , x ) dx 2 1 2 2π sin θ 1 2 1 r Z−∞ iθ−πi/2 iω2 ∞ iω x −ax i x −ax 2 e − ( 1 2) + ( 1 2) = e 2 tan θ e sin θ 2 tan θ Ψ(x − ax , x ) dx (36) 2π sin θ 1 2 2 1 r Z−∞ implies

iaωx iθ−πi/2 iω2 ∞ iωx ix2 −2iax x +ia2x2 − 2 e − 1 + 1 1 2 2 e sin θ Φ(ω, x )= e 2 tan θ e sin θ 2 tan θ e 2 tan θ Ψ(x − ax , x ) dx 2 2π sin θ 1 2 2 1 r Z−∞ − iax x ia2x2 2 1 2+ 2 θ =(Fˆ(θ) ⊗ Iˆ) e 2 tan Ψ(x1 − ax2, x2) . (37) Finally, applying Lemma 5,

inf{H(ˆx1 + axˆ2)+ H(ˆx1 cos θ +ˆp1 sin θ)} |Ψi ∞ ∞ 2 2 = inf H |Φ(ω, x2)| dx2 + H |Ψ(x1 − ax2, x2)| dx2 |Ψi Z−∞ Z−∞ ∞ iaωx ∞ − iax x ia2x2 2 2 2 2 1 2+ 2 − θ θ = inf H e sin Φ(ω, x2) dx2 + H e 2 tan Ψ(x1 − ax2, x2) dx2 |Ψi ( Z−∞ Z−∞ !)

= 1 + ln π + ln | sin θ|. (38)

Similarly,

inf{H(ˆx1)+ H(ˆx1 cos θ +ˆp1 sin θ + axˆ2)} = 1 + ln π + ln | sin θ|. (39) |Ψi

Theorem 3 (Theorem 1 in H2).

inf{H(a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2)+ H(b1xˆ1 + b2pˆ1 + b3xˆ2 + b4pˆ2)} |Ψi

= 1 + ln π + ln |[a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2, b1xˆ1 + b2pˆ1 + b3xˆ2 + b4pˆ2]|. (40)

Proof. The idea is to reduce the most general case to more and more simpler cases by using lemmas proved previously. Assume b2 = b4 = 0 without loss of generality. Otherwise, we apply local rotations (Lemma 3), which preserve the commutator. We only need to prove

inf{H(a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2)+ H(b1xˆ1 + b3xˆ2)} |Ψi

= 1 + ln π + ln |[a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2, b1xˆ1 + b3xˆ2]|. (41)

7 Applying global rotations (Lemma 4), which preserve the commutator, we assume b3 = 0. We only need to prove

inf{H(a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2)+ H(b1xˆ1)} = 1 + ln π + ln |[a1xˆ1 + a2pˆ1 + a3xˆ2 + a4pˆ2, b1xˆ1]|. |Ψi (42) Assume a4 = 0 by applying local rotations. It suﬃces to show

inf{H(a1xˆ1 + a2pˆ1 + a3xˆ2)+ H(b1xˆ1)} = 1 + ln π + ln |[a1xˆ1 + a2pˆ1 + a3xˆ2, b1xˆ1]|, (43) |Ψi which is equivalent to (due to Lemma 1)

inf{H(ˆx1 cos θ +ˆp1 sin θ + a3xˆ2/a)+ H(ˆx1)} = 1 + ln π + ln |[ˆx1 cos θ +ˆp1 sin θ, xˆ1]| |Ψi = 1 + ln π + ln | sin θ|, (44) where a1 = a cos θ, a2 = a sin θ. Now we see that Eq. (44) is precisely Eq. (39).

We have completed our discussion in H2. Generally, Theorem 1 in Hn(n > 2) can be proved similarly with only the following minor revision (no essentially new ideas in- cluded). We should introduce R ∈ SO(n) (n-dimensional rotation) to replace the role of Rθ (two-dimensional rotation) in Lemma 4, simply because the global rotation becomes an n-dimensional rotation in Hn. The n-dimensional version of Eq. (26) can be veriﬁed by direct computation and making use of R ∈ SO(n), or by simply using the fact that R is a symplectic transformation. We have completed the proof of Theorem 1. Due to the concavity of the diﬀerential Shannon entropy, we observe that Eq. (7) also holds for mixed states, which are probabilistic mixtures of pure states.

3 Discussions

Proposition 1. Equation (7) implies the variance-based uncertainty principle (5). It also implies Seraﬁni’s multidimensional uncertainty principle—(8) in Ref. [6].

Proof. We ﬁrst provide an equivalent description of (5). Let d,d′ be two 2n-dimensional real ′ ′ T ′ ′ ′ T vectors: d =(a1 a1 a2 ··· an an) and d =(b1 b1 b2 ··· bn bn) . Obviously,

γ + iΩ ≥ 0 ⇐⇒ (d + id′)†(γ + iΩ)(d + id′) ≥ 0 ∀d,d′. (45)

If we deﬁne operators A,ˆ Bˆ as Eq. (8), it is equivalent to

n ˆ ˆ ′T ′ ′ ˆ ˆ ∆A + ∆B ≥|d Ωd| = (aibi − biai) = |[A, B]|, (46)

i=1 X because dT γd = 2∆Aˆ and d′T γd′ = 2∆Bˆ. We thus see that (5) is simply a direct consequence of the Heisenberg uncertainty principle. H(Aˆ) and ∆Aˆ are, respectively, the diﬀerential

8 Shannon entropy and the variance of the same distribution. From [21, 22], we have ∆Aˆ ≥ e2H(Aˆ)−1/(2π). By basic inequalities, we obtain

e2H(Aˆ)+2H(Bˆ)−2 e1+ln π+ln |[A,ˆ Bˆ]|−1 ∆Aˆ + ∆Bˆ ≥ 2 ∆Aˆ∆Bˆ ≥ 2 ≥ = |[A,ˆ Bˆ]|. (47) s 4π2 π p Finally, Eq. (7) implies Seraﬁni’s uncertainty principle for multimode states ((8) in Ref. [6]) because (8) in Ref. [6] is a necessary condition of (5) [6].

4 Conclusion and outlook

I have derived the commutator-based entropic uncertainty relation Eq. (7), which holds for more general Hermitian operators on multidimensional position and momentum spaces, twofold generalizing the previous entropic uncertainty relation Eq. (6). The lower bound in Eq. (7) is optimal, and Eq. (7) implies the multidimensional variance-based uncertainty principle (5). Every time we use (5) in applications, we might think of using Eq. (7) instead to produce better results. One might try to seek for a simpliﬁed proof of Theorem 1 by using the Stone–von Neu- mann theorem [23]. However, the present proof at least has the advantage of being elemen- tary. A fundamental and interesting problem is to study how far we can generalize Eq. (7). We restrict A,ˆ Bˆ to be of the form Eq. (8) in the present work, but does Eq. (7) hold for general Hermitian operators? At least we should modify Eq. (7) in the case that [A,ˆ Bˆ] is not a number operator. I propose the following open conjecture, which (if holds) implies the Heisenberg uncertainty principle Eq. (1).

Conjecture 1. For arbitrary Hermitian operators A,ˆ Bˆ on multidimensional position and momentum spaces, we have

H(A,ˆ |Ψi)+ H(B,ˆ |Ψi) ≥ 1 + ln π + ln hΨ|[A,ˆ Bˆ]|Ψi . (48)

If this conjecture is false, then can we add some loose restrictions on A,ˆ Bˆ (not as strong as the restriction that A,ˆ Bˆ should take the form of Eq. (8)) so that (48) holds?

References

[1] R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki. Quantum entanglement. Reviews of Modern Physics, 81(2):865–942, 2009. [2] O. G¨uhne and G. T´oth. Entanglement detection. Physics Reports, 474(1–6):1–75, 2009. [3] L.-M. Duan, G. Giedke, J. I. Cirac, and P. Zoller. Inseparability criterion for continuous variable systems. Physical Review Letters, 84(12):2722–2725, 2000. [4] R. Simon. Peres-Horodecki separability criterion for continuous variable systems. Phys- ical Review Letters, 84(12):2726–2729, 2000.

9 [5] O. Gühne. Characterizing entanglement via uncertainty relations. Physical Review Let- ters, 92(11):117903, 2004. [6] A. Serafini. Multimode uncertainty relations and separability of continuous variable states. Physical Review Letters, 96(11):110402, 2006. [7] Y. Huang. Entanglement criteria via concave-function uncertainty relations. Physical Review A, 82(1):012335, 2010. [8] Y. Huang. Erratum: Entanglement criteria via concave-function uncertainty relations [Phys. Rev. A 82, 012335 (2010)]. Physical Review A, 82(6):069903, 2010. [9] N. Gisin, G. Ribordy, W. Tittel, and H. Zbinden. Quantum cryptography. Reviews of Modern Physics, 74(1):145–195, 2002. [10] S. Wehner and A. Winter. Entropic uncertainty relations—a survey. New Journal of Physics, 12(2):025009, 2010. [11] C. H. Bennett and G. Brassard. Quantum cryptography: Public key distribution and coin tossing. Theoretical Computer Science, 560:7–11, 2014. [12] E. J. Candès, J. Romberg, and T. Tao. Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information. IEEE Transactions on Information Theory, 52(2):489–509, 2006. [13] X. Guanlei, W. Xiaotong, and X. Xiaogang. Generalized entropic uncertainty principle on fractional Fourier transform. Signal Processing, 89(12):2692–2697, 2009. [14] H. P. Robertson. The uncertainty principle. Physical Review, 34(1):163–164, 1929. [15] R. Simon, N. Mukunda, and B. Dutta. Quantum-noise matrix for multimode systems: U(n) invariance, squeezing, and normal forms. Physical Review A, 49(3):1567–1583, 1994. [16] S. Pirandola, A. Serafini, and S. Lloyd. Correlation matrices of two-mode bosonic systems. Physical Review A, 79(5):052327, 2009. [17] W. Beckner. Inequalities in Fourier analysis. Annals of Mathematics, 102(1):159–182, 1975. [18] I. Bialynicki-Birula and J. Mycielski. Uncertainty relations for information theory in wave mechanics. Communications in Mathematical Physics, 44(2):129–132, 1975. [19] H. Maassen and J. B. M. Uffink. Generalized entropic uncertainty relations. Physical Review Letters, 60(12):1103–1106, 1988. [20] V. Namias. The fractional order Fourier transform and its application to quantum mechanics. IMA Journal of Applied Mathematics, 25(3):241–265, 1980. [21] C. E. Shannon. A mathematical theory of communication. The Bell System Technical Journal, 27(3):379–423, 1948. [22] C. E. Shannon. A mathematical theory of communication. The Bell System Technical Journal, 27(4):623–656, 1948.

10 [23] M. H. Stone. Linear transformations in Hilbert space. III. Operational methods and group theory. Proceedings of the National Academy of Sciences of the United States of America, 16(2):172–175, 1930.