CP-Invariance and Complete Positivity in Quantum Information Theory

Transcription

1 in Quantum Information Theory D. W. Kribs, V. I. Paulsen, R. Pereira, and E. Størmer University of Guelph December 10, 2011

2 Order of Events Operator Systems on Complex Matrices Positive and Completely Positive Maps Operator Systems Right-CP-Invariant Cones Minimal and Maximal Operator Systems Other CP-Invariant Cones from Quantum Information Theory

3 Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps What are Cones and Completely Positive Maps? We will be mostly concerned with the space of n n complex matrices M n. A cone C M n is a set of Hermitian operators that is invariant under positive scaling. A cone C is called convex if X + Y C whenever X, Y C.

4 Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps What are Cones and Completely Positive Maps? We can also consider cones of linear maps acting on M n (i.e., in L(M n )). A cone C L(M n ) is a set of Hermiticity-preserving maps that is invariant under positive scaling. An adjoint map is a map of the form X A XA, where A denotes the adjoint operator of A. We will denote this map by Ad A. The set of adjoint maps is a cone. Its convex hull is the cone of completely positive maps.

5 What is an Operator System? Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps An (abstract) operator system on M n is a family of convex cones C m M m M n (one cone for each m N) that satisfy two properties: 1. C 1 = M + n, the cone of positive-semidefinite operators in M n ; and 2. for each m 1, m 2 N and A M m1,m 2 we have (Ad A id n )(C m1 ) C m2. Intuitively, these restrictions say that each cone C m is somehow like the cone M + n.

6 Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps Examples of Operator Systems on M n Quantum information theorists are actually familiar with some operator systems on M n already. The most natural operator system on M n is the one that arises by making the association M m M n = Mmn in the natural way and letting C m = M + mn; the cone of positive semidefinite operators. Keep this naïve operator system in mind! We will denote this naïve operator system simply by M n. We will denote other general operator systems on M n by things like O 1 and O 2.

7 Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps Examples of Operator Systems on M n An operator X M m M n is called separable if it can be written as X = i p i ψ i ψ i φ i φ i, p i 0 i. If we let S m be the cone of separable operators in M m M n, then {S m } is an operator system (which we will denote by OMAX ). An operator X M m M n is called block positive if ( a b )X ( a b ) 0 for all a C m, b C n. If we let P m be the cone of block positive operators in M m M n, then {P m } is an operator system (which we will denote by OMIN).

8 Positive Maps Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps A map on M n is called positive if it preserves positive semidefiniteness. That is, Φ L(M n ) is positive if Φ(X ) M + n whenever X M + n. Positive maps play an important role in quantum information theory (particularly in entanglement theory). We will denote the set of postive maps by P. We often want a stronger condition than positivity when dealing with operator systems though. We don t just want the cones C 1 to be preserved, but we want all of the cones {C m } to be preserved.

9 Completely Positive Maps Cones and Completely Positive Maps Operator Systems Positive and Completely Positive Maps Suppose we are given operator systems O 1 and O 2, defined by cones {C m } and {D m }, respectively. A map Φ L(M n ) is called completely positive from O 1 to O 2 if (id m Φ)(C m ) D m for all m. The set of completely positive maps from O 1 to O 2 will be denoted by CP(O 1, O 2 ). CP(M n, M n ) is the usual set of completely positive maps that quantum information theorists know and love. For brevity, we will denote it simply by CP.

10 Uniqueness of Operator Systems Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones The main workhorse that allows us to deal with operator systems is the following result, which says that they are uniquely determined by the nth cone that is, the cone C n M n M n. Proposition Let C n M n M n be a convex cone such that S n C n P n ; and (Ad A id n )(C n ) C n for all A M n. Then there exists a unique family of cones {C m } m n such that {C m } m=1 defines an operator system on M n.

11 Uniqueness of Operator Systems Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones Corollary Let Φ : M n M n and let O 1 and O 2 be operator systems defined by families of cones {C m } m=1 and {D m} m=1, respectively. Then Φ CP(O 1, O 2 ) if and only if (id n Φ)(C n ) D n. The above result generalizes the well-known result of Choi that says that Φ is completely positive if and only if it is n-positive.

12 What is a Right-CP-Invariant Cone? Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones Observe that if Φ 1 CP(O 1, O 2 ) and Φ 2 CP(O 2, O 3 ), then Φ 2 Φ 1 CP(O 1, O 3 ). In particular, if Φ 1 CP and Φ 2 CP(M n, O), then Φ 2 Φ 1 CP(M n, O). We use the term right-cp-invariant for cones C L(M n ) with the property that C CP = C.

13 Left-CP-Invariant and Mapping Cones Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones We just saw that CP(M n, O) is always right-cp-invariant. Similarly, CP(O, M n ) is always left-cp-invariant. A cone that is both left-cp-invariant and right-cp-invariant is called a mapping cone such cones are well-studied for independent reasons in operator theory. Examples of mapping cones include the cones of positive, completely positive, and entanglement-breaking maps (denoted by P, CP, and S, respectively).

14 Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones Recall that CP(M n, O) is always right-cp-invariant. It turns out that right-cp-invariance completely characterizes the possible cones of completely positive maps. Theorem Let C L(M n ) be a convex cone. The following are equivalent: C is right-cp-invariant with S C P. There exists an operator system O such that C = CP(M n, O). Furthermore, if O is defined by the cones {C m } m=1 then C n is the cone of Choi matrices of maps in C. Thus, O is uniquely determined by C.

15 Basic Examples Uniqueness of Operator Systems Right-CP-Invariant Cones Operator Systems Right-CP-Invariant Cones The previous theorem established a bijection between operator systems and right-cp-invariant cones, and furthermore showed that bijection is essentially the same as the Choi-Jamio lkowski isomorphism. The naïve operator system M n corresponds to the cone of standard completely positive maps CP(M n, M n ). The operator system OMIN, defined by the cones of block-positive operators {P m }, corresponds to the cone of positive maps P. OMIN is the largest possible operator system. The operator system OMAX, defined by the cones of separable operators {S m }, corresponds to the cone of entanglement-breaking maps S. OMAX is the smallest possible operator system.

16 k-positive Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps A linear map Φ L(M n ) is called k-positive if id k Φ is positive. k-positive maps play a role in entanglement theory analogous to the role of positive maps they help distinguish states with Schmidt number k from those with Schmidt number > k. The cone of k-positive maps is right-cp-invariant. The cones that define the corresponding operator system are the cones of k-block positive operators. That is, operators such that v X v 0 whenever the Schmidt rank of v is k. This operator system is called the super k-minimal operator system, denoted OMIN k. It is the largest operator system {C m } such that C m = M + mn for 1 m k.

17 k-entanglement Breaking Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps A linear map Φ L(M n ) is called k-entanglement breaking if (id n Φ)(X ) has Schmidt number k for any X M + n 2. The cone of k-entanglement breaking maps is right-cp-invariant. The operator system associated with the cone of k-entanglement breaking maps is defined by the cones of operators with Schmidt rank k. This operator system is called the super k-maximal operator system, denoted OMAX k. It is the smallest operator system {C m } such that C m = M + mn for 1 m k.

18 Anti-Degradable Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps A map Φ CP is called anti-degradable if there exists Ψ CP such that Ψ Φ C = Φ, where Φ C is the complementary map of Φ. The cone of anti-degradable maps is convex and right-cp-invariant. This cone is not left-cp-invariant. The operator system it gives rise to does not seem to be one that has been studied in operator theory.

19 Anti-Degradable Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps An operator X (M m M n ) + is called shareable if there exists X (M m M n M n ) + such that Tr 2 ( X ) = Tr 3 ( X ) = X. We will denote by H m M m M n the set of shareable operators. The Choi matrix of an anti-degradable map is a shareable operator, and a shareable operator is always the Choi matrix of some anti-degradable map. The operator system associated with the cone of anti-degradable maps is {H m }.

20 Local k-broadcasting Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps A map Φ CP is called local k-broadcasting if there exists a completely positive Φ : M n Mn k with the property that Φ = Tr i Φ for all 1 i k. In the k = 2 case, these are exactly the anti-degradable maps (slightly non-trivial, but not horribly so). These cones, like the cone of anti-degradable maps, are convex and right-cp-invariant. These cones are not left-cp-invariant (except in the trivial k = 1 and k = cases, in which case we get the cones of completely positive and entanglement breaking maps, respectively).

21 Local k-broadcasting Maps k-positive and k-entanglement Breaking Maps Anti-Degradable and Local k-broadcasting Maps An operator X (M m M n ) + is called k-shareable if there exists X (M m M k n ) + such that Tr i ( X ) = X for all 2 i k + 1. We will denote by H k m M m M n the set of k-shareable operators. The Choi matrix of a local k-broadcasting map is a k-shareable operator, and a k-shareable operator is always the Choi matrix of some local k-broadcasting map. The operator system associated with the cone of local k-broadcasting maps is {H k m}. These operator systems provide a natural (infinite) hierarchy that interpolates between M n and OMAX.

22 Open Questions Open Questions Further Reading 1. What other natural convex cones are right-cp-invariant? a) A possible example is the cone of locally entanglement-annihilating maps: maps Φ such that (Φ Φ)(X ) S n for all X (M n M n ) +. This cone is right-cp-invariant (and left-cp-invariant), but is it convex? b) What about the set of entanglement binding maps? This set is right-cp-invariant. However, it is convex (and hence corresonds to an operator system) if and only if all bound entangled states are PPT (decade-old open problem). 2. This seems to be a strong link between quantum information theory and operator theory. Find applications going in either direction!

23 Further Reading Open Questions Further Reading N. J., D. W. Kribs, V. I. Paulsen, and R. Pereira, Minimal and maximal operator spaces and operator systems in entanglement theory. J. Funct. Anal , (2011). E-print: arxiv: [math.oa] N. J. and E. Størmer, Mapping cones are operator systems. Preprint (2011). E-print: arxiv: [math.oa]