Abstract
This paper studies the extremals and other faces of the completely positive and positive semidefinite-preserving linear transformations.
1. Introduction
This paper is the third in a sequence giving the related theory of the cone of positive semidefinite-preserving linear transformations on the complex vector space of complex matrices of order , denoted by , and its self-dual subcone of the completely positive linear transformations. Following the papers of Barker et al. [1] and Yopp and Hill [2], this third paper studies the extremals and other faces of these cones. The cone of completely copositive linear transformations, , fits nicely in this work as well.
For a list of all the cone-theoretic definitions used here, see [2]. For a list of characterization theorems for as well as , the Hermitian preservers, which are an ambient space for all the cones of this paper, see [3].
2. Classes of Extremals of
Yopp and Hill [2, 4] have characterized the extremals of and as follows.
Theorem 2.1. Let . Then,(i) if and only if there exists such that ; (ii)if , where , then if and only if for all ; (iii) if and only if there exist and such that .
Theorem 2.2. Let be an element of . Then,(i) if and only if for some ;(ii)if , where , then if and only if for all ;(iii) if and only if there exist and such that .
The remainder of this section focuses on the extremals of . The major result of [5] follows.
Theorem 2.3. Let . If or , then is an extremal of , and when the rank of is 1 or , the extremals are exposed.
The following result, originally given in a different setting by Loewy and Schneider [6], yields a (different) large class of extremals, namely, the nonsingular maps. Further, we note that the analogous result does not hold for . We first give an alternative proof for the theorem.
Theorem 2.4. Every nonsingular element of is an extremal of .
Proof. Let be a nonsingular element of . Since every nonsingular linear transformation is a vector space isomorphism, it follows from [2, Theorem 1.1(iv)] that . By [7, Theorem 2.B.2], it follows that is an extremal of .
The following example shows that this result does not apply to . Let , and let . Then, the linear transformation with matrix representation is a nonsingular element of . However, we observe that there exists no such that . It follows from [2, Theorem 3.4(i)] that is not an extremal of .
From the subcones and , we find two more classes of extremals of .
Theorem 2.5. Every extremal of and of is also an extremal of .
Proof. Assume that is an extremal of . By [2, Theorem 3.4(i)], is an extremal of if and only if there exists such that . Then, [5, Lemma 3.2] gives us that is an extremal of . An analogous argument implies that every extremal of is also an extremal of .
By definition, all extreme rays of and are also extreme rays of . Also, Theorem 2.5 and [2, Theorem 3.4] give us the following extremals for .
Theorem 2.6. Let . Then,(i)if or , where , and if for all , then ; (ii) if there exist and such that or , then .
3. Other Faces of and
It is well known that the transpose map is an element of but not . Since , it follows that . Thus, but . While it is well known that is a subcone but not a face of , it is an open question whether every proper face of (in the sense of a proper subset) is a face of . Examples of such faces do exist. The Siler cone is a one-dimensional face of both and [8, page 33].
In [2, Theorem 3.2], Yopp and Hill have characterized all the faces of as follows.
Theorem 3.1. if and only if there exist such that
One way in which we could show that a face of is not a face of would be to exhibit (at least) one element of that lies in the interior (equivalently, not in the boundary) of , as we could then apply the following result due to Barker and Schneider [9, Corollary 2.18].
Theorem 3.2. If and , then .
In [13], Kye gives us a criterion by which we can determine whether a given linear transformation is an element of int as follows. Let be nonzero, and let denote the one-dimensional projection matrix corresponding to the projection onto the subspace spanned by . Note that one-dimensional projections are precisely the extremals of , as characterized by Barker and Carlson [11] and further by Hill and Waters [12, Theorem 3.8]. This leads to Kye's result [13, Proposition 4.1].
Theorem 3.3. The linear map is an interior point of if and only if is nonsingular for all extremals .
A more common characterization of the interior of a positive cone of linear maps when specialized to gives the following: Thus, instead of requiring that the image of any nonzero positive semidefinite matrix be positive definite, as in the standard characterization, Kye's characterization instead requires that the image of any extremal be nonsingular.
We continue this section with two examples due to Kye [13].
The first example is the trace map, which is easily seen to be an element of . For nonzero , we have that which is nonsingular, giving us that the trace map is an element of int .
Kye's second example is the linear map where is the standard square root of a positive definite matrix (cf. [10]). We know that . Since is positive definite, is also positive definite, and Sylvester's law of inertia gives us that and have the same inertias, so that . Thus, for all positive definite matrices . Kye claims, for any , . Actually, it is not (making it an element of bd ). To see this, let , which is positive definite. Then, . In this case, for , that is, is simply the transpose map, an extremal of , as we observed at the beginning of this section, and thus not an interior point of .
4. Toward a Result Concerning Faces in Shared Boundaries
In Section 3, we have encountered a number of linear maps that lie in the intersection of the boundaries of and . We further study this area, which leads us to a still-open question: is a face of which lies in the boundary of necessarily also a face of ?
We do have the following.
Theorem 4.1. Let be a chain of (sub)cones. If , then .
Proof. Assume that and that , and . Since , it follows that . Since , we have that . Thus, .
In [2], Yopp and Hill give several results concerning the boundary of , one of which, when combined with its analogue for (also found in [2]), yields a sufficient condition for a linear map to be in the intersection of their boundaries.
Theorem 4.2. Let , and suppose that there exist such that . If , then .
Theorem 4.3. Let , and suppose that there exist such that where . If , then .
Combining these results gives the following.
Theorem 4.4. Let , and suppose that there exist such that . If , then .
Proof. Assume . Since for . If we let for , then and the conclusion is immediate.
Finally, combining Theorems 3.1 and 4.4, we obtain the following sufficient condition for a face of to also be a face of :
Theorem 4.5. is a face of both and if and only if there exist such that
Results by Yopp and Hill [2] finish the known material of this section.
Theorem 4.6. Let . Then, if and only if there exist linearly independent matrices such that .
Theorem 4.7. Let . If , then for every .
Proof. By Corollary 2.10 of [9, page 222], if and only if there exists such that . Therefore, for every , . It follows that for every .
The material of this paper leaves an open problem that we first state in generality and then in the setting of the paper.
Conjecture 4.8. Let , , and be cones such that . If and , then .
Conjecture 4.9. Let be a proper subcone of . If and , then .
Acknowledgment
The material of this paper forms a part of the Doctor of Arts thesis written by Klimas under the direction of Hill at Idaho State University.