We define and study the properties of channels which are analogous to unital qubit channels in several ways. A full treatment can be given only when the dimension d = p m a prime power, in which case each of the d + 1 mutually unbiased bases (MUB) defines an axis. Along each axis the channel looks like a depolarizing channel, but the degree of depolarization depends on the axis. When d is not a prime power, some of our results still hold, particularly in the case of channels with one symmetry axis. We describe the convex structure of this class of channels and the subclass of entanglement breaking channels. We find new bound entangled states for d = 3.For these channels, we show that the multiplicativity conjecture for maximal output p-norm holds for p = 2. We also find channels with behavior not exhibited by unital qubit channels, including two pairs of orthogonal bases with equal output entropy in the absence of symmetry. This provides new numerical evidence for the additivity of minimal output entropy. * Partially supported by by the National Science Foundation under Grants DMS-0314228 and DMS-0604900 and by the National Security Agency and Advanced Research and Development Activity under Army Research Office contract number DAAD19-02-1-0065.
IntroductionThe results presented here are motivated by the desire to find channels for dimension d > 2 whose properties are similar to those of the unital qubit channels, particularly with respect to optimal output purity. A channel is described by a completely positive, trace-preserving (CPT) map. The channels we construct are similar to unital qubit channels in the sense that their effect on a density matrix can be defined in terms of multipliers of components along different "axes" defined in terms of mutually unbiased bases (MUB). When all multipliers are positive, these channels are very much like unital qubit channels with positive multipliers. However, when some of the multipliers are negative the new channels can exhibit behavior not encountered for unital qubit channels.For a fixed orthonormal basis B = {|ψ k }, the quantum-classical (QC) channelprojects a density matrix ρ onto the corresponding diagonal matrix in this basis. A convex combination J t J Ψ QC J (ρ) of QC channels in a collection of orthonormal bases B J = {|ψ J k } is also a channel; in fact, it is an entanglement breaking (EB) channel. We consider channels which are a linear combination of the identity map I(ρ) = ρ and a convex combination of QC channels whose bases are mutually unbiased, i.e., satisfySuch channels can be written in the form 3The first condition ensures that Φ is trace-preserving (TP), and the pair that it is completely positive (CP), as will be shown in Section 2.It is well-known that C d can have at most d + 1 MUB and that this is always possible when d = p m is a prime power. We are primarily interested in channels of the form (3) when such a full set of d + 1 MUB exist. In that case, it is natural to generalize the Bloch sphere representation so that a density matri...