Multiplicativity for polarized Werner–Holevo channels
- Field
- Topics
Problem
For every integer \(d\geq3\), every \(x\in(0,1)\), and every \(1<p<2\), is the maximal output Schatten \(p\)-norm of the polarized Werner–Holevo channel multiplicative on two identical copies? Define the Werner–Holevo channel and its polarized interpolation with the identity channel by
where the transpose in Eq. (1) is taken in a fixed basis. For a channel \(\Phi\) with \(d\)-dimensional input, set
With the convention in Eq. (2), determine whether
holds throughout the stated parameter range.
Source
Ruskai explicitly asked for Eq. (3) for \(x\in[0,1]\) and \(1\leq p\leq2\) [Rus07]. The formulation above removes all regimes settled by the results below.
Progress
For \(d=2\), the channel in Eq. (1) is a unital qubit channel, so King’s theorem gives multiplicativity with an arbitrary companion for every \(p\geq1\). The cases \(p=1\) and \(x=1\) are also immediate from trace preservation and the identity channel, respectively [Kin02].
At \(x=0\), Datta proved multiplicativity for two Werner–Holevo channels of arbitrary dimensions throughout \(1\leq p\leq2\). This settles the unpolarized endpoint but not any \(x\in(0,1)\) [Dat04].
Michalakis proved Eq. (3) at \(p=2\) for every \(d\geq2\) and every \(x\in[0,1]\). The proof is specific to the output \(2\)-norm and leaves \(1<p<2\) open [Mic07].
Comment
The only unresolved regime of the source problem is precisely \(d\geq3\), \(x\in(0,1)\), and \(1<p<2\), with two identical copies as in Eq. (3).