Comments on Hastings' additivity counterexamples (Q982455): Difference between revisions

The additivity problem is one of the most profound mathematical problems of quantum information theory. It concerns additivity, with respect to tensor products, of several entropic quantities related to the classical capacity of quantum channel. From an analytical point of view it is closely related to the multiplicative property of norms of maps on operator spaces equipped with the Schatten \(p\)-norms. The problem remained open for more than a dozen of years when in 2007 important findings of \textit{A. Winter} and \textit{P. Hayden} [Commun. Math. Phys. 284, No. 1, 263--280 (2008; Zbl 1201.94066)] appeared which showed existence of a pair of channels of random unitary channels breaking the multiplicativity for all values of the parameter \(p>1\). In 2008, basing on this progress, \textit{M. B. Hastings} [[H]: ``A counterexample to additivity of minimum output entropy'', Nature Physics 5, 255--257 (2009)] gave a sketch of argument for existence of channels breaking the basic additivity conjecture for the key quantity -- the minimal output entropy, corresponding to \(p=1\), in very high dimensions. The paper of Fukuda, King and Moser gives the ultimate mathematical proof of existence of a counterexample following the main points of [H]. In addition to careful elucidation and improvement of the detail of Hastings' argument, it presents concrete bounds for the minimal dimensions needed to obtain a counterexample following this method. The accurate estimates made by Fukuda, King and Moser reveal overwhelmingly high values of the dimensions in which this construction can lead to a break in additivity (in fact very tiny, of the order \(10^{-5}\)). While this does not exclude possibility of better estimates, based perhaps on a different (but yet unknown) approach, it casts doubt on finding concrete counterexamples by computer simulation of random unitary channels. It also leaves a possibility that in small and moderate dimensions the additivity could still hold generically, or its violation is so small that it could not be caught by numerical simulations. More recent work [\textit{G. Aubrun, S. Szarek} and \textit{E. Werner}, ``Hastings' additivity counterexample via Dvoretzky's theorem'', \url{arXiv:1003.4925}] points out on the deep relation of this proof to the Dvoretzky-Milman theorem in the asymptotic theory of finite-dimensional normed spaces.