Kay, A.R. and Kay, B.S. (2001) Monotonicity with volume of entropy and of mean entropy for translationally invariant systems as consequences of strong subadditivity. Journal of Physics A: Mathematical and General, 34 (3). pp. 365-382. ISSN 0305-4470Full text not available from this repository.
We consider some questions concerning the monotonicity properties of entropy and of mean entropy for translationally invariant states on translationally invariant systems (classical lattice, quantum lattice and quantum continuous). By taking the property of strong subadditivity, which for quantum systems was proven rather late in the historical development, as one of four primary axioms (the other three being simply positivity, subadditivity and translational invariance), we are able to obtain results, some new, some proved in a new way, which appear to complement in an interesting way results proved around 30 years ago on limiting mean entropy and related questions. In particular, we prove that as the sizes of boxes in ν or ν increase in the sense of set inclusion, (1) their mean entropy decreases monotonically and (2) their entropy increases monotonically. We include a proof of (2) based on the notion of m-point correlation entropies, which we introduce and which generalize the notion of index of correlation (see e.g. Horodecki R 1994 Phys. Lett. A 187 145). We mention a number of further results and questions concerning monotonicity of entropy and of mean entropy for more general shapes than boxes and for more general translationally invariant (/homogeneous) lattices and spaces than ν or ν. We also obtain some further results on monotonicity of entropy in these more general situations by adjoining a fifth axiom, which embodies yet another general property of entropy (which we call the `strong triangle inequality').
|Academic Units:||The University of York > Mathematics (York)|
|Depositing User:||York RAE Import|
|Date Deposited:||17 Apr 2009 09:27|
|Last Modified:||17 Apr 2009 09:27|
|Publisher:||Institute of Physics|
Actions (login required)