Wednesday, April 4, 2012

1204.0741 (Matthias Christandl et al.)

Eigenvalue Distributions of Reduced Density Matrices    [PDF]

Matthias Christandl, Brent Doran, Stavros Kousidis, Michael Walter
Given a random quantum state of multiple (distinguishable or indistinguishable) particles, we provide an algorithm, rooted in symplectic geometry, to compute the joint probability distribution of the eigenvalues of its one-body reduced density matrices, and hence some associated physical invariants of the state. As a corollary, by taking the support of this probability distribution, which is a convex polytope, we recover a complete solution to the one-body quantum marginal problem, i.e., the problem of characterizing the one-body reduced density matrices that arise from some multi-particle quantum state. In the fermionic instance of the problem, which is known as the one-body N-representability problem, the famous Pauli principle amounts to one linear inequality in the description of the convex polytope. We obtain the probability distribution by reducing to computing the corresponding distribution of diagonal entries (i.e., to the quantitative version of a classical marginal problem), which is then determined algorithmically. This reduction applies more generally to symplectic geometry, relating invariant measures for a compact Lie group action to that for the maximal torus action; we state and prove our results in this more general symplectic setting. Our approach is in striking contrast to the existing solution to the computation of the supporting polytope by Klyachko and by Berenstein and Sjamaar, which made crucial use of non-Abelian features. In algebraic geometry, Duistermaat-Heckman measures correspond to the asymptotic distribution of multiplicities of irreducible representations in the associated coordinate ring. In the case of the one-body quantum marginal problem, these multiplicities include bounded height Kronecker and plethysm coefficients. A quantized version of the Abelianization procedure provides an efficient algorithm for their computation.
View original:

No comments:

Post a Comment