Reduced density matrices and semidefinite optimization
Bastiaan
Braams
Deptartment of Mathematics, Courant Institute of Mathematical Sciences
Tuesday, June 17, 2003 15:00-16:00,
Admin. Bldg. Lecture Room F
Gaithersburg
Tuesday, June 17, 2003 13:00-14:00,
Room 4550
Boulder
Abstract:
Reduced density matrices play an important role in the search for
methods of electronic structure calculation that offer a systematic
route to better accuracy and also offer favorable computational
scaling properties for large systems. In particular, the 2-body
reduced density matrix (2-RDM) contains in it enough information to
express exactly (through a known linear functional) the complete
electron Hamiltonian as well as all other ground-state properties of
interest. However, the constructive use of the 2-RDM is hindered by
the problem of N-representability that was first clearly described and
studied by Coleman and by Garrod and Percus. We describe recent work
that uses semidefinite programming to solve the variational problem
for the 2-RDM subject to a subset of representability conditions, and
we discuss the quality of the resulting approximation.
Contact: G. B. McFadden
Note: Visitors from outside NIST must contact
Robin Bickel; (301) 975-3668;
at least 24 hours in advance.
|
