Fri, 16 Nov 2007
13:15
13:15
DH 1st floor SR
The impact of inside information on learning and trading behaviour under parameter uncertainty
Andrew Ng
(University of Oxford)
For atomistic (and related) material models, global minimization
gives the wrong qualitative behaviour; a theory of equilibrium
solutions needs to be defined in different terms. In this talk, a
process based on gradient flow evolutions is presented, to describe
local minimization for simple atomistic models based on the Lennard-
Jones potential. As an application, it is shown that an atomistic
gradient flow evolution converges to a gradient flow of a continuum
energy, as the spacing between the atoms tends to zero. In addition,
the convergence of the resulting equilibria is investigated, in the
case of both elastic deformation and fracture.