ABSTRACTSTUDIES OF CHARGE NEUTRAL FCC LATTICE GAS WITH YUKAWA INTERACTION AND ACCELERATED CARTESIAN EXPANSION METHODByHe Huang In this thesis, I present the results of studies of the structural properties and phase transition of a charge neutral FCC Lattice Gas with Yukawa Interaction and discuss a novel fast calculation algorithm -- Accelerated Cartesian Expansion (ACE) method. In the first part of my thesis, I discuss the results of Monte Carlo simulations carried out to understand the finite temperature (phase transition) properties and the ground state structure of a Yukawa Lattice Gas (YLG) model. In this model the ions interact via the potential qiqjexp(-κr>ij)/rij where qi,j are the charges of the ions located at the lattice sites i and j with position vectors Ri and Rj; rij =Ri-Rj, ê is a measure of the range of the interaction and is called the screening parameter. This model approximates an interesting quaternary system of great current thermoelectric interest called LAST-m, AgSbPbmTem+2. I have also developed rapid calculation methods for the potential energy calculation in a lattice gas system with periodic boundary condition bases on the Ewald summation method and coded the algorithm to compute the energies in MC simulation. Some of the interesting results of the MC simulations are: (i) how the nature and strength of the phase transition depend on the range of interaction (Yukawa screening parameter ê) (ii) what is the degeneracy of the ground state for different values of the concentration of charges, and (iii) what is the nature of two-stage disordering transition seen for certain values of x. In addition, based on the analysis of the surface energy of different nano-clusters formed near the transition temperature, the solidification process and the rate of production of these nano-clusters have been studied. In the second part of my thesis, we have developed two methods for rapidly computing potentials of the form R-í. Both these methods are founded on addition theorems based on Taylor expansions. Taylor's series has a couple of inherent advantages: (i) it forms a natural framework for developing addition theorem based computational schemes for a range of potentials; (ii) only Cartesian tensors (or products of Cartesian quantities) are used as opposed to special functions. This makes creating a fast scheme possible for potential of the form R-í. Indeed, it is also possible to generalize the proposed methods to several potentials that are important in mathematical physics. An interesting consequence of the approach has been the demonstration of the equivalence of FMMs that are based on traceless Cartesian tensors to those based on spherical expansions for í = 1. Two methods are introduced; the first relies on exact translation of the origin of the multipole whereas the second relies on cascaded Taylor's approximations. Finally, we have shown the application of this methodology to computing Coulombic, Lennard-Jones, Yukawa potentials and etc. We have also demonstrated the efficacy of this scheme for other (non-integer) potential functions.
Read
