Xiulin Ruan Purdue University Thermal transport properties are key issues limiting many important energy transfer and conversion applications such as thermal management of electronic devices, thermoelectrics, solar thermal and photovotaics. It is desired to be able to predict such properties from their atomic structures. In this talk I will give examples of predicting thermal conductive and radiative properties of solids from their atomic structures, through multiscale multiphysics simulation tools that bridge first-principles calculations, molecular dynamics, and finite difference/element methods. For solar thermal and photovoltaic applications using CNT and Si nanowire arrays, we predict the dielectric function using first-principles methods together with Fermi’s Golden Rule. The results are then implemented into finite difference time domain (FDTD) calculations to demonstrate the extremely high optical absorption, which is desirable for solar applications. For thermoelectric applications, we use ab initio calculations to develop empirical interatomic potentials for heavy metal compounds including Bi2Te3 and PbTe, and then we perform classical molecular dynamics simulations to predict thermal conductivities of bulk, nanowires, and few-quintuple thin films. Using a spectral energy density approach, the spectral phonon lifetime and mean-free-path are derived which provide very useful insights of thermal conductivity size effects in these nanomaterials.