Monte Carlo Algorithm for Nanoscale Electro-Thermal Optimization
A.W. Skorek, S.V. Ble, G. Anders and J. Nazarko
University of Quebec at Trois-Rivieres, CA
electro-thermal, modeling, optimization, nanoscale, nanoelectronics, Monte Carlo, parallel computing
In this paper, we will present a Monte Carlo method based algorithm established for a sample nanoelectronics device. We are assuming that at the nanoscale level like at the highest the thermal phenomena are important in the conception of device. In addition, we will justify that the effect of a single electron can be important to the thermal phenomena linked to a nanodevice. To illustrate this importance we will consider the single electron transistor. Thus, we must be able to know the position of electrons for modeling thermal phenomena of such devices. For this, we will define the governing equation from which we will be able to do so. These equations will be the phonons equation and the Schrödinger waves equations. We are aware that we can’t know for sure the position of an electron, but these equations will give us the probability of their position. From this point, one of these equations will be choose for a particular application and for further deepening. Another aspect that we will consider is that the thermal analysis of a nanoscale device will take a lot of computations. So, this is why we will present how to use parallel algorithm to solve the chosen equation. After parallelization, we will show implementation of the equation’s parallel algorithm on a cluster of 4 and 256 computers and a strategy of computing distribution in the series of numerical experiments. The results will then be analyzed and will conclude the paper with some recommendations.
Back to Program
Nanotech 2007 Conference Program Abstract