ICHEC Seminar 2007

Computational Nanoscience: Solving big problems for small objects on big computers

Prof. Stefano Sanvito, School of Physics and CRANN, Trinity College, Dublin 2, Ireland.

Abstract:

Density functional theory has revolutionarized our way to do materials science and it is now a fundamental asset for research in Physics, Chemistry, Biology and Nanoscience. This is mainly due to a combination of conceptual simplicity, rigorous theoretical foundation and efficient numerical algorithms. The Smeagol project has the ambitious goal of setting the same revolution in the field of ab initio quantum transport, i.e. of being able of predicting the properties of a device by just knowing the position of its constituent atoms.

Smeagol, in its present form, calculates the I-V characteristics of nanoscale two-probe devices without adjustable parameters and it is constructed with three main goals in mind. First it must be accurate. For this reason exchange and correlation potentials including strong correlation corrections (LDA+U or LDA+SIC) have been implemented and demonstrated effective for the transport. Secondly, it must be able to scale, and therefore capable of accessing massive parallel machines. Finally it must be reasonably user friendly to serve a large community.

In this talk I will review the basic ideas behind the Smeagol project and present one of our most interesting results. This is the investigation of the conducting state of DNA, a long-standing problem to both the nano- and bio-science communities. In particular I will demonstrate that short DNA strands behave as mid-gap semiconductors. In addition I will discuss how the measurement of a transverse current can be used for DNA sequencing at low cost and high yield, a protocol that, if experimentally successful will change the course of preventive medicine.