from 16:00 to 17:00
|Where||Seminar room 115, 1st floor, Rogowski building, Schinkelstr. 2|
|Add event to calendar||
The FHI-AIMS Code: Experiences from a Large Scientific Software Project
Theory Department, Fritz Haber Institut of the Max Planck Society
Electronic structure theory is "big business" today in computational materials and molecular science: Modern implementations and computational resources enable the parameter-free prediction of properties for many classes of molecules and solids, based only on fundamental, controlled approximations to Schroedinger's Equation. While the underlying math is well defined, the list of scientific "features" to be requested is almost endless, and hard to cover in one framework by simple, homegrown code. Instead, pre-constructed general purpose codes are now the norm in much of the academic and industrial research community.
In this talk, I describe the development of our own general-purpose implementation of electronic structure theory, the FHI-aims code . Begun from scratch in 2004, the code intends to bridge a number of requirements of the field - periodic vs. non-periodic system geometries, density functional theory vs. non-local and/or many-body formalisms, small vs. large systems and, not least, ability to utilize massively parallel computers effectively. I discuss the basic algorithmic choices made (the rationale for writing a new code), the development choices made, and current capabilities of the code for large surfaces and biomolecules.
 The Fritz Haber Institute ab initio molecular simulations (FHI-aims) package. More information