TYC LUNCHTIME GET-TOGETHER

TYC@UCL Lunchtime Get-Togethers consist of two short presentations from either students, visitors or academics talking place over lunch and drinks. They are a great opportunity to get to meet other students, catch up with your friends and colleagues, and listen to members and visitors talk about their research. 

Food and drink will be provided so please just bring yourselves!

These events happen on the first Friday of the month and we particularly welcome all those people who've not been to a TYC event before

This month's talks are from:

Linear-scaling implementation of exact exchange using numerical atomic orbitals and three-center reduction integrals

Lionel Truflandier

The exchange interaction is a fundamental concept in the description and the interpretation of electronic structure of molecules and solids. Its genuine representation is found in the so-called "Fock" exchange for Hartree-Fock (HF) and correlated wave-function approaches, and "exact" exchange for Kohn-Sham (KS) density functional theory (DFT). Unfortunately, the expensive computational resources needed for the evaluation of the 4-center two-electron repulsion integrals (ERIs) represents a serious bottleneck. Within the framework of O(N) density-matrix based methodologies and numerical atomic orbitals, we will introduce an effcient and simple scheme which allows
computation of exact exchange without explicit evaluation of the ERIs. This defines a formal scaling about N3 for the pre-factor, which reaches asymptotically O(N2). Besides the methodological aspects, the accuracy and the reliability of the new algorithm will be demonstrated. The principles
of the parallelization within the linear-scaling code CONQUEST will be briefly described. Finally, generalization of the formalism to the plane-wave based DFT as found in many periodic electronic structure codes will be discussed.


What makes a good ice former?

Steve Cox

The freezing of water is an everyday phenomenon yet the
physical processes that determine the nucleation rates of water are
still not fully understood. In the absence of impurities, water may
exist in a supercooled state down to approximately -40 degree C but in the
presence of a surface, the freezing temperature may be significantly
higher. It has been seen experimentally that different surfaces can
exhibit differing nucleating abilities, for example, the minerals
kaolinite and muscovite are found to freeze water much more
efficiently than quartz or calcite.The reasons why one surface
may be a more effective ice forming nucleus (IN) than another are
not completely known, though it has been suggested that the presence
of active sites (e.g. dislocations, steps) and a good lattice match
between the IN and bulk ice play an important role. Here we
shall discuss our findings from work using Grand Canonical Monte Carlo
at a model hexagonal surface, that suggest the concept of lattice
match in ice formation is complicated by the formation of
non-hexagonal arrangements.