June 2014 summaries

Symmetry breaking in the high temperature superconductor CsFe2Se2, Jon Goff (Royal Holloway)

The new family of iron based superconductors AxFe2-ySe2 where A = K, Rb and Cs show some remarkable differences to the iron-pnictide high-Tc's. Superconductivity and magnetism appear to co-exist with antiferromagnetism with an unusually high ordering temperature, TN~559 K, and a large ordered moment of about 3.3 μB per Fe. This project aimed to expand our understanding of these materials by studying the lattice dynamics across a symmetry-breaking phase transition with momentum-resolved spectroscopy measurements performed at the Institut Laue-Langevin (ILL) and ab-initio calculations using the CASTEP code on ARCHER.

This project has enabled the explanation of a number of published experimental results. Notably, the symmetry breaking observed in Raman and inelastic neutron scattering (INS) is well described by our proposed ferrimagnetic phase. It has also enabled the assignment of modes from the ILL experiment. On the basis of our combined experimental and computational work we have been granted a further week of ILL beamtime to try to directly measure the ferrimagnetic moment with polarised neutrons. This project has significantly contributed to the PhD thesis of David Voneshen and provided his first exposure to national computing resources. Furthermore, the geometry optimisations have already been included in a recent publication, Phys. Rev. B 91, 144114 (2015).

Full-body scale 3D simulation of arterial haemodynamics, C. Alberto Figueroa (Kings College London)

The objective of our project is to examine the role of arterial vessel stiffness in both normal and diseased states. Increases in arterial stiffness have been clinically shown to be an indicator of risk of fatal cardiovascular events such as strokes and heart attacks. The risk associated with arterial stiffening is typically assessed indirectly through the speed of blood pressure waves travelling throughout the arterial network. By measuring the speed of these waves, referred to as the pulse wave velocity, the risk of cardiovascular events can be assessed. The faster the pulse wave velocity, the stiffer the arterial vessel and the greater the cardiovascular risk. Variation in the arterial stiffness are therefore reflected in increases in the pulse wave velocity, however the specific correlation between the degree of abnormal arterial stiffening and cardiovascular risk is not fully understood.

By using computational models we are able to quantitatively examine the relationship between variations in arterial stiffness throughout the arterial network and the corresponding pulse wave velocity. Using our finite element software we are able to simulate blood flow in a given arterial geometry, under a range of different arterial stiffness that represent healthy, elderly (increased stiffness in all blood vessels) and diseased (localised stiffening) states of the arterial network. Analysis of the resulting haemodynamics provides further insight into the relationship between arterial stiffness, pulse wave velocity and the resultant workload on the heart.

Based on medical images, we have built a model of the arterial system that includes the aorta and the main vessels of the head and neck, arms, and legs. Consisting of a network of approximately 100 vessel, the resulting model was discretised into a finite element mesh consisting of ~52 million tetrahedral elements. Using this mesh, simulations of pulsatile blood flow have been performed on 4512 processors in parallel - resulting in the largest 3D, patient specific, full body scale, simulation of blood flow performed in the world to date. These results are currently being processed and will be submitted for publication in our next paper, in which we intend to fully acknowledge the support of the ARCHER RAP.

Using ARCHER has also enable us to perform scaling tests on our code. Simulations performed on 3384, 4512 and 6144 processors have indicated that optimal performance on ARCHER was obtained with 4512 parallel processes. We are currently working with ARCHER staff to further improve our code in order to address larger and more complex simulation of blood flow in the near future.