We image the internal structure of the Sun, stars and Earth using computational, theoretical and data-analysis techniques

Helioseismic tomography

We use techniques of seismology to image the internal structure and dynamics of the Sun, stars and Earth. We employ a range of computational and theoretical methods in order to interpret analyses of seismic observations. Over the past several years, a major effort has been focused on computationally realising the adjoint method, which we now use as a means of computing kernels around heterogeneous 3D models. Large seismic datasets are extracted from high-resolution observations of the Sun – typically resulting in high-dimensional inverse problems. Appropriately conditioning and parametrising the inverse problem so as to accurately recover internal structure is another topic of interest. We apply these ideas to the inference of meridional circulation and differential rotation.

Normal-mode coupling is another significant focus area of the group. We measure the degree of coupling between normal modes in order to infer non-axisymmetric and time-varying perturbations. We are interested in using this technique to investigate the properties of interior convection and inferring Lorentz stresses.

Machine learning

We train machines on solar observations to predict magnetic processes such as flares and field emergence. Our focus is not so much prediction as it is to appreciate the governing mechanism. Machines, which are successful in predicting aspects of magnetic behaviour which other contemporary analyses have been unable to capture, are of particular interest. We apply a variety of tests on such machines to try to extract the patterns they have learnt.

kredi veren bankalar
instagram takipçi satın al
bahisnow giriş
canlı casino
deneme bonusu

Comments are closed.