Tiago Pereira

• Position:

  Post-doc

• A brief description of your research aims:

  My current research is focused on the studying the 3D effects on line formation in stellar atmospheres, in particular with regards with abundance determinations from 3D models. There are two kinds of calculations I'm performing: 3D non-LTE radiative transfer for a few lines, and high-resolution calculations of the full spectrum coming from 3D models (still under LTE).

• A brief description of the nature of the computations to be performed (resources required, interactive/batch job, i/o requirements, single processor/parallel jobs, scalability):

  The computations I will be running will be mostly batch jobs, in parallel using 8-32 cores. I'm hoping to use COALA for convergence testing and probing the parameter space. Thus I plan to do short runs (typically 1h-4h) that will be very helpful in setting up the runs on a larger supercomputer. The 3D NLTE code (Multi3D) is MPI-domain decomposed and at present resolutions the optimal number of cores is 32. The full spectrum code (ASSET) will scale linearly to thousands of cores, although I don't expect to run it with more than 32. In terms of disk-space, each run takes about 25 Gb. It would be most helpful if there could be some common storage in COALA with a low-latency access, so that I don't have to transfer all these files across the network (which is often very time-consuming). I would need perhaps 200 Gb for the testing of a few snapshots.