Procuro aluno(a)s interessado(a)s em Iniciação Científica em computação de alto desempenho aplicada à astrofísica. O projeto consiste em acelerar ray tracing ao redor de buracos negros—a la Interestelar—usando GPUs.
Cursar ciência da computação, engenharia elétrica ou matemática aplicada
Conforto com a linguagem C
Conhecimentos básicos–ou forte interesse–em otimização e paralelização
Pontos extras: noções de programação em GPUs (OpenCL ou CUDA)
Basic idea: take observed black hole systems and treat them as engines. Measure what flows in (fuel) and what comes out (exhaust). Dividing the energy of what comes out by the amount of energy that flows in, we get an energy efficiency.
We measured this efficiency for 27 nearby supermassive black holes which produce jets (radio galaxies). Our analysis is based on recent Chandra observations and current ideas on accretion disks.
The results imply that black hole engines are producing more energy than flows in: efficiencies > 100%. This is completely opposite to the general idea that everything disappears inside black holes. Black holes are actually quite “green” (energy-efficient).
Does this violate energy conservation? No. The extra energy that comes out would be extracted from the spin energy of the black hole.
What constraints do these results put on black holes physics? In order to explain the data, black holes need to: (i) be rapidly rotating, (ii) be surrounded by strong magnetic fields (see footnote below). Such “dynamically-important” magnetic fields act as catalysts of the energy extraction.
The paper got some coverage in the brazilian press here and here.
ps. The technical name for “strong magnetic fields” is “magnetically arrested disk” (abbreviated as MAD). This is a fashionable idea among theorists that are simulating black hole accretion flows. My collaborator (Tchekhovskoy) is one of the theorists doing such fantastic GRMHD simulations.