Course on General Relativistic Magnetohydrodynamics by Yosuke Mizuno

Yosuke Mizuno (Goethe University, Frankfurt) taught an advanced course on general relativistic magnetohydrodynamics on August 13-17 at our institute. General relativistic magnetohydrodynamics—or GRMHD—is an essential tool to model high-energy astrophysical phenomena such as accreting black holes and relativistic jets—precisely the type of phenomena that our group loves and cherishes. This course was very useful for everybody in the group. 

The slides are available on this website.

IMG_2194
Yosuke Mizuno lecturing about GRMHD at IAG USP. Credit: Rodrigo Nemmen.

His visit was supported by FAPESP grant 2013/10559-5.

Group will teach analysis of Fermi LAT data in dark matter school

On the final week of July, members of our research group (Fabio, Raniere & Rodrigo) will be teaching a lesson on the analysis of Fermi LAT gamma-ray observations in the School and Workshop on Dark Matter and Neutrino Detection at ICTP-SAIFR. In this hands-on activity, we will teach how to analyze gamma-ray data for a dwarf galaxy, do a simple estimate of the dark matter cross section and reproduce the analysis described in Ackermann et al. (2015).

It will be fun!

NVIDIA donation to the group

I am excited to announce that today we received a generous donation from NVIDIA through its GPU Grant Program, which will allow us to accelerate our science to another scale. We received a Quadro P6000 GPU, packing a powerful punch of 12 TFLOPS of FP32 processing power and 24 GB. This GPU will allow us to severely speed up the calculations of electromagnetic spectra from black holes that our group is developing. One estimate that always amazes me: this GPU is almost faster than the whole computer cluster—called Alphacrucis—that our institute hosts, which has 2300 CPUs (~20 TFLOPS) and was purchased in 2011.

Thanks to NVIDIA for the great gift, which will soon be put to good use. Let’s see those fans spin at maximum RPM! Science!

IMG_2003
Some of the grad students in our group with the new GPU donated by NVIDIA. From left to right: Fabio, Gustavo, Raniere and Ivan.
IMG_2005
Rodrigo happily holding the Quadro GPU

Group training on GPU programming w/ OpenACC

Last Friday we had a group meeting with João Paulo Navarro from NVIDIA. João Paulo gave an OpenACC tutorial to the group, demonstrating how easy it is to accelerate scientific codes on GPUs. With just two lines of code (yes, I said two lines), we made a partial differential equation solver run almost 10x faster on a GPU! It is truly impressive.

We have great plans ahead, with some group projects where we anticipate huge speedups using GPUs. Stay tuned!

IMG_1987 (1)
The Black Hole Group after the OpenACC training with J. P. Navarro (NVIDIA; sitting at the rightmost side).

New grad student in the group obtains FAPESP scholarship

Let’s congratulate Artur Vemado─the newest graduate student in the group─who got a prestigious FAPESP scholarship. Artur just graduated with an astronomy degree at USP. His project will consist of incorporating radiative cooling in the energy equation of hot accretion flows, in order to investigate state transitions in black hole binaries.

Postdoctoral position ICTP-SAIFR/USP to join our group

There is a FAPESP postdoctoral position opening for working in our group and collaborating with researchers at ICTP-SAIFR. This position is for a double appointment at USP and ICTP-SAIFR.

Review of applications will begin in January 2018 and will continue until all positions have been filled. Although there is no strict deadline, applications before December 15, 2017 are strongly recommended for positions to begin in 2018.

More information in the website.

 

Participação de membro do grupo no Jornal Band News

Ontem à noite, participei do Jornal Band News contando a importância da grande descoberta astronômica sobre a colisão das duas estrelas de nêutrons.

Foi um momento histórico, o início da era da astronomia dos multi-mensageiros: observação de ondas gravitacionais e luz vindos de uma mesma fonte astronômica! E a resposta ao mistério da origem dos elementos mais pesados que o ferro da tabela periódica.

Espero que tenha passado os principais aspectos da descoberta, ressaltando a participação de muitos astrônomos no Brasil em vários estados—RN, RJ, SP, SC e SE—e em várias instituições—INPE e UFRN (times do LIGO/VIRGO), e USP, ON, UFRJ, UFSE e UFSC (contrapartida eletromagnética).

Speeding up black hole radiative transfer with GPUs

Today we’ve had the first meeting of the group with the goal of accelerating radiative transfer calculations around black holes with graphical processing units (GPUs). The idea is that by porting the code to exploit the massive parallelism of GPUs, we will see a dramatic speedup (hopefully ~10-100x). We have an enthusiastic group of students!

We decided to proceed with NVIDIA CUDA rather than OpenCL due to the better documentation and more astrophysical examples available in CUDA to guide us.

We will try to have a beta version of the GPU-accelerated code by December, and a release candidate on May 2018. The main challenges–as usual with GPGPU–is writing the appropriate kernels and GPU memory management in order to reduce host-device data transfers as much as possible.

Stay tuned for exciting news on astrophysical high-performance computing by our group!

IMG_0965