Karel Adámek is a Post Doctoral Research Associate working in radio astronomy software development with Dr Wes Armour. Karel Adámek finished his PhD study of Theoretical Physics and Astrophysics at Silesian University in Opava, Czech Republic in year 2016. His research interests include parallel algorithms in signal processing with emphasis on GPU implementation. He is also interested in numerical simulation of processes around compact objects. Current projects relate to Astro-Accelerate and the Square Kilometre Array.
Single pulse detection
Single pulse detection algorithm is used to detect pulses in de-despersed time series. It is used as a stand alone algorithm for detection fast radio bursts (FRBs) or other isolated events. It could be also used in conjunction with periodicity search. The conventional pulse detection algorithm involves using matched filter to identify pulses in data. We are using a boxcar filter as our matched filter.
Shared memory Fast Fourier transform on GPUs
Fast Fourier transformation (FFT) is well known algorithm and it is commonly available in libraries for both CPU and GPU. Fourier domain acceleration search (FDAS) utilizes convolutions in its search for accelerated pulsars. In order to perform these convolutions in shared memory custom FFT is needed. I have implemented FFT for Nvidia GPUs which could be used inside a CUDA threadblock and works in shared memory. The advantage of this approach is that one does not require to use slower global memory for FFT calculations.
Code on GitHub is here.
The polyphase filter is a standard tool in digital signal processing and as such a well established algorithm. We have implemented the polyphase filter for real-time data processing in radio astronomy on three generations of NVIDIA GPU cards (Fermi, Kepler, Maxwell), on the Intel Xeon CPU and Xeon Phi (Knights Corner) platforms. The polyphase filter (PPF) is a two step algorithm. First step is to apply a linear filter and second step performs fast Fourier transformation (FFT). We have focused on the first step and for FFT algorithm we have used external libraries specific to the platform used.
G. Török, K. Goluchová, M. Urbanec, E. Šrámková, K. Adámek, G. Urbancová, T. Pecháček, P. Bakala, Z. Stuchlík, J. Horák; Constraining Models of Twin-peak Quasi-periodic Oscillations with Realistic Neutron Star Equations of State, The Astrophysical Journal, Volume 833, Number 2, Published 2016 December 21 2016 (http://adsabs.harvard.edu/abs/2016ApJ...833..273T)
K. Adámek, J. Novotný, W. Armour; A polyphase filter for many-core architectures: Astronomy and Computing; Volume 16, July 2016, Pages 1–16 (http://dx.doi.org/10.1016/j.ascom.2016.03.003)
G. Török, M. Urbanec, K. Adámek, G. Urbancová; Appearance of innermost stable circular orbits of accretion discs around rotating neutron stars: Astronomy & Astrophysics, 2014 (http://adsabs.harvard.edu/abs/2014A%26A...564L...5T)
K. Adámek, Z. Stuchlík; Magnetized tori in the field of Kerr superspinars: Classical and Quantum Gravity, 2013 (http://adsabs.harvard.edu/abs/2013CQGra..30t5007A)
- ADASS XXVI (2016) - A real-time Single Pulse detection algorithm for GPUs (http://www.adass2016.inaf.it/images/posters/Adamek_P3.1.pdf)
- GTC 2016 - Implementation of a Real-Time Polyphase Filter in Radio Astronomy, selected for GTC16 Poster Fast Forward (http://on-demand.gputechconf.com/gtc/2016/posters/GTC_2016_Astronomy_and_Astrophysics_AA_04_P6281_WEB.pdf)
- OeRC, Oxford University, Oxford e-Research Centre, Oxford 2014: The Implementation of a real-time polyphase filter bank
- Perspectives of GPU Computing in Physics and Astrophysics, Rome 2014: The Implementation and Comparison of a Polyphase Filter on Many-Core Systems
• PhD in Theoretical Physics and Astrophysics at Silesian University in Opava (thesis title: Disc structures in the field of compact objects) (2008-2016)
• Research assistant at Silesian University in Opava (2010-2015)
• Cooperated with OeRC on polyphase filter on GPU - (2013-2015)
• Research intern at astrophysics, University of Oxford, as a part of Erasmus exchange program