arXiv subjects

Wesley Armour

Publications and source records attributed to Wesley Armour.

33 records · Page 2Linked to original sources

arXiv2021

Implementation of 3D degridding algorithm on the NVIDIA GPUs using CUDA

Practical aperture synthesis imaging algorithms work by iterating between estimating the sky brightness distribution and a comparison of a prediction based on this estimate with the measured data ("visibilities"). Accuracy in the latter step is crucial but is made difficult by irregular and non-planar sampling of data by the telescope. In this work we present a GPU implementation of 3d de-gridding which accurately deals with these two difficulties and is designed for distributed operation. We address the load balancing issues caused by large variation in visibilities that need to be computed. Using CUDA and NVidia GPUs we measure performance up to 1.2 billion visibilities per second.

astro-ph.IM↗

arXiv2020

Single Pulse Detection Algorithms for Real-time Fast Radio Burst Searches using GPUs

The detection of non-repeating or irregular events in time-domain radio astronomy has gained importance over the last decade due to the discovery of fast radio bursts. Existing or upcoming radio telescopes are gathering more and more data and consequently the software, which is an important part of these telescopes, must process large data volumes at high data rates. Data has to be searched through to detect new and interesting events, often in real-time. These requirements necessitate new and fast algorithms which must process data quickly and accurately. In this work we present new algorithms for single pulse detection using boxcar filters. We have quantified the signal loss introduced by single pulse detection algorithms which use boxcar filters and based on these results, we have designed two distinct "lossy" algorithms. Our lossy algorithms use an incomplete set of boxcar filters to accelerate detection at the expense of a small reduction in detected signal power. We present formulae for signal loss, descriptions of our algorithms and their parallel implementation on NVIDIA GPUs using CUDA. We also present tests of correctness, tests on artificial data and the performance achieved. Our implementation can process SKA-MID-like data 266$\times$ faster than real-time on a NVIDIA P100 GPU and 500x faster than real-time on a NVIDIA Titan V GPU with a mean signal power loss of 7%. We conclude with prospects for single pulse detection for beyond SKA era, nanosecond time resolution radio astronomy.

astro-ph.IM↗

arXiv2020

GPU Fast Convolution via the Overlap-and-Save Method in Shared Memory

We present an implementation of the overlap-and-save method, a method for the convolution of very long signals with short response functions, which is tailored to GPUs. We have implemented several FFT algorithms (using the CUDA programming language) which exploit GPU shared memory, allowing for GPU accelerated convolution. We compare our implementation with an implementation of the overlap-and-save algorithm utilizing the NVIDIA FFT library (cuFFT). We demonstrate that by using a shared memory based FFT we can achieved significant speed-ups for certain problem sizes and lower the memory requirements of the overlap-and-save method on GPUs.

cs.MS↗

arXiv2020

Development of production-ready GPU data processing pipeline software for AstroAccelerate

Upcoming large scale telescope projects such as the Square Kilometre Array (SKA) will see high data rates and large data volumes; requiring tools that can analyse telescope event data quickly and accurately. In modern radio telescopes, analysis software forms a core part of the data read out, and long-term software stability and maintainability are essential. AstroAccelerate is a many core accelerated software package that uses NVIDIA(R) GPUs to perform realtime analysis of radio telescope data, and it has been shown to be substantially faster than realtime at processing simulated SKA-like data. AstroAccelerate contains optimised GPU implementations of signal processing tools used in radio astronomy including dedispersion, Fourier domain acceleration search, single pulse detection, and others. This article describes the transformation of AstroAccelerate from a C-like prototype code to a production-ready software library with a C++ API and a Python interface; while preserving compatibility with legacy software that is implemented in C. The design of the software library interfaces, refactoring aspects, and coding techniques are discussed.

astro-ph.IM↗

arXiv2019

Searching for pulsars in extreme orbits -- GPU acceleration of the Fourier domain 'jerk' search

Binary pulsars are an important target for radio surveys because they present a natural laboratory for a wide range of astrophysics for example testing general relativity, including detection of gravitational waves. The orbital motion of a pulsar which is locked in a binary system causes a frequency shift (a Doppler shift) in their normally very periodic pulse emissions. These shifts cause a reduction in the sensitivity of traditional periodicity searches. To correct this smearing Ransom [2001], Ransom et al. [2002] developed the Fourier domain acceleration search (FDAS) which uses a matched filtering technique. This method is however limited to a constant pulsar acceleration. Therefore, Andersen and Ransom [2018] broadened the Fourier domain acceleration search to account also for a linear change in the acceleration by implementing the Fourier domain "jerk" search into the PRESTO software package. This extension increases the number of matched filters used significantly. We have implemented the Fourier domain "jerk" search (JERK) on GPUs using CUDA. We have achieved 90x performance increase when compared to the parallel implementation of JERK in PRESTO. This work is part of the AstroAccelerate project Armour et al. [2019], a many-core accelerated time-domain signal processing library for radio astronomy.

astro-ph.IM↗

arXiv2018

A GPU implementation of the harmonic sum algorithm

Time-domain radio astronomy utilizes a harmonic sum algorithm as part of the Fourier domain periodicity search, this type of search is used to discover single pulsars. The harmonic sum algorithm is also used as part of the Fourier domain acceleration search which aims to discover pulsars that are locked in orbit around another pulsar or compact object. However porting the harmonic sum to many-core architectures like GPUs is not a straightforward task. The main problem that must be overcome is the very unfavourable memory access pattern, which gets worse as the dimensionality of the harmonic sum increases. We present a set of algorithms for calculating the harmonic sum that are more suited to many-core architectures such as GPUs. We present an evaluation of the sensitivity of these different approaches, and their performance. This work forms part of the AstroAccelerate project which is a GPU accelerated software package for processing time-domain radio astronomy data.

astro-ph.IM↗

arXiv2018

A GPU implementation of the Correlation Technique for Real-time Fourier Domain Pulsar Acceleration Searches

The study of binary pulsars enables tests of general relativity. Orbital motion in binary systems causes the apparent pulsar spin frequency to drift, reducing the sensitivity of periodicity searches. Acceleration searches are methods that account for the effect of orbital acceleration. Existing methods are currently computationally expensive, and the vast amount of data that will be produced by next generation instruments such as the Square Kilometre Array (SKA) necessitates real-time acceleration searches, which in turn requires the use of High Performance Computing (HPC) platforms. We present our implementation of the Correlation Technique for the Fourier Domain Acceleration Search (FDAS) algorithm on Graphics Processor Units (GPUs). The correlation technique is applied as a convolution with multiple Finite Impulse Response filters in the Fourier domain. Two approaches are compared: the first uses the NVIDIA cuFFT library for applying Fast Fourier Transforms (FFTs) on the GPU, and the second contains a custom FFT implementation in GPU shared memory. We find that the FFT shared memory implementation performs between 1.5 and 3.2 times faster than our cuFFT-based application for smaller but sufficient filter sizes. It is also 4 to 6 times faster than the existing GPU and OpenMP implementations of FDAS. This work is part of the AstroAccelerate project, a many-core accelerated time-domain signal processing library for radio astronomy.

astro-ph.IM↗

arXiv2017

Improved Acceleration of the GPU Fourier Domain Acceleration Search Algorithm

We present an improvement of our implementation of the Correlation Technique for the Fourier Domain Acceleration Search (FDAS) algorithm on Graphics Processor Units (GPUs) (Dimoudi & Armour 2015; Dimoudi et al. 2017). Our new improved convolution code which uses our custom GPU FFT code is between 2.5 and 3.9 times faster the than our cuFFT-based implementation (on an NVIDIA P100) and allows for a wider range of filter sizes then our previous version. By using this new version of our convolution code in FDAS we have achieved 44% performance increase over our previous best implementation. It is also approximately 8 times faster than the existing PRESTO GPU implementation of FDAS (Luo 2013). This work is part of the AstroAccelerate project (Armour et al. 2002), a many-core accelerated time-domain signal processing library for radio astronomy.

astro-ph.IM↗

arXiv2016

A Real-time Single Pulse Detection Algorithm for GPUs

The detection of non-repeating events in the radio spectrum has become an important area of study in radio astronomy over the last decade due to the discovery of fast radio bursts (FRBs). We have implemented a single pulse detection algorithm, for NVIDIA GPUs, which use boxcar filters of varying widths. Our code performs the calculation of standard deviation, matched filtering by using boxcar filters and thresholding based on the signal-to-noise ratio. We present our parallel implementation of our single pulse detection algorithm. Our GPU algorithm is approximately 17x faster than our current CPU OpenMP code (NVIDIA Titan XP vs Intel E5-2650v3). This code is part of the AstroAccelerate project which is a many-core accelerated time-domain signal processing code for radio astronomy. This work allows our AstroAccelerate code to perform a single pulse search on SKA-like data 4.3x faster than real-time.

astro-ph.IM↗

arXiv2015

Pulsar Acceleration Searches on the GPU for the Square Kilometre Array

Pulsar acceleration searches are methods for recovering signals from radio telescopes, that may otherwise be lost due to the effect of orbital acceleration in binary systems. The vast amount of data that will be produced by next generation instruments such as the Square Kilometre Array (SKA) necessitates real-time acceleration searches, which in turn requires the use of HPC platforms. We present our implementation of the Fourier Domain Acceleration Search (FDAS) algorithm on Graphics Processor Units (GPUs) in the context of the SKA, as part of the Astro-Accelerate real-time data processing library, currently under development at the Oxford e-Research Centre (OeRC), University of Oxford.

astro-ph.IM↗

arXiv2012

Observations of transients and pulsars with LOFAR international stations and the ARTEMIS backend

The LOw Frequency ARray - LOFAR - is a new radio interferometer designed with emphasis on flexible digital hardware instead of mechanical solutions. The array elements, so-called stations, are located in the Netherlands and in neighbouring countries. The design of LOFAR allows independent use of its international stations, which, coupled with a dedicated backend, makes them very powerful telescopes in their own right. This backend is called the Advanced Radio Transient Event Monitor and Identification System (ARTEMIS). It is a combined software/hardware solution for both targeted observations and real-time searches for millisecond radio transients which uses Graphical Processing Unit (GPU) technology to remove interstellar dispersion and detect millisecond radio bursts from astronomical sources in real-time.

astro-ph.IM↗

arXiv2011

Monte Carlo simulation of monolayer graphene at non-zero temperature

We present results from lattice simulations of a monolayer graphene model at non-zero temperature. At low temperatures for sufficiently strong coupling the model develops an excitonic condensate of particle-hole pairs corresponding to an insulating phase. The Berezinskii-Kosterlitz-Thouless phase transition temperature is associated with the value of the coupling where the critical exponent delta governing the response of the order parameter at criticality to an external source has a value close to 15. The critical coupling on a lattice with temporal extent N_t=32 (T=1/(N_t a_t) where a_t is the temporal lattice spacing) and spatial extent N_s=64 is very close to infinite coupling. The value of the transition temperature normalized with the zero temperature fermion mass gap Delta_0 is given by T_BKT/Delta_0=0.055(2). This value provides an upper bound on the transition temperature, because simulations closer to the continuum limit where the full U(4) symmetry is restored may result in an even lower value. In addition, we measured the helicity modulus Upsilon and the fermion thermal mass Delta_T(T), the later providing evidence for a pseudogap phase with Delta_T>0 extending to arbitrarily high T.

cond-mat.str-el↗

arXiv2011

Magnetic monopole plasma phase in (2+1)d compact quantum electrodynamics with fermionic matter

We present the first evidence from lattice simulations that the magnetic monopoles in three dimensional compact quantum electrodynamics (cQED3) with N_f=2 and N_f= 4 four-component fermion flavors are in a plasma phase. The evidence is based mainly on the divergence of the monopole susceptibility (polarizability) with the lattice size at weak gauge couplings. A weak four-Fermi term added to the cQED3 action enabled simulations with massless fermions. The exact chiral symmetry of the interaction terms forbids symmetry breaking lattice discretization counterterms to appear in the theory's effective action. It is also shown that the scenario of a monopole plasma does not depend on the strength of the four-Fermi coupling. Other observables such as the densities of "isolated" dipoles and monopoles and the so-called specific heat show that a crossover from a dense monopole plasma to a dilute monopole gas occurs at strong couplings. The implications of our results on the stability of U(1) spin liquids in two spatial dimensions are also discussed.

hep-lat↗

arXiv2010

Chiral symmetry breaking and monopole dynamics in non-compact QED3 coupled to a four-fermi interaction

We present results from the first lattice simulations of three dimensional non-compact quantum electrodynamics (QED3) with N_f four-component fermion flavors coupled to a weak Z(2) chirally invariant four-fermi interaction. Results with $N_f \ge 4$ show that the scaling near the strong coupling chiral transition or sharp crossover is determined by the 3d Gross Neveu ultraviolet-stable renormalization group fixed point. Small deviations of the N_f=4 critical exponents from the respective Gross Neveu ones, hint at evidence for non-zero fermion mass generated by the gauge fields dynamics that might have been enhanced by the four-fermi coupling. It is also shown that the scaling region is suppressed at weak four-fermi couplings and large N_f values. Measurements of i) a monopole susceptibility which is the polarizability of the monopole configurations, and ii) the density of isolated monopoles, imply that for $N_f \ge 1$ and weak gauge couplings the monopoles do not affect the theory's confining properties, because they are shielded.

hep-lat↗

arXiv2009

Lattice Simulations near the Semimetal-Insulator Phase Transition of Graphene

We present results from Monte Carlo simulations of a three dimensional fermionic field theory which can be derived from a model of graphene in which electrons interact via a screened Coulomb potential. For our simulations we employ lattice gauge theory methods used in elementary particle physics. We show that the theory undergoes a second order phase transition and we provide estimates for the critical exponents. The estimated value of the physical critical coupling implies that graphene in vacuum is an insulator. We also present the first results for the quasiparticle dispersion relation.

cond-mat.str-el↗