The Square Kilometre Array (SKA)

Abstract:

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.

Abstract:

RF system-on-chip (RFSoC) devices provide the potential for implementing a complete radio astronomy receiver on a single board, but performance of the integrated analogue-to-digital converters is critical. We have evaluated the performance of the data converters in the Xilinx ZU28DR RFSoC, which are 12-bit, 8-fold interleaved converters with a maximum sample speed of 4.096 Giga-sample per second (GSPS). We measured the spurious-free dynamic range (SFDR), signal-to-noise and distortion (SINAD), effective number of bits (ENOB), intermodulation distortion (IMD) and cross-talk between adjacent channels over the bandwidth of 2.048 GHz. We both captured data for off-line analysis with floating-point arithmetic, and implemented a real-time integer arithmetic spectrometer on the RFSoC. The performance of the ADCs is sufficient for radio astronomy applications and close to the vendor specifications in most of the scenarios. We have carried out spectral integrations of up to 100 s and stability tests over tens of hours and find thermal noise-limited performance over these timescales.

Abstract:

The cosmic dipole anomaly -- the observation of a significant mismatch between the dipole observed in the matter distribution and that expected given the kinematic interpretation of the cosmic microwave background dipole -- poses a serious challenge to the Cosmological Principle upon which the standard model of cosmology rests. Measurements of the dipole in a given sample crucially depend on having control over other large-scale power so as to avoid biases, in particular those potentially caused by correlations among multipoles during fitting, and those by local source clustering. Currently, the most powerful catalogue that exhibits the cosmic dipole anomaly is the sample of 1.6 million mid-infrared quasars derived from CatWISE2020. We therefore analyse clustering properties of this catalogue by performing an inference analysis of large-scale multipoles in real space, and by computing its angular power spectrum on small scales to test for convergence with LCDM. After accounting for the known trend of the quasar number counts with ecliptic latitude, we find that any other large-scale power is consistent with noise, find no evidence for the presence of an octupole ( ) in the data, and quantify the clustering dipole's proportion to be marginal. Our results therefore reaffirm the anomalously high dipole in the distribution of quasars.