Context. Cosmological measurements require the calculation of nontrivial quantities over large datasets. The next generation of survey telescopes (such as DES, PanSTARRS, and LSST) will yield measurements of billions of galaxies. The scale of these datasets, and the nature of the calculations involved, make cosmological calculations ideal models for implementation on graphics processing units (GPUs).
Aims. We consider two cosmological calculations, the two-point angular correlation function and the aperture mass statistic, and aim to improve the calculation time by constructing code for calculating them on the GPU. Methods. Using CUDA, we implement the two algorithms on the GPU and compare the calculation speeds to comparable code run on the CPU.
Results. We obtain a code speed-up of between 10 – 180x faster, compared to performing the same calculation on the CPU. The code has been made publicly available.
Conclusions. GPUs are a useful tool for cosmological calculations, even for datasets the size of current surveys, allowing calculations to be made one or two orders of magnitude faster.
We have implemented code to perform calculations of the two point angular correlation function and the aperture mass statistic on the GPU. We have demonstrated that this implementation can reduce compute times for these calculations by factors of 100x-300x, depending on the amount of data to be processed. The code for making this calculation is publicly available from Github. And can be cloned by anyone who has git installed on their system. Along with the code, we have provided sample datasets and scripts to run and test your installation. Each package has its own README that details how to build and run the software. This software is licensed under the MIT License.
Faster compute speeds mean that a full MC-based calculation of the errors for the angular correlation function can reasonably be performed, without approximations or assumptions that are required to make the calculation reasonable for CPU codes.We intend to evaluate this in future work. The increasing size of astronomical dat sets will require a new approach to data analysis. We expect that the use of the GPU in everyday cosmological calculations will become more common in the next few years, especially since faster compute times allows experimentation in techniques used to make the calculation
and rapid comparison of the results. We expect this application to be extended to other computationally challenging calculations, such as the three-point and higher order angular correlation functions, and the shear correlation functions.
Category: Physical Science