Search

The analysis of scientific data, specially in different kinds of cosmological studies, has to deal with the increment in data volume. These studies include the calculation of correlation functions such as the Two-point Three-Dimensional Correlation Function. To get the final estimator value for these functions, it is necessary to construct histograms for storing large number counts. Histograms are a very common way of representing data and summarizing information in science. However, they have a high computational cost, which is worsened by the increase of the standard sample size. This increment leads directly to two problems: first of all, the large processing time and, secondly, the lack of accuracy of the result. Therefore, the implementations of correlation functions need to maintain high accuracy and affordable processing times. In order to reduce the high processing times, GPU computing is being widely used. In this work, the bin-recycling strategy is implemented and evaluated in the Two-Point Three-Dimensional Correlation Function. We show that this implementation outperforms others which also correctly process a large number of galaxies. As a result of this work, an accuracy-aware implementation of the Two-Point Three-Dimensional Correlation Function on GPU is described and evaluated to ensure the correctness of the results.

In this paper, several versions of a signal extraction algorithm, pertaining to the entry stage of the Cherenkov Telescope Array's Real Time Analysis pipeline, were implemented and optimised using SSE2, POSIX threads and CUDA. Results of this proof of concept let us gain an insight into the suitability of each platform, and the performance each one can deliver, to carry out this particular task.

Cosmology is a data intensive field that will become more so in the coming years. In this scenario, GPU-computing is becoming very relevant for analysing large catalogs within a reasonable amount of time. In this work, we investigate the acceleration of the analysis by using a hybrid MPI-CUDA implementation.

The Cherenkov Telescope Array (CTA) observatory will be one of the biggest ground-based very-high-energy (VHE) gamma-ray observatory. CTA will achieve a factor of 10 improvement in sensitivity from some tens of GeV to beyond 100 TeV with respect to existing telescopes. The CTA observatory will be capable of issuing alerts on variable and transient sources to maximize the scientific return. To capture these phenomena during their evolution and for effective communication to the astrophysical community, speed is crucial. This requires a system with a reliable automated trigger that can issue alerts immediately upon detection of gamma-ray flares. This will be accomplished by means of a Real-Time Analysis (RTA) pipeline, a key system of the CTA observatory. The latency and sensitivity requirements of the alarm system impose a challenge because of the anticipated large data rate, between 0.5 and 8 GB/s. As a consequence, substantial efforts toward the optimization of high-throughput computing service are envisioned. For these reasons our working group has started the development of a prototype of the Real-Time Analysis pipeline. The main goals of this prototype are to test: (i) a set of frameworks and design patterns useful for the inter-process communication between software processes running on memory; (ii) the sustainability of the foreseen CTA data rate in terms of data throughput with different hardware (e.g. accelerators) and software configurations, (iii) the reuse of non-real-time algorithms or how much we need to simplify algorithms to be compliant with CTA requirements, (iv) interface issues between the different CTA systems. In this work we focus on goals (i) and (ii).

The fitting of experimental or observational data to specific functional forms requires high computational capacities in order to tackle the complexity of the calculations. This complexity makes compulsory the use of efficient search procedures such as evolutionary algorithms. Evolutionary algorithms have proved their capability to find suboptimal, high-quality solutions to problems with large search spaces. In this context, a particle swarm algorithm and differential evolution are used to fit a data set to a serial expansion of Legendre polynomials. Concerning the data set, 56 rotation curves of spiral galaxies are used to build up a serial expansion—physically meaningless—retaining the essential information of the curves. The ultimate goal of this work is twofold: first, to provide a theoretical functional form representing the features of the rotational curves of spiral galaxies in order to couple it to other computational models; and second, to demonstrate the applicability of evolutionary algorithms to the matching between astronomical data sets and theoretical models.