First Deployment of a Novel Imaging Correlator for Radio Astronomy with the Long Wavelength Array

天文无线电干涉术是指将来自多个射电望远镜的信号组合在一起，形成单一的相干图像。计算机硬件的进步使日益大规模的干涉仪成为可能。这些仪器将探测早期宇宙，并实时探测射电天空。该项目将实施一种新的算法，以在大型阵列上实现计算效率的干涉测量。理论上证明了该算法的有效性。Thyagarajan和合作者将在真实数据上进行演示。他们将使用新墨西哥州拉塞维莱塔的长波长阵列。他们将在专门的硬件中对这一算法进行编码，以实现对天文无线电瞬变的调查。除了对天文学有益外，这种算法在使用干涉测量法的任何地方都可能得到应用。应用包括雷达、用于海洋、国家安全和医学的声学干涉测量。射电天文学的未来进步将由高性能、低成本的数字信号处理驱动的超大干涉仪驱动。走向大规模干涉测量需要重新考虑干涉仪的相关器。相关器相当于光学系统中的摄像机。它将来自成对天线组合的信号互相关联。它的计算需求为O(N^2)，其中N是天线的数量。它主导着大型阵列的成本和功耗。该项目将实现模优化频率傅里叶(MOF)相关器的概念，其中在孔径平面上的空间快速傅立叶变换将计算复杂度降低到O(N LogN)。这种潜在的变革性进步使更大的阵列和对快速瞬变的研究成为可能。该项目将实现该算法并使用来自La Sevilleta的长波长阵列的真实数据进行演示，部署基于图形处理单元(GPU)的实现，并使用该方法来演示瞬时搜索流水线。该项目将证明未来大规模干涉测量项目的可行性，如HERA第三阶段。

Astronomical radio interferometry means combining signals from multiple radio telescopes to form a single coherent image. Advances in computer hardware enable increasingly large-scale interferometers. Such instruments will probe the early Universe and explore the radio sky in real-time. This project will implement a new algorithm to enable computationally efficient interferometry on large arrays. The algorithm has been shown to be effective in theory. Thyagarajan and collaborators will demonstrate it on real data. They will use the Long Wavelength Array in La Sevilleta, New Mexico. They will encode this algorithm in specialized hardware to enable a survey of astronomical radio transients. In addition to benefitting astronomy, this algorithm may find application wherever interferometry is used. Applications include radar, acoustic interferometry used in oceanography, national security applications and medicine.Coming advances in radio astronomy will be enabled by extremely large interferometers driven by high-performance, low-cost digital signal processing. Progress toward large scale interferometry requires rethinking the correlator of an interferometer. The correlator is the radio equivalent of a camera in an optical system. It cross-correlates signals from pairwise combinations of antennae. Its computational requirements scale as O(N^2), where N is the number of antennae. It dominates the cost and power consumption of large arrays. This project will implement the Modular Optimal Frequency Fourier (MOFF) correlator concept, in which spatial Fast Fourier Transforms in the aperture plane reduce the computational complexity to O(N log N). This potentially transformative advance enables larger arrays and the study of fast transients. This project will implement the algorithm and demonstrate it on real-world data from the Long Wavelength Array at La Sevilleta, deploy a Graphical Processing Unit (GPU) based implementation, and use this method to demonstrate a transient search pipeline. This project will demonstrate feasibility for future large scale interferometry projects such as HERA phase III.