8 August 2016 Efficient receiver tuning using differential evolution strategies
Author Affiliations +
Abstract
Differential evolution (DE) is a powerful and computationally inexpensive optimization strategy that can be used to search an entire parameter space or to converge quickly on a solution. The Kilopixel Array Pathfinder Project (KAPPa) is a heterodyne receiver system delivering 5 GHz of instantaneous bandwidth in the tuning range of 645-695 GHz. The fully automated KAPPa receiver test system finds optimal receiver tuning using performance feedback and DE. We present an adaptation of DE for use in rapid receiver characterization. The KAPPa DE algorithm is written in Python 2.7 and is fully integrated with the KAPPa instrument control, data processing, and visualization code. KAPPa develops the technologies needed to realize heterodyne focal plane arrays containing ~1000 pixels. Finding optimal receiver tuning by investigating large parameter spaces is one of many challenges facing the characterization phase of KAPPa. This is a difficult task via by-hand techniques. Characterizing or tuning in an automated fashion without need for human intervention is desirable for future large scale arrays. While many optimization strategies exist, DE is ideal for time and performance constraints because it can be set to converge to a solution rapidly with minimal computational overhead. We discuss how DE is utilized in the KAPPa system and discuss its performance and look toward the future of ~1000 pixel array receivers and consider how the KAPPa DE system might be applied.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Caleb H. Wheeler, Caleb H. Wheeler, Trevor G. Toland, Trevor G. Toland, } "Efficient receiver tuning using differential evolution strategies", Proc. SPIE 9913, Software and Cyberinfrastructure for Astronomy IV, 99130Q (8 August 2016); doi: 10.1117/12.2231363; https://doi.org/10.1117/12.2231363
PROCEEDINGS
8 PAGES


SHARE
Back to Top