Colloquium: Dr. Hendrik Müller | NRAO
In-Person PHYS 401
Location
Physics : 401
Date & Time
March 5, 2025, 11:00 am – 12:00 pm
Description
TITLE: "Towards an automatized, high precision and high performance radio camera”
ABSTRACT: For decades data processing in radio interferometry has been a highly involved manual procedure, mainly based on monolithic data manipulation and inverse modelling steps nested in multiple loops around each other. The development of pipeline heuristics rooted in this traditional framework has provided us with the ability to carry out surveys with reduced manual interaction. However, they are challenged by the needs of the next generation of radio telescopes (e.g. ngVLA, ALMA WSU, SKA, DSA2000) which will operate at much higher data rates and sensitivity, questioning various approximations done during the data analysis previously. In this talk, I will review recent proposals to solve the data processing problem for the next generation of radio interferometers. The talk will cover accelerations of the deconvolution (minor loop) by multiscale techniques, Bayesian algorithms, and artificial intelligence, high throughput computing solutions for gridding (major loop), alternatives to classical self-calibration, and recent proposals for automated hyperparameter selections and pipeline heuristics.
ABSTRACT: For decades data processing in radio interferometry has been a highly involved manual procedure, mainly based on monolithic data manipulation and inverse modelling steps nested in multiple loops around each other. The development of pipeline heuristics rooted in this traditional framework has provided us with the ability to carry out surveys with reduced manual interaction. However, they are challenged by the needs of the next generation of radio telescopes (e.g. ngVLA, ALMA WSU, SKA, DSA2000) which will operate at much higher data rates and sensitivity, questioning various approximations done during the data analysis previously. In this talk, I will review recent proposals to solve the data processing problem for the next generation of radio interferometers. The talk will cover accelerations of the deconvolution (minor loop) by multiscale techniques, Bayesian algorithms, and artificial intelligence, high throughput computing solutions for gridding (major loop), alternatives to classical self-calibration, and recent proposals for automated hyperparameter selections and pipeline heuristics.
