Presentation Title

Hydrogen Epoch of Reionization (HERA) Calibrated FFT Correlator Simulation

Faculty Mentor

Aaron Parsons

Start Date

18-11-2017 11:00 AM

End Date

18-11-2017 11:15 AM

Location

9-279

Session

Physical Sciences 1

Type of Presentation

Oral Talk

Subject Area

physical_mathematical_sciences

Abstract

The Hydrogen Epoch of Reionization Array (HERA) project is an astronomical radio interferometer array with a redundant baseline configuration. Interferometer arrays are being used widely in radio astronomy because they have a variety of advantages over single antenna systems. For example, they produce images (visibilities) closely matching that of a large antenna (such as the Arecibo observatory), while both the hardware and maintenance costs are significantly lower. However, this method has some complications; one being the computational cost of correlating data from all of the antennas. A correlator is an electronic device that cross-correlates the data between the individual antennas; these are what radio astronomers call visibilities. HERA, being in its early stages, utilizes a traditional correlator system. The correlator cost scales as N2, where N is the number of antennas in the array. The purpose of a redundant baseline configuration array setup is for the use of a more efficient Fast Fourier Transform (FFT) correlator. FFT correlators scale as Nlog2N. The data acquired from this sort of setup, however, inherits geometric delay and uncalibrated antenna gains. This particular project simulates the process of calibrating signals from astronomical sources. Each signal “received” by an antenna in the simulation is given random antenna gain and geometric delay. The “linsolve” Python module was used to solve for the unknown variables in the simulation (complex gains and delays), which then gave a value for the true visibilities. This first version of the simulation only mimics a one dimensional redundant telescope array detecting a small amount of sources located in the volume above the antenna plane. Future versions, using GPUs, will handle a two dimensional redundant array of telescopes detecting a large amount of sources in the volume above the array.

Summary of research results to be presented

The simulation was capable of solving for the gains and delays with a small error discrepancy. Using a core i5 cpu, large data was correlated in seconds. The next step is to convert the code for GPU programming to handle a two dimensional array.

This document is currently not available here.

Share

COinS
 
Nov 18th, 11:00 AM Nov 18th, 11:15 AM

Hydrogen Epoch of Reionization (HERA) Calibrated FFT Correlator Simulation

9-279

The Hydrogen Epoch of Reionization Array (HERA) project is an astronomical radio interferometer array with a redundant baseline configuration. Interferometer arrays are being used widely in radio astronomy because they have a variety of advantages over single antenna systems. For example, they produce images (visibilities) closely matching that of a large antenna (such as the Arecibo observatory), while both the hardware and maintenance costs are significantly lower. However, this method has some complications; one being the computational cost of correlating data from all of the antennas. A correlator is an electronic device that cross-correlates the data between the individual antennas; these are what radio astronomers call visibilities. HERA, being in its early stages, utilizes a traditional correlator system. The correlator cost scales as N2, where N is the number of antennas in the array. The purpose of a redundant baseline configuration array setup is for the use of a more efficient Fast Fourier Transform (FFT) correlator. FFT correlators scale as Nlog2N. The data acquired from this sort of setup, however, inherits geometric delay and uncalibrated antenna gains. This particular project simulates the process of calibrating signals from astronomical sources. Each signal “received” by an antenna in the simulation is given random antenna gain and geometric delay. The “linsolve” Python module was used to solve for the unknown variables in the simulation (complex gains and delays), which then gave a value for the true visibilities. This first version of the simulation only mimics a one dimensional redundant telescope array detecting a small amount of sources located in the volume above the antenna plane. Future versions, using GPUs, will handle a two dimensional redundant array of telescopes detecting a large amount of sources in the volume above the array.