CubiCal: a fast radio interferometric calibration suite exploiting complex optimisation
- Authors: Kenyon, Jonathan
- Date: 2019
- Subjects: Interferometry , Radio astronomy , Python (Computer program language) , Square Kilometre Array (Project)
- Language: English
- Type: text , Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10962/92341 , vital:30711
- Description: The advent of the Square Kilometre Array and its precursors marks the start of an exciting era for radio interferometry. However, with new instruments producing unprecedented quantities of data, many existing calibration algorithms and implementations will be hard-pressed to keep up. Fortunately, it has recently been shown that the radio interferometric calibration problem can be expressed concisely using the ideas of complex optimisation. The resulting framework exposes properties of the calibration problem which can be exploited to accelerate traditional non-linear least squares algorithms. We extend the existing work on the topic by considering the more general problem of calibrating a Jones chain: the product of several unknown gain terms. We also derive specialised solvers for performing phase-only, delay and pointing error calibration. In doing so, we devise a method for determining update rules for arbitrary, real-valued parametrisations of a complex gain. The solvers are implemented in an optimised Python package called CubiCal. CubiCal makes use of Cython to generate fast C and C++ routines for performing computationally demanding tasks whilst leveraging multiprocessing and shared memory to take advantage of modern, parallel hardware. The package is fully compatible with the measurement set, the most common format for interferometer data, and is well integrated with Montblanc - a third party package which implements optimised model visibility prediction. CubiCal's calibration routines are applied successfully to both simulated and real data for the field surrounding source 3C147. These tests include direction-independent and direction dependent calibration, as well as tests of the specialised solvers. Finally, we conduct extensive performance benchmarks and verify that CubiCal convincingly outperforms its most comparable competitor.
- Full Text:
- Date Issued: 2019
- Authors: Kenyon, Jonathan
- Date: 2019
- Subjects: Interferometry , Radio astronomy , Python (Computer program language) , Square Kilometre Array (Project)
- Language: English
- Type: text , Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10962/92341 , vital:30711
- Description: The advent of the Square Kilometre Array and its precursors marks the start of an exciting era for radio interferometry. However, with new instruments producing unprecedented quantities of data, many existing calibration algorithms and implementations will be hard-pressed to keep up. Fortunately, it has recently been shown that the radio interferometric calibration problem can be expressed concisely using the ideas of complex optimisation. The resulting framework exposes properties of the calibration problem which can be exploited to accelerate traditional non-linear least squares algorithms. We extend the existing work on the topic by considering the more general problem of calibrating a Jones chain: the product of several unknown gain terms. We also derive specialised solvers for performing phase-only, delay and pointing error calibration. In doing so, we devise a method for determining update rules for arbitrary, real-valued parametrisations of a complex gain. The solvers are implemented in an optimised Python package called CubiCal. CubiCal makes use of Cython to generate fast C and C++ routines for performing computationally demanding tasks whilst leveraging multiprocessing and shared memory to take advantage of modern, parallel hardware. The package is fully compatible with the measurement set, the most common format for interferometer data, and is well integrated with Montblanc - a third party package which implements optimised model visibility prediction. CubiCal's calibration routines are applied successfully to both simulated and real data for the field surrounding source 3C147. These tests include direction-independent and direction dependent calibration, as well as tests of the specialised solvers. Finally, we conduct extensive performance benchmarks and verify that CubiCal convincingly outperforms its most comparable competitor.
- Full Text:
- Date Issued: 2019
PyMORESANE: A Pythonic and CUDA-accelerated implementation of the MORESANE deconvolution algorithm
- Authors: Kenyon, Jonathan
- Date: 2015
- Subjects: Radio astronomy , Imaging systems in astronomy , MOdel REconstruction by Synthesis-ANalysis Estimators (MORESANE)
- Language: English
- Type: Thesis , Masters , MSc
- Identifier: vital:5563 , http://hdl.handle.net/10962/d1020098
- Description: The inadequacies of the current generation of deconvolution algorithms are rapidly becoming apparent as new, more sensitive radio interferometers are constructed. In light of these inadequacies, there is renewed interest in the field of deconvolution. Many new algorithms are being developed using the mathematical framework of compressed sensing. One such technique, MORESANE, has recently been shown to be a powerful tool for the recovery of faint difuse emission from synthetic and simulated data. However, the original implementation is not well-suited to large problem sizes due to its computational complexity. Additionally, its use of proprietary software prevents it from being freely distributed and used. This has motivated the development of a freely available Python implementation, PyMORESANE. This thesis describes the implementation of PyMORESANE as well as its subsequent augmentation with MPU and GPGPU code. These additions accelerate the algorithm and thus make it competitive with its legacy counterparts. The acceleration of the algorithm is verified by means of benchmarking tests for varying image size and complexity. Additionally, PyMORESANE is shown to work not only on synthetic data, but on real observational data. This verification means that the MORESANE algorithm, and consequently the PyMORESANE implementation, can be added to the current arsenal of deconvolution tools.
- Full Text:
- Date Issued: 2015
- Authors: Kenyon, Jonathan
- Date: 2015
- Subjects: Radio astronomy , Imaging systems in astronomy , MOdel REconstruction by Synthesis-ANalysis Estimators (MORESANE)
- Language: English
- Type: Thesis , Masters , MSc
- Identifier: vital:5563 , http://hdl.handle.net/10962/d1020098
- Description: The inadequacies of the current generation of deconvolution algorithms are rapidly becoming apparent as new, more sensitive radio interferometers are constructed. In light of these inadequacies, there is renewed interest in the field of deconvolution. Many new algorithms are being developed using the mathematical framework of compressed sensing. One such technique, MORESANE, has recently been shown to be a powerful tool for the recovery of faint difuse emission from synthetic and simulated data. However, the original implementation is not well-suited to large problem sizes due to its computational complexity. Additionally, its use of proprietary software prevents it from being freely distributed and used. This has motivated the development of a freely available Python implementation, PyMORESANE. This thesis describes the implementation of PyMORESANE as well as its subsequent augmentation with MPU and GPGPU code. These additions accelerate the algorithm and thus make it competitive with its legacy counterparts. The acceleration of the algorithm is verified by means of benchmarking tests for varying image size and complexity. Additionally, PyMORESANE is shown to work not only on synthetic data, but on real observational data. This verification means that the MORESANE algorithm, and consequently the PyMORESANE implementation, can be added to the current arsenal of deconvolution tools.
- Full Text:
- Date Issued: 2015
- «
- ‹
- 1
- ›
- »