Multi-GPU maximum entropy image synthesis for radio astronomy

The maximum entropy method (MEM) is a well known deconvolution technique in radio-interferometry. This method solves a non-linear optimization problem with an entropy regularization term. Other heuristics such as CLEAN are faster but highly user dependent. Nevertheless, MEM has the following advantages: it is unsupervised, it has a statistical basis, it has a better resolution and better image quality under certain conditions. This work presents a high performance GPU version of non-gridding MEM, which is tested using real and simulated data. We propose a single-GPU and a multi-GPU implementation for single and multi-spectral data, respectively. We also make use of the Peer-to-Peer and Unified Virtual Addressing features of newer GPUs which allows to exploit transparently and efficiently multiple GPUs. Several ALMA data sets are used to demonstrate the effectiveness in imaging and to evaluate GPU performance. The results show that a speedup from 1000 to 5000 times faster than a sequential version can be achieved, depending on data and image size. This allows to reconstruct the HD142527 CO(6-5) short baseline data set in 2.1 minutes, instead of 2.5 days that takes a sequential version on CPU.

[1]  Robert K. Pina,et al.  BAYESIAN IMAGE RECONSTRUCTION: THE PIXON AND OPTIMAL IMAGE MODELING , 1993 .

[2]  Morphological Analysis of the Centimeter-Wave Continuum in the Dark Cloud LDN 1622 , 2005, astro-ph/0511283.

[3]  Cagatay Candan,et al.  The discrete fractional Fourier transform , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[4]  T. Cornwell,et al.  A simple maximum entropy deconvolution algorithm , 1985 .

[5]  Imaging VLBI polarimetry data from Active Galactic Nuclei using the Maximum Entropy Method , 2013, 1310.6187.

[6]  Kristen Menou,et al.  DYNAMICAL STABILITY OF IMAGED PLANETARY SYSTEMS IN FORMATION: APPLICATION TO HL TAU , 2015, 1502.05099.

[7]  Valentin Christiaens,et al.  ACCRETION KINEMATICS THROUGH THE WARPED TRANSITION DISK IN HD 142527 FROM RESOLVED CO(6–5) OBSERVATIONS , 2015, 1505.07732.

[8]  Vladimir Temlyakov,et al.  CAMBRIDGE MONOGRAPHS ON APPLIED AND COMPUTATIONAL MATHEMATICS , 2022 .

[9]  Weidong Chen,et al.  The ill-posedness of the sampling problem and regularized sampling algorithm , 2011, Digit. Signal Process..

[10]  European Southern Observatory,et al.  FIRST RESULTS FROM HIGH ANGULAR RESOLUTION ALMA OBSERVATIO NS TOWARD THE HL TAU REGION , 2015, 1503.02649.

[11]  William H. Press,et al.  Numerical recipes in C (2nd ed.): the art of scientific computing , 1992 .

[12]  D K Smith,et al.  Numerical Optimization , 2001, J. Oper. Res. Soc..

[13]  J. Högbom,et al.  APERTURE SYNTHESIS WITH A NON-REGULAR DISTRIBUTION OF INTERFEROMETER BASELINES. Commentary , 1974 .

[14]  R. Narayan,et al.  Maximum Entropy Image Restoration in Astronomy , 1986 .

[15]  Robert A. Shaw,et al.  Astronomical data analysis software and systems IV : meeting held at Baltimore, Maryland, 25-28 September 1994 , 1995 .

[16]  D. Broguiere,et al.  THE 2014 ALMA LONG BASELINE CAMPAIGN: FIRST RESULTS FROM HIGH ANGULAR RESOLUTION OBSERVATIONS TOWARD THE HL TAU REGION , 2015 .

[17]  Eric Anterrieu,et al.  CLEAN and WIPE , 1997 .

[18]  S. Gull,et al.  Image reconstruction from incomplete and noisy data , 1978, Nature.

[19]  I. Johnstone,et al.  Maximum Entropy and the Nearly Black Object , 1992 .

[20]  T. Cornwell,et al.  A multi-scale multi-frequency deconvolution algorithm for synthesis imaging in radio interferometry , 2011, 1106.2745.

[21]  T. J. Cornwell,et al.  Radio-interferometric imaging of very large objects , 1988 .

[22]  C. Carilli,et al.  Synthesis Imaging in Radio Astronomy II , 1999 .

[23]  F. Ménard,et al.  DUST AND GAS IN THE DISK OF HL TAURI: SURFACE DENSITY, DUST SETTLING, AND DUST-TO-GAS RATIO , 2015, 1508.00584.

[24]  E. Fomalont,et al.  The unusual radio galaxy 3C 288 , 1989 .

[25]  M. Kunz,et al.  Bayesian inference for radio observations , 2015, 1501.05304.

[26]  A. Warmuth,et al.  Thermal and nonthermal hard X-ray source sizes in solar flares obtained from RHESSI observations. I. Observations and evaluation of methods , 2013 .

[27]  Pierre Maréchal,et al.  Fourier synthesis via partially finite convex programming , 2009, Math. Comput. Model..

[28]  Kim M. Hazelwood,et al.  Where is the data? Why you cannot debate CPU vs. GPU performance without the answer , 2011, (IEEE ISPASS) IEEE INTERNATIONAL SYMPOSIUM ON PERFORMANCE ANALYSIS OF SYSTEMS AND SOFTWARE.

[29]  Wilhelm Burger,et al.  Principles of Digital Image Processing - Core Algorithms , 2009, Undergraduate Topics in Computer Science.

[30]  Edmund C. Sutton,et al.  Optimal Image Reconstruction in Radio Interferometry , 2006, astro-ph/0604331.

[31]  Stephanie Thalberg,et al.  Interferometry And Synthesis In Radio Astronomy , 2016 .

[32]  F. Owen,et al.  THE COMPLEX NORTH TRANSITION REGION OF CENTAURUS A: RADIO STRUCTURE , 2015, 1502.05066.

[33]  Santiago,et al.  Bayesian Image Reconstruction Based on Voronoi Diagrams , 2007, 0712.4140.

[34]  Valentin Christiaens,et al.  A COMPACT CONCENTRATION OF LARGE GRAINS IN THE HD 142527 PROTOPLANETARY DUST TRAP , 2015, 1505.07743.

[35]  S. Markoff,et al.  LOFAR - low frequency array , 2006 .

[36]  Blythe G. Clark.  Coherence in Radio Astronomy , 1989 .

[37]  Patrick Marais,et al.  Montblanc: GPU accelerated Radio Interferometer Measurement Equations in support of Bayesian Inference for Radio Observations , 2015, Astron. Comput..

[38]  Max Grossman,et al.  Professional CUDA C Programming , 2014 .

[39]  B. Winkel,et al.  Cygrid: A fast Cython-powered convolution-based gridding module for Python , 2016, 1604.06667.

[40]  Alexander S. Szalay,et al.  Delivering SKA Science , 2015, ArXiv.

[41]  D. Iono,et al.  High-resolution ALMA observations of SDP.81. I. The innermost mass profile of the lensing elliptical galaxy probed by 30 milli-arcsecond images , 2015, 1503.07605.

[42]  Martin Kay,et al.  Morphological Analysis , 1973, COLING.