Development of automated data reduction and analysis pipelines
Led by RUB: With the advance of large, all-sky surveys it becomes important to be able to reduce and analyze most data sets in an automated manner. Specifically for MeerKAT, several large scale surveys are planned. Due to the huge amount of raw data coming out from the telescope, quick data reduction is imperative as it becomes increasingly difficult to store them over long periods of time. For this purpose several data reduction and analysis pipelines are being developed. One of the most advanced data product that can come from an interferometer is a spectral data cube. In order to get such a cube many other products, such as continuum images are automatically produced. As such the current research effort in this project is on the development of a data reduction pipeline that produces spectral data cubes from the raw MeerKAT data in automated fashion. The data products from this pipeline should, in the end, be science ready neutral hydrogen line emission data cubes as well as images for all polarization parameters. Additionally, kinematical analyses pipelines and tools are being developed in order to process the finalized data products from the data reduction pipeline and study astrophysical implications.
Data quality and search algorithms
Led by UBi & MPIfR: The aim of this project is searching for pulsars using artificial intelligence techniques. While pulsars do show a characteristic periodicity, they are hard to detect due to their faintness. Current search techniques are able to detect pulsars along with many, often orders of magnitude more, false positive candidates. These candidates can only be classified properly after further, computationally-expensive, processing. By using a powerful telescope like MeerKAT, the number of pulsar candidates will only grow larger. Especially for pulsars that show interesting features, such as acceleration or extreme nulling, the sensitivity of current search algorithms decreases considerably.
We aim to build a search pipeline, employing the power of artificial neural networks, that can — run in real time while the observations are being performed; produce far lower false-positive rate; and be tuned to have higher sensitivity to exotic systems. The neural network have been shown to perform very well on a wide variety of classification tasks. We are exploring convolutional neural networks and recurrent neural networks for the detection of very faint signals in very long time-series data. The pipeline will be applied to the data from MeerKAT's TRAPUM LSP.
Multi-spectral imaging and calibration
Led by MPA & TUM: The Information Field Theory (IFT) group at the MPI for Astrophysics is developing novel imaging and calibration algorithms based on IFT. These explore and exploit correlations of signals in spatial, temporal and spectral domains. For MeerKAT, an enhancement of the RESOLVE algorithms to provide spatial-spectal image cubes directly from the raw visibilities is in development. Performance of the RESOLVE algorithm on a real dataset is shown in the figure.
Algorithm development for multi-wavelength source characterization
Led by LMU: The main objective of this project is to build tools that, by means of the available multwavelength database, will allow the end users to associate radio sources detected in the upcoming MeerKAT radio-frequency surveys to their galaxy counterparts and, to then characterize the physical properties of those sources. Radio continuum emission is characterized by a featureless spectrum mainly produced by relativistic electrons accelerated in the host object's magnetic field, known as the synchrotron emission. Synchrotron emission from radio-loud active galactic nuclei (AGN) and star-forming galaxies have, statistically speaking, similar continuum spectra and hence discerning the properties of the host galaxy is of paramount importance to enable the full scientific exploitation of deep radio observations, like the surveys carried by the MeerKAT array over the coming decade. Our efforts are currently focused on (1) counterpart identification, (2) photometric redshift estimation and (3) classification of radio sources as star forming or AGN.
Sky survey with the MPIfR-MT SKA-protoype dish
Led by UBi & MPIfR: The 15-m aperture MPIfR-MT SKA-prototype dish is an excellent instrument to perform fast, sensitive, broad-bandwidth polarization surveys of the entire southern sky. The prototype dish will provide spectro-polarimetric data in the frequency range 1.7 to 3.6 GHz at 1 degree angular resolution and thereby the opportunity to explore an entirely new parameter space in studies of the Galactic magneto-ionic medium. Some of the exciting scientific prospects that are being pursued are:
- Separate the synchrotron and the free-free emission from the Milky Way by using the broad-bandwidth radio continuum data alone.
- Quantify Faraday depolarization through broad-bandwidth spectro-polarimetry and study the nature of turbulence in the Mikly Way's magneto-ionic medium.
- Study the origin and nature of large-scale polarized structures in the Milky Way.
- Pin down the contribution of the foreground synchrotron + free-free emission components to the cosmic microwave background (CMB) at few μK accuracy per 1 degree2, and the contribution of the polarized synchrotron emission at sub-μK accuracy. This will be crucial in detecting signatures of primordial gravitational waves and the reionization history of the Universe imprinted on the E- and B-modes of the polarized CMB.
- Monitor time-variability of bright active galactic nucleii (AGN) in all Stokes parameters for several years with roughly one month cadence. This will provide insights into the connection between AGN polarization variability related to black hole accretion and jet launching mechanisms.
Members of the D-MeerKAT consortium: