Simulations of Radio Observations

Record of the efforts in simulating radio observations

Structure of Meqtrees module

MeqTrees is a software package for implementing Measurement Equations. This makes it uniquely suited for simulation and calibration of radioastronomical data, especially that involving new radiotelescopes and observational regimes.
RIME: Radio Interferometer Measurement Equation

A major challenge in the new-generation interferometers is
  • Direction-Dependent effects, i.e instrument response varies with the direction within the beam.
Advantages of Meqtrees are
  1. It explicitly solves for the RIME
  2. It has build-in features for generalised selfcal which allows arbitrary RIMEs with arbitrary parameterization
  3. User friendly application: modelling in python and numierical computation effeciency in C++
  4. Visualisation tools
Important terms needed are
  1. Tree:
  2. Leaf:
  3. Root:
  4. Node:
  5. Parent Node:
  6. Child Node:
  7. Node Class:
Implementation od Meqtrees have three major components:
  1. the Tree Definition Language (TDL) is a Python-based language for building expression trees. It allows one to succinctly specify nodes and their connections by means of class, name, children, and other options.
  2. the "meqserver" is the computational back-end of MeqTrees. It is mostly implemented in C++. A meqserver process takes care of constructing and evaluating trees, and interfacing them to datasets;
  3. the meqbrowser is a separate GUI (implemented in Python, and running in a separate process) for controlling meqservers. It parses TDL scripts, instructs servers to build the corresponding trees, and takes care of visualizing the results. (Note that it is also possible to run meqservers in noninteractive mode, without a browser.)
Existing framework in the context of radio astronomy
  • Meow: Measurement Equation Object frameWork
  • Siamese: two higher-level frameworks for simulation
  • Calico: calibration
Major steps in the simulation and later calibration,
  1. For simulation, the RIME (plus an optional noise term) predicts the output of a [real or theoretical] telescope observing a model sky
  2. Given a sky model, and a set of Jones matrices for the DDE and DIE components, MeqTrees constructs a set of per-baseline trees corresponding to a RIME
  3. These trees are evaluated for a series of times and frequencies specified by a Measurement Set (MS). The resulting simulated visibility data is then written out to the MS.
  4. Calibration is, essentialy, a model fitting. The RIME is used to predict model visibilities, but the outputs of the subtrees are then treated as the model functions and fitted (by solving for their parameters) to the observed data, which is read from an MS. The resulting residuals are also written out to the MS.