Skip to content

Latest commit

 

History

History
210 lines (145 loc) · 7.08 KB

INSTALL.md

File metadata and controls

210 lines (145 loc) · 7.08 KB
Raw

Installation

Steps to install the UVMultiFit package:

Install dependencies

CASA

These instructions assume you have a casa version 5.x installed, either the normal release or with the ALMA and/or VLA pipeline included. The instructions below use an installation of casa-pipeline-release-5.6.1-8.el7 (ALMA pipeline) as an example. Earlier versions (4.x) are neither tested nor supported.

Find out, where casa is installed on your computer as you will need this information during the installation of UVMultiFit.

For Linux (Ubuntu-like), in a terminal:

sudo apt-get install libgsl-dev libfftw3-dev

For Mac OS, in a terminal:

If using Mac Ports:

sudo port install gsl fftw-3

If using Homebrew:

brew install gsl fftw

Clone the repository or update your installation

Pick a directory where you want to install UVMultiFit, e.g.

cd ~
mkdir -p casa/NordicARC
cd casa/NordicARC
git clone https://github.com/onsala-space-observatory/UVMultiFit.git
cd UVMultiFit

If you already had a git based version installed, pull in the latest changes:

git pull

Compile the C++ module

Open the Makefile and edit the first line, where it says

CASADIR=/home/olberg/Python/casa-pipeline-release-5.6.1-8.el7

and replace with the path of your own casa installation. If you have write access to the directories underneath $CASADIR, you may then run

make install

This should make uvmultifit available also to other users, running from the same casa installation.

Alternatively, you can install to your local user PYTHONPATH by running

make user

You may check that you can load the module and execute some of its support functions. Move into the test folder and start up casa

$ cd test
$ <CASADIR>/bin/casa   # replace `CASADIR` with the correct path from above!

and run the short unit test in that folder:

CASA <1>: %run test_uvmultimodel.py

which should produce output like this

C++ shared library loaded successfully

....

Ran 4 tests in 0.000s

OK

Running the test suite

Still in the test directory, and with casa running, run any of the tests via

CASA <2>: execfile("<test>.py")

where <test> is one of:

  • PROFILES_ELLIPTICITY
  • MULTICOMP
  • WIDEFIELD_MOSAIC
  • GAINS_EXPERIMENTAL
  • IMMULTIFIT
  • ONLYFLUX
  • FRINGE_FITTING

TEST 1+2: Fitting source ellipticity with different radial intensity profiles

All sources appear unresolved in the image plane. The axis ratio and position angle are, within uncertainties, compatible with the ones used in the simulation.

The residuals are all very small, with the exception of the "power-2" source:

Take into account that the "power-2" source has an infinite flux density, since the source extends to infinity and its 2d integral does not converge. Hence, the shortest baselines (i.e. the ones corresponding to the spatial scales of the order of the image size used in the simulation) have to be removed before the fitting (they are biased toward lower flux densities).

Furthermore, this "window effect" from the image used by simobserve introduces artifacts in the residual image. But such bad residuals are not indicative of a wrong fit (these are due to the fact that we have not used an exact "power-2" model for the simulation, since the model image needs to be a truncated representation of the source).

A similar thing happens with the "expo" model, which also presents some residual artifacts (notice the hint of a "missing-flux" effect in the image).

TEST 3: Fitting multiple sources simultaneously.

The source is a disc with a hole (the whole is modelled using an inner disc with a negative flux density) and a gaussian with an offset position. Notice that for the hole model to work, the surface brightness of the two discs must be the same (in absolute value). All sources appear unresolved in the image plane.

TEST 4: Widefield mosaic.

Simulations of an l-band vla mosaic with three pointings, covering an area of about 1 degree squared. the sources are distributed in a grid accross the mosaic area. This dataset is useful to check the primary-beam correction and the mosaic re-projection algorithm.

TEST 5: A Gaussian source observed with PdB.

A time-varying gain amplitude is added to one of the antennas. We use the gain solver feature in UVMULTIFIT, to solve for the source parameters and the antenna gain simultaneously, Approximating the gain with a "piece-wise" time function.

This gain-solver feature is still quite experimental. In realistic situations, a lot of fitting parameters may be needed, to solve for all the antenna gains with enough time resolution. This may slow down the code significantly (and there may be risks of falling in a local minimum of the chi2), although this method should be superior to the conventional scheme, where the gaincal has no memory of the gain values close by in time (and the calibration is solved independently of the source model parametrization).

TEST 6: IMMULTIFIT.

A simple model (two Gaussians) fitted to the gridded visibilities in the image plane.

TEST 7: A test of the "only_flux" mode,

which can be used to fit many (tens or hundreds!) source components whose shapes and coordinates are fixed in the model (i.e., only the flux densities are fitted). This mode can be used as an interface to implement modelling based on compressed-sensing techniques.

Notice that simobserve introduces a small sub-pixel shift in RA, which introduces small systematic residuals when the sources are fixed to their nominal positions (look at the plot of residuals).

TEST 8: An example of global fringe fitting, a-la UVMULTIFIT.

A synthetic C-band dataset is generated with realistic noise and 128 channels (covering 32mhz). The gains are estimated first by fitting them to the peaks in delay-rate space (using the new helper rutine "QuinnFF"). These estimates are then used as a-priori for the ordinary global fringe fitting.

Notice that calibration tables are not generated (that feature is beyond the purpose of UVMULTIFIT), but the calibrated data can still be generated, either by dividing the data by the (gain-corrected) model (which UVMULTIFIT writes in the model column when you set write='model') or by setting write='calibrated'. the master script in the test directory shows how to do this.

A fixed point source is used for the gff (centered at the nominal source position), but of course more complicated structures (and even *.model images generated with clean/tclean) can be used as models and you can even fit the source parameters together with the gains!

Running your own model:

Now, this can be done from any directory.

  • Start up casa and run

      CASA <1>: from NordicARC import uvmultifit as uvm
  CASA <2>: help(uvm.uvmultifit)                 # to get the help text
  CASA <3>: myfit = uvm.uvmultifit(vis=..., ...) # fit your model ...

  • ENJOY!

Any feedback and bug report should be sent either to the ARC Nordic Node (contact@nordic-alma.se) or to the source maintainer (michael.olberg@chalmers.se).