WODEN demonstrated via examples

Below are a number of example simulations with various settings. You can find and run these examples in the WODEN/examples directory. I’ll let you know how much storage you’ll need for each set of simulation and images (things can get large with radio data).

Two of the southern hemisphere sky models are largish (a total of about 100 MB), so instead of bundling them into the github, I’ve added links into the relevant instructions to download them. For the LOFAR examples, the sky models are linked within the notebook, so you can download them from source.

Note

For all simulation times reported in the below, when using a GPU, I used an NVIDIA GeForce GTX 1080 Ti with 12 GB of RAM. The CPU used was a AMD Ryzen™ 7 2700X x16.

The examples are:

Fornax A simulation - two examples that compare a point/Gaussian model to a shapelet model. This serves as a general introduction on how to simulate an MWA observation using a metafits file, with a few extra commands. It also serves as a comparison of running with woden_float and woden_double.

MWA EoR1 simulation - demonstrates using a larger (>300,000) source catalogue

EDA2 Haslam Map simulation - this demonstrates using WODEN without a metafits file, using a text file to describe the array layout, and using the EDA2 beam.

MWA dipole amplitudes and flags - this demonstrates using WODEN when supplying either dipole flags or amplitudes to the MWA FEE beam (via hyperbeam). Note, currently this only works with the FEE beam, not the analytic MWA beam

Polarisation - demonstrates how to get create polarised sky models, simulate them, and gives examples of how to inspect the results.

LOFAR LoTSS DR2 cutout matching - demonstrates how to get grab a chunk of the LoTSS DR2 gaussian catalogue, convert to WODEN format, and simulate it with the LOFAR HBA beam.

LOFAR LBA NCP - demonstrates how to simulate the LOFAR LBA beam with a sky model of the North Celestial Pole. Also has a cheeky bit of cross-matching, and discusses how borked plotting images at the NCP can be.

Warning

If you have a GPU with small amounts of RAM (say 2GB) some of these simulations won’t work to DOUBLE precision, you won’t have enough memory. You can add the --precision=float argument to switch to a lower memory requirement (for the loss of accuracy).