Skip to content

Latest commit

 

History

History
89 lines (56 loc) · 4.22 KB

README.md

File metadata and controls

89 lines (56 loc) · 4.22 KB
Raw

Dataset used for DataComb (Plunkett et al. 2023)

List of datasets:

  1. M100 - ALMA observations from Casaguide
  2. "Skymodel" - model and simulated ALMA observations

For the example, the setup assumes that the data is present in data, physically or via a (sym)link. If you cloned this Github repository, then the data directory should already exist. You can modify this for your system as needed.

1) M100

This M100 Band3 Casaguide walks you through a combination with Feather. We utilize the same dataset for Feather and for the other combination methods, so they can be compared.

To get the full data, see Obtaining the Data

You may prefer to download the nearest data, but any of these paths should work if using e.g. wget:

For 12m use directory: M100Band3_12m For 7m/TP use directory: M100Band3ACA

You will need the 12m, 7m and TP data. For example, in NA (else, replace the root of the URL with the EU/EA ones above):

 wget https://bulk.cv.nrao.edu/almadata/sciver/M100Band3_12m/M100_Band3_12m_CalibratedData.tgz
 wget https://bulk.cv.nrao.edu/almadata/sciver/M100Band3ACA/M100_Band3_7m_CalibratedData.tgz
 wget https://bulk.cv.nrao.edu/almadata/sciver/M100Band3ACA/M100_Band3_ACA_ReferenceImages_5.1.tgz

An alternative can be a single large tar file (if the archives are too cumbersome):

 wget https://ftp.astro.umd.edu/pub/teuben/DataComb/M100_big.tar

Note: We are also planning on a trimmed version of these big data to immediately jump into data combination. You can download trimmed M100 data, which were designed for the QAC benchmark. These are gridded to the same 70 channels corresponding to the TP data. For testing purposes, another reason for this trimmed version is to ensure that all data have been sorted correctly, as some programs (certainly in the past) could not deal with data sorted differently by frequency.

2) "Skymodel"

These data follow the method described by Koda et al. (2019) to generate a model representative of a GMC, with a power spectrum distribution of emission. Then we inserted two "point sources." Simulated observations were generated, as described in Plunkett et al. (2023).

The new (since late 2022) link to the DataComb data is via ftp or http (they are identical)

https://ftp.astro.umd.edu/pub/teuben/DataComb
(local at UMD) /n/ftp/pub/teuben/DataComb

In here you will find the following files:

skymodel-b.fits                  original skymodel plus two fuzzy "point" sources
skymodel-c.fits                  skymodel with a different random seed, same powerlaw as 'b'
skymodel-b.sim.tar               MS simulation files from Toshi (large: 7GB)
skymodel-c.sim.tar	             MS sim files
pointSrc_gaussOnly.sim.tar       MS sim files
pointSrc_pointOnly.sim.tar       MS sim files
pointSrc_pointSrcGauss.sim.tar   MS sim files
qac_bench5.tar.gz                M100 data : 70 channels at 5 km/s for 12m, 7m and TP

Examples for retrieving data:

wget https://ftp.astro.umd.edu/pub/teuben/DataComb/skymodel-b.fits 
wget https://ftp.astro.umd.edu/pub/teuben/DataComb/skymodel-c.fits
curl https://ftp.astro.umd.edu/pub/teuben/DataComb/skymodel-b.sim.tar | tar xf -
curl https://ftp.astro.umd.edu/pub/teuben/DataComb/skymodel-c.sim.tar | tar xf -

The "skymodel" model data are big, but best solved by placing them in data as well (or a link)

Examples:

skymodel-b.sim/skymodel-b_120L        full 8GB multi-day dataset
skymodel-c.sim/skymodel-c_120L        full 8GB multi-day dataset
gmcSkymodel/gmc_2L/                   an older - kind of wrong - but small < 1GB data for testing

Also note that the code to generate the simulations is available inside the dataset, with the comment that these were produced with CASA 5.6.1.

Other

More datasets have been tested, and may be documented in the future.