file_name, bin_width, do_scatter Read UV data from a random-groups UV FITS file. EXAMPLE ------- 0>observe /scr/mcs/fits/gbt_spec.uvf Reading UV FITS file: /scr/mcs/fits/gbt_spec.uvf AN table 1: 7 integrations on 45 of 45 possible baselines. Apparent sampling: 1 visibilities/baseline/integration-bin. Found source: 1641+399 There are 2 IFs, and a total of 128 channels: IF Channel Frequency Freq offset Number of Overall IF origin at origin per channel channels bandwidth ------------------------------------------------------------- (Hz) 01 1 1.53575e+10 125000 64 8e+06 02 65 1.53655e+10 125000 64 8e+06 Polarization(s): RR LL RL LR Read 353 lines of history. Reading 161280 visibilities. 0> All lines following the 'observe' command line are informational output messages from the observe command. PARAMETERS ---------- file_name - The name of the UV FITS file to be read. bin_width - Default=0.0 (seconds) This argument is used to specify how visibilities are to be collected into integrations. This is discussed further in the CONTEXT section below. There are two alternatives: 1. If binwid<1.0 seconds then 'observe' will assign an integration to every time-stamp in the file. This is ok if visibilities on different baselines within the same integration have the same time-stamps, but disasterous otherwise. 2. If binwid>=1.0 seconds then 'observe' will bin visibilities onto a regular integration grid of spacing 'binwid' seconds anchored at 0 UTC on the first day of the observation. Where multiple visibilities fall on the same baseline of a given integration, they will be vector averaged using a weighted mean using the same algorithm as uvaver. do_scatter - Default=false If do_scatter=true the weights from the FITS file will be supplanted by weights estimated from the scatter of the data within each integration bin. There must be at least two visibilities per baseline within each integration bin to do this. Where this is not the case the output visibility will be flagged, and be assigned the input visibility weight. CONTEXT ------- Before interferometer data can be processed, it must be read into difmap. The observe command reads interferometer data in the form of single source UV FITS files. The older Caltech VLBI merge format files are no longer acceptible. The caltech VLBI package program 'MERGEFITS' may be used to convert merge files to FITS. UV FITS files do not provide any means to map visibilities on different baselines into integrations. Each visibility has its own time-stamp, which need not aggree with those on other baselines within the same logical integration. In fact UV FITS files do not even limit visibilities on different baselines to have the same integration time. Difmap, on the other hand does require that visibilities be grouped into integrations. This is the reason for the 'binwid' argument of the 'observe' command. If the visibilities do not lie on an integration grid then you must specify a suitable integration time into which visibilities should be binned into integrations. Depending on how the FITS file has been processed, it may already have visibilities grouped into integrations with identical time-stamps assigned to each grouped visibility, in which case no 'binwid' argument will be required. This is the case for files written by difmap. If you do not know what state your file is in, then try to read it with the observe command without specifying an integration time. Then if 'observe' reports something along the following lines, then either run the 'uvaver' command to re-grid the data or equivalently re-run the observe command with a suitable integration time. observe poor_sampling.uvf ... Apparent sampling: 0.245305 visibilities/baseline/integration. *** This seems a bit low - see "help observe" on the binwid argument. ... Other symptoms of in-completely binned integrations are that selfcal flags all of your data due to the lack of closure quantities, and that station based editing in vplot behaves like baseline based editing. While the 'binwid' and 'scatter' options are provided primarily as a mechanism to group data into integrations, they may also be employed to average ones data to longer integration times for efficiency reasons. The effect is identically equivalent to reading the data with observe and then running the 'uvaver' command with the same arguments, except that in the latter case you might not have sufficient memory to read the data before averaging and reading in the larger file would take longer. LIMITATIONS ----------- Only single source, time-ordered UV FITS files will be accepted by difmap. Multi-source files must be split into one or more single source files before presenting them to difmap. AIPS provides the SPLIT task to do this. Similarly, AIPS UVSRT may be used to re-sort files into time order (eg. TB order). Difmap accepts multiple sub-arrays within a file, multiple polarizations, spectral-line channels and multiple IFs. While difmap does accept multiple IFs, it does not make any allowances for spectral-index effects over the range of frequencies covered by the IFs. If the source structure changes significantly between IFs, you should SPLIT the IFs into separate files and process them independantly within difmap. Also note that difmap requires that the frequencies assigned to each IF do not change over the duration of the file. To enforce this difmap rejects files containing multiple frequency IDs. SIDE EFFECTS ------------ Any previously read UV data will be discarded before the new file is read. Any existing models and CLEAN windows will also be discarded. All weight options are restored to their defaults. The current map-grid will be cleared, but its dimensions and cell-size will not be changed. RELATED COMMANDS ---------------- wobs - Write out a modified UV FITS file. select - Select a polarization and channels to be processed. uvaver - Used to re-grid UV data.