Aips++/Gildas test
Data Set Description
The data set was obtained in the following way:
-
Select an approximately two hour period of one day of the Phase I data set: (25-mar-1997, scans 7306 to 7429)
-
Create from this data a N-antenna Plateau de Bure data set:
- Header data for each antenna is cloned from that of antenna N (modulo 5) of the originalheader data (5 was the actual number of available antennas at Plateus de Bure in 1997).
- All visibilities are replaced by point source visibilities, using the source flux for calibrators, and 0.07 Jy for the target source (GGTau). Random thermal noise is added corresponding tothe channel width and system temperatures.
- Antenna positions are replaced with random positions (with a dispersion of 50m, maxbaseline 300m), and a minimum separation of 24m. U and V are computed accordingly).
-
Convert the data set into ALMATI-FITS.
Four data sets were prepared with N = 8, 16, 32, 64. The data set sizes are 139, 465, 1774, and 6979 Megabytes (fits files).
The actual integration time of the original data set was 106 min. The simulated visibility data rate is thus about 1.1 Mb/s (12.5% of the specified average visibility rate, if we take into account that those fits files contain 8-byte visibilities, not 4-byte as specified for ALMA).
Data Set Processing on iralx7
The processing was made on a AMD 1532 Mhz PC with 768 Mbytes of memory and 256kButes of cache memory. All data was on a local disk.
- Gildas processing is made by CLIC program (special version for 8,16,32 and 64 antennas) built in February 2003.
- Aips++ processing is made from a binary installation (using aupdate) of the software.