Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.atnf.csiro.au/computing/software/askapsoft/sdp/docs/current/utils/delaysolver.html
Дата изменения: Unknown Дата индексирования: Tue Apr 12 13:03:09 2016 Кодировка: ISO8859-5 Поисковые слова: п п р р р р р р п п п п п п п п |
Usage:
delaysolver -c delaysolver.parset [-s scheduling_block] [-f measurement_set]
This utility is intended to assist initial delay calibration of the array after maintenance or power cycle. It processes a short observation of some dominant continuum calibrator (e.g. Virgo) and produces a file directly suitable for uploading into the FCM. The tool also supports a number of useful options to assist commissioning. It is intended to be run on Galaxy, but will work, albeit a bit slow, on aktos01 (which can be handy to avoid waiting for the data transfer).
The measurement set can be given either in the command line, or in the parset (see below). If the parset contains cp.ingest.tasks.FringeRotationTask.params.fixeddelays keyword, the delay corrections solved for by this tool will be added to the delays specified in this parameter (the name is deliberately chosen to match the fcm key and assist copy-pasting). The result is reported to the log and is written into an ascii file called corrected_fixeddelay.parset in the format accepted by fcm put. Another useful alternative, which is intended to be the main method used in operations, is to specify the scheduling block ID via the -s command line option. In this case, both previous delay values and the measurement set file name are taken from the scheduling block.
With the correctly set cutoff, the tool was expected to deal with birdies automatically and provide a delay solution automatically. However, it pays to inspect that the system is working well and doesn’t have other correlator artefacts in addition to occasional spikes. Otherwise, an incorrect delay could be inserted which can make things worse. As an additional diagnostics, the tool produces an ASCII file called avgspectrum.dat which contains phase spectra for all baselines before they are passed to the delay solver (i.e. with flagging and averaging applied). In the case of troubleshooting, it is handy to inspect these spectra which should show a reasonable phase slope (note, phase wraps are normal - they are handled well by the delay solution algorithm). The format of this file contains 4 columns:
antenna1_id (from 0 to 5)
antenna2_id (from 0 to 5)
channel_id (0 based, after averaging; 0 corresponds to 0 in the original dataset)
phase_in_degrees
Parameters understood by delaysolverr are given in the following table and should not contain any prefix:
Parameter | Type | Default | Description |
---|---|---|---|
stokes | string | “XX” | Polarisation product to use. The code does no conversion, so the parameter should correspond to a product which actually has been observed. |
resolution | double | 1e6 | Spectral resolution in Hz to average data to before solving for delays. Averaging of integer number of channels is done to match the requested resolution as close as possible. If the dataset has already equal or more coarse resolution, nothing is done to the data prior to solving |
beam | int | 0 | Beam to work with |
cutoff | double | -1 | If positive, the spectral channels which have the amplitude greater than the cutoff value are flagged. This is done before any averaging. |
exclude13 | bool | false | If true, AK01-AK03 baseline is excluded from the solution |
sbpath | string | ”./” | Path to the directory which contains scheduling blocks (only used with -s command line option |
ms | string | “” | A full path to the measurement set to use (required if ms is not given in the command line or via the scheduling block) |
cp.ingest.tasks.FringeRotationTask.params.fixeddelays | vector<double> | None | Current fixed delays (can only be used if no scheduling block is given) |
Typical parset for use on Galaxy:
The tool is used with -s command line option and takes the measurement set name and the fixed delay setting used during observations from the supplied scheduling block
stokes = XX
resolution = 1e6
beam = 0
# the cutoff may need an adjustment if beams are formed with different normalisation
# the following value seems to be good for single port beams we currently use
cutoff = 0.23
# We exclude AK01-AK03 baseline due to cross-talk and interference, otherwise it
# can skew the solution
exclude13 = true
sbpath = /scratch2/askap/askapop/archive/askap-scheduling-blocks