The scripts included with LSL are primarily intended to be used as a guide to writing new scripts to accomplished particular tasks. However, some of the scripts are genuinely useful.
At 200 kB per stand per polarization per second, it is easy to end up with large TBN files. splitTBN.py makes it easy to remove only a small section of a particular file for analysis. The advantage of this is that is makes it easier to perform a quick look using scripts like tbnSpectra.py. All of the options to splitTBN.py are safe and have no adverse side-effects. To split off the first hour of a large TBN file:
python splitTBN.py -c 3600 TBN_file.dat
For the second hour, use:
python splitTBN.py -o 3600 -c 3600 TBN_file.dat
These commands create files with names of the form TBN_file_s<first frame number>.dat. A more convenient way to name them is using the ‘-d’ flag with appends the date and time of the first frame in the newly created file.
splitTBN.py is also useful for aligning TBN files with full frames. In some situations, the first frame in a TBN file does not correspond to the first stand. splitTBN.py examines the start of every file and removes the first capture if it is missing some of the frames.
tbnSpectra.py provides a way to plot integrated spectra for all data in a TBN file. For short TBN captures (up to a few minutes) it can be directly run on the file. However, longer TBN files should be pared down with splitTBN.py to ensure that tbnSpectra.py does not take hours to run. To use tbnSpectra.py:
python tbnSpectra.py -l 1024 -o spectra.png TBN_file.dat
The above command creates a 1,024 channel spectra of the given TBN file, displays the spectra to the screen, and then saves the plots to the file spectra.png.
For files that contain more data than can fit in the machine’s memory at once, tbnSpectra.py ‘chunks’ the data into 400,000 frame sections. These chunks are averaged together to generate the global average of the file. In addition, the relative difference between the average within each of the chunks and the global average is show in the spectra plot as variations around 0 dB.
Here is a Python code snippet for reading in TBN data:
1 2 3 4 5 6 7 | >>> from lsl.reader import tbn, errors
>>> fh = open('TBN_file.dat', 'rb')
>>> frame = tbn.readFrame(fh)
>>> print frame.parseID()
(1, 0)
>>> print frame.data.iq.mean()
5.03+1.03j
|
In the above code, line 3 reads the raw TBN frame into a lsl.reader.tbn.Frame object. Lines 4 and 6 access the Frame objects various attributes. Line 4, for example, parses the TBN ID field and returns a two-element tuple containing the stand number and polarization. Line 6 prints the mean value of the I/Q data associated with this frame.
Warning
The TBN reader can throw various errors when reading in a TBN frame if the frame is truncated or corrupted. These errors are defined in lsl.reader.errors. tbn.readFrame should be wrapped inside a try...except to check for these.
Note
With the full LWA-1 station, the output frames of TBN are interleaved and, thus, do not have a predefined order. The lsl.reader.buffer module provides a simple ring buffer than can be used in conjunction with the reader to re-order the frames. See if include tbnSpectra.py script for an example implementation.
After the TBN data have been read in, spectra can by computed and plotted using the function lsl.correlator.fx.calcSpectra(). For example:
1 2 | >>> from lsl.correlator import fx as fxc
>>> freq, spec = fxc.calcSpectra(data, LFFT=2048, SampleRate=1e5, CentralFreq=38e6, DisablePool=True)
|
Where data is a 2-D array of where the first dimension loops through stands and the second samples. Unlike TBW data, the additional keywords ‘SampleRate’ and ‘CentralFreq’ are needed to create the correct frequencies associated with the FFTs. The sample rate can be obtained from the data using:
>>> sampleRate = tbn.getSampleRate(fh)
which uses the time tags of sequetial frames to determine the sample rate. Currently there is not a way to determine the central frequency of the observations from the data frames.
LSL 0.4.0 introduces a new way to compute spectra with the lsl.correlator.fx.SpecMaster() function. This function uses a C extension and OpenMP to provide better overall performance. SpecMaster is called in the same way as the original calcSpectra function:
>>> freq, spec = fxc.SpecMaster(data, LFFT=2048, SampleRate=1e5, CentralFreq=38e6)
Once the spectra have been computed, they can be plotted via matplotlib via:
1 2 3 4 5 6 7 | >>> import numpy
>>> from matplotlib import pyplot as plt
>>> fig = plt.figure()
>>> ax = fig.gca()
>>> ax.plot(freq/1e3, numpy.log10(spec[0,:])*10.0)
>>> ax.set_xlabel('Frequency [kHz]')
>>> ax.set_ylabel('PSD [Arb. dB]')
|
Note
In the above example, the thread pool has been disabled for lsl.correlator.fx.calcSpectra() which forces the function to run single-threaded. By default, calcSpectra runs with 4 threads and this can cause problems if a Ctrl-C is issued. Ctrl-C kills the main python thread but leaves the worker threads running.
For post-acquisition beam forming, you need need an azimuth (in degrees) and elevation (in degrees) to point the beam towards. For planets, this can be accomplished using the pyephem package that is required by lsl. For example, compute the location of Jupiter at LWA-1 on 12/17/2010 at 21:18 UTC (JD 2,455,548.38787):
1 2 3 4 5 6 7 8 9 10 | >>> import math
>>> import ephem
>>> from lsl.common import stations
>>> lwa1 = stations.lwa1
>>> lwaObserver = lwa1.getObserver(2455548.38787, JD=True)
>>> jove = ephem.Jupiter()
>>> jove.compute(lwaObserver)
>>> print "Jupiter: az -> %.1f, el -> %.1f" % (jove.az*180/math.pi,
... jove.alt*180/math.pi)
Jupiter: az -> 112.4, el -> 24.4
|
Line 4 defines the location for LWA-1 as a lsl.common.stations.LWAStation object while line 5 create an ephem.Observer object that can be used to calculate the sky positions of various bodies. The position of Jupiter is calculated using this Observer object on lines 6 and 7.
Note
When working with positions from pyephem objects, all values are in radians. For more information about pyehem, see http://rhodesmill.org/pyephem/
For fixed positions, use:
1 2 3 4 5 6 7 | >>> cyga = ephem.FixedBody()
>>> cyga._ra = '19:59:28.30'
>>> cyga._dec = '+40:44:02'
>>> cyga.compute(lwaObserver)
>>> print "Cygnus A: az -> %.1f, el -> %.1f" % (cyga.az*180/math.pi,
... cyga.alt*180/math.pi)
Cygnus A: az -> 10.0, el -> 83.2
|
After TBN data have been read in and a pointing position has been found, a beam can be formed. For example, forming a N-S beam via integer sample delay-and-sum on Cygnus A for data taken on JD 2,455,548.38787:
1 2 3 4 5 6 7 8 | >>> from lsl.misc import beamformer
>>> antennas = []
>>> for ant in lwa1.getAntennas():
... if ant.pol == 0:
... antennas.append(ant)
...
>>> beamdata = beamformer.intDelayAndSum(antennas, data, sampleRate=1e5,
... azimuth=10.0, elevation=83.2)
|
Lines 2 through 5 retrieves the list of antennas used for observations and selects only antennas with N-S polarization. This information is needed in order to get the correct delays geometric and cable delays to use for the beam forming.