--This is a pipeline use 2DFFT method to search Fast Radio burst
Here we proposed a different FRB searching algorithm which basically trace a curve in frequency-time image. This algorithm is mainly realized by 2 dimensional Fast Fourier Transform(2DFFT). We take a 2DFFT on I(f^-2^,t) data map, Then trace the signal along the angle of straight line. In this searching method, it's easier to remove RFI in large scale and will bring a speed up benefit in well-developed 2D FFT library both in CPU and GPU code.
Add* /2DFFT_transient_search/src to your PYTHONPATH and LD_LIBRARY_PATH varible in your .bashrc file.
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:path_to_download/2DFFT_transient_search/src"
export PYTHONPATH="$PYTHONPATH:path_to_download/2DFFT_transient_search/src"
cd src/
Use
python __main__.py -h:
Usage: mpirun -n <number of processor> python __main__.py [options]
Options:
-h, --help show this help message and exit
-f FILE, --file=FILE Put filterbank file want to search
-t THRESHOLD, --threshold=THRESHOLD
Threshold(sigma) for candidates pick
--dm=DM Set DM range, Suggest use default [50,2000]
--pixel=PIXEL Pixels number at 2nd 1DFFT map
--mask_cycle=MSK_CYCLE
Mask abnormal lines at 1st 2DFFt map, only when RFI is
terrible
--nsamps_gulp=T_LEN Samples number for onece process, suggest use self-
calculate(def)
--nbin=NBIN number of channels after re-bin step, suggest use
self-calculate(def)
--wp=WP Set width of pulse to search, Suggest use default
10(ms).
--angle=ANGLE Angle range within [0,90], For de-buging.
-v, --verbose Show details of process
-p, --plot Make overview plot for final result
--Plot_proc=PLOT_PROC
Input process step key words want to make signle plot.
Key words Including: {raw, rebin, 1stFFT,
polar_sets_3D, polar_sets_2D, 2ndFFT_3D, 2ndFFT_2D}
(This function Remain updates)
==Input file is required for filterbank file (*.fil) . From SIGPYPROC Readers.py is used as one of our dependent package ==
If we use '-p' in parameter options, we will get a final plot like this:
Plot for FRB090625 search result. Data has 4620288 time samples and 1024 frequency channels. Top left give the candidate plot at 2nd 1DFFT map. Right top plot the candidates pixels number at each DM value and significance. Bottom plot is like traditional plot for DM and time, but time here are instead by time samples interval.