Scripts for manipulating timing data stored in GW frame files.
So far, using these should just involve placing them in a folder in your path and then calling the commands by name. Most of them take arguments involving start/end times and possibly channel names.
To see detailed usage instructions, run the command followed by an
Checks and Plots for EVNT Log and Timing Witness Paper
After an event, you will need to generate a bunch of timing plots. Generate the IRIG-B decoded times and plots with:
irig-b-decode-commands.py -t GPSTIME -g GRACEID
Make DuoTone statistics plots showing the zero-crossing at the time of the event as well as in the surrounding 10 minutes for each site:
duotone_delay.py --stat --ifo H1 -t GPSTIME duotone_delay.py --stat --ifo L1 -t GPSTIME
Make overlay plots for IRIG-B and DuoTone as well as zero-crossing plots for the DuoTone in the +/-15 minutes surrounding the event:
geco_overlay_plots.py -t GPSTIME
Here is how I have been using these scripts to e.g. dump all files for October 2015.
First, dump the slow channel data (using the
-l flag to specify the
slow-channel list) from the entire month of October to the
~/oct directory (each channel will get its own directory within
-p is not specified, then
~ will be used by default):
geco_dump -s 'Oct 1 2015' -e 'Nov 1 2015' -p ~/oct -l slow
You could also just specify a couple of channels you want to dump if you don't need all of the timing slow channels:
geco_dump -s 'Oct 1 2015' -e 'Nov 1 2015' -p ~/oct \ "H1:SYS-TIMING_C_MA_A_PORT_2_SLAVE_CFC_TIMEDIFF_1" \ "H1:SYS-TIMING_X_FO_A_PORT_9_SLAVE_CFC_TIMEDIFF_1"
If you have already dumped data from some of these frame files, they will not be re-dumped. The time spent dumping should not depend on the number of slow channels, since each channel dump gets its own separate process.
While the dump is in progress, you can monitor how many frame files
out of the total have been dumped so far by adding the
to your previous query:
geco_dump -s 'Oct 1 2015' -e 'Nov 1 2015' -p ~/oct -l slow -P
If you need to print a comma-separated timeseries to
(instead of saving each 64-second interval produced by
framecpp_dump_channel to its own file without editing, which is the
default behavior), you can specify the
-T flag (for "timeseries"):
geco_dump -s 'Oct 1 2015' -e 'Nov 1 2015' -T "H1:SYS-TIMING_C_MA_A_PORT_2_SLAVE_CFC_TIMEDIFF_1"
The output of this command is a comma-separated timeseries suitable for
immediate reading into any plotting script, or piping into another utility.
Note that you can only do this with one channel at a time, for obvious
If you have already dumped this time range to files, you can get the
same comma-separated timeseries read straight from text files for a big
speed-up using the
-R flag rather than the
-T flag. This is, of
course, also useful if you are running the script off of datagrid, and
you don't have any frame files handy:
geco_dump -s 'Oct 1 2015' -e 'Nov 1 2015' -R -p ~/oct "H1:SYS-TIMING_C_MA_A_PORT_2_SLAVE_CFC_TIMEDIFF_1"
Note that you have to tell the script where the files were saved originally;
in this case, assuming we already ran the first example command, they would
be saved to the
Finally, once you have dumped the files, you can change to the
directory and zip them all up with:
After dumping all that data. you should probably zip up your output directories to make it easier to transfer the data elsewhere. Compressing the data will speed up file transfers, but more importantly, representing all of the files in your output directories as a single zipped archive will make file transfer much faster (downloading tens of thousands of small files will take many, many times longer than downloading a single large file, even if the total amount of data is nominally the same). You can do this like so:
zip -r output_file_name.zip input_directory
You can then do something clever like, say, uploading your data to a remote
scp or something like the pure-