|
This page is gives some usage information regading the PEGASUS Worklow Management System. If you would like to obtain more information abut this software, please contact burston [at] mps.mpg.de.
Before embarking on running the Track'n'Map pipeline yourself, please make sure that there is no existing data ingested into the DRMS that you could use before making possible duplicate records under a different series name. There is no benefit to this.
Setting up
Download this ".pegasusrc" file to your home directory.
First you need to verify you can use pegasus
>show-properties
(problems?)
Edit/create your ".pegasusrc" file.
Pay attention to this setting:
pegasus.catalog.replica.file
Get grid certification
>grid-cert-request
Read the screen and follow the instructions. (Problems?)
Initiate your grid proxy (lasts for 12 or 24 hours, I can't remember).
>grid-proxy-init
Make a directory where you will run the pipeline from and download this tar package and unpack the files (maptrack.jsd, params.in, CreateDAX.java).
Compile the javascript
> javac CreateDAX.java
Now you should be set up to go and not have to repeat any of the above (except the grid-proxy-init).
Running the pipeline
1. Create a series (if the one you want to add to doesn't exist already).
For this you need to edit the maptrack.jsd file
You need to modify these parameters:
Seriesname:
Author: Owner: #======= Segments =======
Data:
The parameters in "Data:" must match the params.in file in the next step.
The "Seriesname" field can be anything after the punkt, but before that must be "mps_yourusuername"
(to delete the series: >delete_series seriesname )
Also, just check the type of data in the "Data:" line (float, double, int etc) and the value of the "Retention". The Rentention is measured in days and is how long before the series (or only the records??) is deleted.
Now create the series
>create_series maptrack.jsd
(Problems?)
2. Modify the params.in file
daxfile=MapTrackDAX.dax --- don't touch this
input_series=hmi_test.V2_45s --- name of the input series you want to get data from
seg_name=Dopplergram --- the segment name of the data (continuum, magnetogram)
cube_series=mps_schunker.V2_mag1_512x512x1919 --- the name of the output series (this must exist - if not create one in step 1.)
t_start=2010.04.04_00:53:15_TAI --- the start time of the data you want to use
t_end=2010.04.15_03:39:45_TAI --- OBSOLETE!
missing_frames=b --- fill any missing frames with blank (there is also an interpolation option - not tested and not recommended)
cadence=45 --- cadence of the data (seconds)
workflowname=trackmap --- don't touch this
namespace=MPS --- don't touch this
mapping_exec=drms_mapping --- don't touch this
mapping_version=50.0 --- make sure this is the correct version (check Updates)
cube_exec=drms_cube_care --- don't touch this
cube_version=1.0 --- make sure this is the correct version (check Updates)
rcfile=/some/directory/ofyours/rc.data --- this should point to where your rc.data file will live
lfmaindir=/another/directory/ofyours/Inputs/ --- this points to where your mapping lists will be dumped
projection_type=postel --- type of projection, for now use only Postel
n_trac_regions=1 --- the number of tracked regions in
n_trac=1919 --- how many frames per cube - MUST BE ODD
n_spg=-960 --- how much to overlap the cubes in time. Negative is overlap, positive puts gaps in between (ignored if n_trac_regions=1)
n_phi=1 --- OBSOLETE (may be used in future)
n_lambda=1 --- OBSOLETE (may be used in future)
phi_cen_first_box=2.268928 --- longitudinal coordinates of centre of mapping at central time (radians)
lambda_cen_first_box=-0.122173 --- latitudinal coordinates of centre of mapping at central time (radians)
nx=512 --- size of output map in x-direction/number of columns (pixels)
ny=512 --- size of output map in y-direction/number of rows (pixels)
dx=0.0005 --- scale of pixels in output map in x-direction/number of columns (radians)
dy=0.0005 --- scale of pixels in output map in y-direction/number of rows (radians)
a0=-0.02893 --- constant of differential rotation profile (for more information on tracking rates see here)
a2=-0.3441 --- coefficient for sin^2 of differential rotation profile
a4=-0.5037 --- coefficient for sin^4 of differential rotation profile
spec=eqcentre --- specified tracking, see below
jid=6 --- job ID (should be unique, keep track of your files in "lfmaindir" argument above )
(Questions?)
2.1 Possible tracking options
centre: Tracks centred on the sub-observation point of HMI at the middle time. Ignores the "phi_cen_first_box" and "lambda_cen_first_box" parameters. Goes to the frame at the middle of the time range and gets OBS_L0 and OBS_B0, it then tracks forward and backward in time to beginning and end of the cube at the rate specified by the tracking coefficients (a0,a2,a4) at OBS_B0.
eqcentre: Tracks centred on the meridian at the equator at the middle time. Ignores the "phi_cen_first_box" and "lambda_cen_first_box" parameters. Goes to the frame at the middle of the time range and gets OBS_L0, it then tracks forward and backward in time to beginning and end of the cube at the rate specified by the tracking coefficients (a0,a2,a4) at the equator (ie. latitude is 0 degrees).
norm: this is 'normal' tracking. It tracks at the latitude specified in "lambda_cen_first_box" centred at the middle frame forward and backward in time to beginning and end of the time series. It does NOT check if it is on the visible side of the Sun.
fixed: Each frame throughout the time series is mapped centred on the given coords ("phi_cen_first_box","lambda_cen_first_box").
-----
SHORT VERSION
centre: calculate the coordinates at disk centre for the central time frame, and then track forward and backward in time for the longitudes.
eqcentre: calculates the longitude of the middle frame and tracks for this keep latitude = 0.
norm: take given coords (phi_cen_first_box,lambda_cen_first_box) for middle frame and calculate the rest depending on tracking coefficients
fixed: keep the fixed given coords (phi_cen_first_box,lambda_cen_first_box) throughout the tracking
-----
(Questions?)
3. Run the java script to create a dax file using params.in
>java -classpath .:$CLASSPATH CreateDAX params.in
Then follow the instructions printed to the screen (yes, read the screen) there are three more steps to get the pipeline running and instructions how to monitor the progress of your pipeline.
(Problems?)
Goodluck!
While you are waiting for it to finish please read about cleaning up or you will shortly run out of disk space.
Known bugs.
From the command line:
seismo_cluster> drms_mapping inlist=my_map.list outlist=my_cube_care.list -d
seismo_cluster> drms_cube_care list=my_cube_care.list ds=mps_schunker.V_test_33 t_start=2010.04.10_04:00:00_TAI t_end=2010.04.10_20:00:00_TAI a0=-0.02893 a2=-0.3441 a4=-0.5037 cadence=45 -b
|
