Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 22 Next »

VCSBeam version

Required: >= v3.2.1

# On Garrawarla, test vcsbeam version
$ module use /pawsey/mwa/software/python3/modulefiles
$ module load vcsbeam
$ make_mwa_tied_array_beam -V
MWA Beamformer v2.18.8_99df2a4

Calibration

Offline Correlator

$ cd /astro/mwavcs/vcs/1320499816/cal/1320499816/vis
$ for freq in {109..132}
do
    sed "s/FREQ/${freq}/" < offline_correlate.sbatch > offline_correlate_${freq}.sbatch
    sbatch offline_correlate_${freq}.sbatch
done
offline_correlate.sbatch 
#!/bin/bash -l

#SBATCH --nodes=1
#SBATCH --mem=370gb
#SBATCH --partition=gpuq
#SBATCH --gres=gpu:1
#SBATCH --time=01:00:00
#SBATCH --account=mwavcs
#SBATCH --export=NONE

module use /pawsey/mwa/software/python3/modulefiles
module load offline_correlator

for t in {8..599}
do

    gpssecond=$(echo "$t + 1320499816" | bc)
    gpubox=$(echo "FREQ - 108" | bc)

    INPUT_DATA_FILE=/astro/mwavcs/vcs/1320499816/combined/1320499816_${gpssecond}_chFREQ.dat
    START_SECOND=${gpssecond}
    DUMPS_PER_SECOND=1 # This sets the output time resolution
                    # (e.g. 20 --> 1/20 = 0.05s = 50 ms)
                    # Minimum allowed resolution is 2 ms
    CHANS_TO_AVERAGE=1 # This sets the output frequency resolution
                   # (e.g. 1 --> 1x10 kHz = 10 kHz)
    GPUBOX_CHANNEL_NUMBER=${gpubox} # This should be chosen to "match" the input channel
                         # This is not easy! (mwalib handles this, but at the
                         # moment, offline_correlator is not using mwalib)
    OUTPUT_PREFIX=1320499816 # Output files begin with this

    srun -N 1 -n 1 offline_correlator \
        -d ${INPUT_DATA_FILE} \
        -s ${START_SECOND} \
        -r ${DUMPS_PER_SECOND} \
        -n ${CHANS_TO_AVERAGE} \
        -c ${GPUBOX_CHANNEL_NUMBER} \
        -o ${OUTPUT_PREFIX}

done

Hyperdrive

Attempt to run Hyperdrive for in-beam calibration (failed!)

$ cd /astro/mwavcs/vcs/1320499816/cal/1320499816/hyperdrive
$ sbatch hyperdrive.sbatch


/astro/mwavcs/vcs/1320499816/cal/1320499816/hyperdrive/hyperdrive.sbatch 
#!/bin/bash -l
#SBATCH --job-name=hyp-$1
#SBATCH --output=hyperdrive.out
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=40
#SBATCH --time=01:00:00
#SBATCH --clusters=garrawarla
#SBATCH --partition=gpuq
#SBATCH --account=mwavcs
#SBATCH --export=NONE
#SBATCH --gres=tmp:50g
#SBATCH --gres=gpu:1

module use /pawsey/mwa/software/python3/modulefiles
module load hyperdrive

set -eux
which hyperdrive

#cd /astro/mwaeor/MWA/data/1090008640
METAFITS=1320499816_cal.metafits

if [[ ! -r srclist_1000.yaml ]]; then
   hyperdrive srclist-by-beam -n 1000 -m ${METAFITS} /pawsey/mwa/software/python3/srclists/master/srclist_pumav3_EoR0aegean_fixedEoR1pietro+ForA_phase1+2.txt srclist_1000.yaml
fi

hyperdrive di-calibrate -s srclist_1000.yaml -d ../vis/*gpubox*.fits ${METAFITS}

Unfortunately, this doesn't work on the gpubox files as written out by the legacy offline_correlator, which doesn't write out most of the metadata info that mwalib needs to correctly parse those files. I'll work on this route later.

On MWAX_CORRELATOR observation 1320412440 (SUCCESS!)

At the moment, Harry is still helping me obtain the files. As it turned out, this obs was already downloaded for Cath and the EoR team, so Harry was able to just copy them to a place I can see them on Garrawarla: /astro/mwavcs/asvo/252007. Then, running them through Hyperdrive is straightfoward:

$ cd /astro/mwavcs/vcs/1320499816/cal/1320412440/hyperdrive
$ sbatch hyperdrive.sbatch
/astro/mwavcs/vcs/1320499816/cal/1320412440/hyperdrive/hyperdrive.sbatch 
#!/bin/bash -l
#SBATCH --job-name=hyp-$1
#SBATCH --output=hyperdrive.out
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=40
#SBATCH --time=01:00:00
#SBATCH --clusters=garrawarla
#SBATCH --partition=gpuq
#SBATCH --account=mwavcs
#SBATCH --export=NONE
#SBATCH --gres=tmp:50g
#SBATCH --gres=gpu:1
 
module use /pawsey/mwa/software/python3/modulefiles
module load hyperdrive
 
set -eux
which hyperdrive
 
#cd /astro/mwaeor/MWA/data/1090008640
METAFITS=/astro/mwavcs/asvo/252007/1320412440.metafits
 
if [[ ! -r srclist_1000.yaml ]]; then
   hyperdrive srclist-by-beam -n 1000 -m ${METAFITS} --tile-flags 57 /pawsey/mwa/software/python3/srclists/master/srclist_pumav3_EoR0aegean_fixedEoR1pietro+ForA_phase1+2.txt srclist_1000.yaml
fi
 
hyperdrive di-calibrate -s srclist_1000.yaml -d /astro/mwavcs/asvo/252007/*.fits ${METAFITS}

This outputs hyperdrive_solutions.bin. The --tile-flags 57 corresponds to TileName HexE2. It is used because doing it without shows the amplitudes for that tile are much 

Using these solutions in vcsbeam requires >= v2.19.13.

Plot Hyperdrive solutions
$ mwa_plot_calibration -m 1320499816_legacy.metafits -c 1320412440.fits -C hyperdrive_solutions.bin -R NONE -U 0,0 -O -X -z nan > solution.txt
$ plot_calibration.py --phases_png phases.png --amps_png amps.png solution.txt


Weirdly, even though I flagged HexE2 in Hyperdrive (corresponding to Antenna #57), it still shows up in these solution plots. This might be a bug in Hyperdrive. Fornow, I'll just also flag it when I try to beamform. Fingers crossed this solution is still suitable for my observation!

RTS (failed!)

Create source list

Create source list
$ cd /astro/mwavcs/vcs/1320499816/cal/1320499816
$ module load srclists/v1.0.0
$ wget -O 1320499816_metafits_ppds.fits http://ws.mwatelescope.org/metadata/fits?obs_id=1320499816&include_ppds=1
$ srclist_by_beam.py -m 1320499816_metafits_ppds.fits -n 1000 -s ${SRCLISTS_DIR}/srclist_pumav3_EoR0aegean_EoR1pietro+ForA.txt

Set up initial RTS job

First, we need a calibration metafits file (mode = 'HW_LFILES'), which for the moment I edited by hand. It doesn't matter whether the file is one with PPDs or without, but apparently some script somewhere is expected the file to be called \[obsid\]_metafits_ppds.fits. So I manually edited the (MWAX VCS) metafits file and just changed the mode, and then made a symbolic link to is called /astro/mwavcs/vcs/1320499816/cal/1320499816/rts/1320499816_metafits_ppds.fits. Then I ran the script from VCSTools:

Run the RTS
cd /astro/mwavcs/vcs/1320499816/cal/1320499816/rts
$ calibrate_vcs.py -O 1320499816 -o 1320499816 -m 1320499816_metafits_ppds.fits -s ../srclist_pumav3_EoR0aegean_EoR1pietro+ForA_1320499816_patch1000.txt --offline --nosubmit

There's no good reason I couldn't have left off the --nosubmit option and had it run, but I just wanted to check it worked before trying it.

Run the RTS

Run the RTS
$ cd /astro/mwavcs/vcs/1320499816/cal/1320499816/rts
$ sbatch /astro/mwavcs/vcs/1320499816/batch/RTS_1320499816.batch


/astro/mwavcs/vcs/1320499816/cal/1320499816/rts/RTS_1320499816.batch 
#!/bin/bash -l

#SBATCH --export=NONE
#SBATCH --output=/astro/mwavcs/vcs/1320499816/batch/RTS_1320499816.out
#SBATCH --account=mwavcs
#SBATCH --clusters=garrawarla
#SBATCH --partition=gpuq
#
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=10240MB
#SBATCH --nice=0
#SBATCH --chdir=/astro/mwavcs/vcs/1320499816/cal/1320499816/rts
#SBATCH --time=0:45:00
#SBATCH --nodes=13
#SBATCH --cpus-per-gpu=1
#SBATCH --gres=gpu:1

ncpus=1
export OMP_NUM_THREADS=1


module use /pawsey/mwa/software/python3/modulefiles/
module load vcstools/master

module load RTS/master


cd /astro/mwavcs/vcs/1320499816/cal/1320499816/rts
export UCX_MEMTYPE_CACHE=n
srun --export=all -N 25 -n 25 rts_gpu /astro/mwavcs/vcs/1320499816/cal/1320499816/rts/rts_1320499816.in
/astro/mwavcs/vcs/1320499816/cal/1320499816/rts/rts_1320499816.in 
ReadAllFromSingleFile=
BaseFilename=/astro/mwavcs/vcs/1318345216/cal/1318345216/vis/*_GPUBOX
ReadGpuboxDirect=0
UseCorrelatorInput=1

ReadMetafitsFile=1
MetafitsFilename=/astro/mwavcs/vcs/1318345216/1318345216

DoCalibration=
doMWArxCorrections=1
doRawDataCorrections=1
doRFIflagging=0
useFastPrimaryBeamModels=0
generateDIjones=1
applyDIcalibration=1
UsePacketInput=0
UseThreadedVI=1

MaxFrequency=170.24
ObservationFrequencyBase=138.895
ObservationTimeBase=2455846.1252314816
ObservationPointCentreHA=-0.35952189154909175
ObservationPointCentreDec=-5.98757073636307
ChannelBandwidth=0.04
NumberOfChannels=32

CorrDumpsPerCadence=128
CorrDumpTime=1.0
NumberOfIntegrationBins=6
NumberOfIterations=1

StartProcessingAt=0

ArrayPositionLat=-26.7033194
ArrayPositionLong=116.6708152
ArrayNumberOfStations=128

ArrayFile=

SourceCatalogueFile=/astro/mwavcs/vcs/1318345216/cal/1318345216/srclist_pumav3_EoR0aegean_EoR1pietro+ForA_1318345216_patch1000.txt
NumberOfCalibrators=1
NumberOfSourcesToPeel=0
calBaselineMin=20.0
calShortBaselineTaper=40.0
FieldOfViewDegrees=1

For an as yet unknown reason, this RTS job timed out, even after allowing 4 hours of wall time.

Beamforming

Beamforming directly from MWAX (which does the fine PFB on GPU memory) requires >= 3.2.1.

Using the Hyperdrive solution (1320412440)

Run beamforming job
$ cd /astro/mwavcs/vcs/1320499816/vcsbeam/coh
$ wget -O 1320412440_metafits_ppds.fits http://ws.mwatelescope.org/metadata/fits?obs_id=1320412440&include_ppds=1
$ sbatch vcsbeam.sbatch
/astro/mwavcs/vcs/1320499816/vcsbeam/coh/pointings.txt 
00:34:08.8703 -07:21:53.409
/astro/mwavcs/vcs/1320499816/vcsbeam/coh/flagged_tilenames.txt 
HexE2
/astro/mwavcs/vcs/1320499816/vcsbeam/coh/vcsbeam.sbatch 
#!/bin/bash -l

#SBATCH --nodes=24
#SBATCH --ntasks-per-node=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=370gb
#SBATCH --partition=gpuq
#SBATCH --gres=gpu:1
#SBATCH --time=01:00:00
#SBATCH --account=mwavcs
#SBATCH --nice=0
#SBATCH --export=NONE
 
module use /pawsey/mwa/software/python3/modulefiles
module load vcsbeam
 
srun -N 24 -n 24 make_mwa_tied_array_beam \
    -m /astro/mwavcs/asvo/252057/1320499816.fits \
    -b 1320499824 \
    -T 592 \
    -f 109 \
    -d /astro/mwavcs/asvo/252057 \
    -P pointings.txt \
    -F flagged_tilenames.txt \
    -c 1320412440_metafits_ppds.fits \
    -C /astro/mwavcs/vcs/1320499816/cal/1320412440/hyperdrive/hyperdrive_solutions.bin \
    -p -R NONE -U 0,0 -O -X
Run beamforming job
$ module load singularity
$ /pawsey/mwa/singularity/presto/presto.sif prepfold -psr J0034-0721 -nosearch -nodmsearch C001_1320499816_00:34:08.87_-07:21:53.41_ch109-132_000?.fits

***(Needs to be updated)***

Single Channel

Changing all the number of nodes "24" to "1" in the sbatch script above produces the following:

A good single pulse for testing purposes

Run beamforming job
$ cd /astro/mwavcs/vcs/1320499816/vcsbeam/coh
$ sbatch vcsbeam_single_pulse.sbatch
/astro/mwavcs/vcs/1320499816/vcsbeam/coh/vcsbeam_single_pulse.sbatch 
#!/bin/bash -l

#SBATCH --nodes=24
#SBATCH --ntasks-per-node=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=370gb
#SBATCH --partition=gpuq
#SBATCH --gres=gpu:1
#SBATCH --time=01:00:00
#SBATCH --account=mwavcs
#SBATCH --nice=0
#SBATCH --export=NONE
 
module use /pawsey/mwa/software/python3/modulefiles
module load vcsbeam
 
srun -N 24 -n 24 make_mwa_tied_array_beam \
    -m /astro/mwavcs/vcs/1320499816/vcsbeam/1320499816_legacy.metafits \
    -b 1320500016 \
    -T 1 \
    -f 109 \
    -d /astro/mwavcs/vcs/1320499816/combined \
    -P pointings.txt \
    -F flagged_tilenames.txt \
    -c 1320412440_metafits_ppds.fits \
    -C /astro/mwavcs/vcs/1320499816/cal/1320412440/hyperdrive/hyperdrive_solutions.bin \
    -p -R NONE -U 0,0 -O -X
$ module load singularity
$ /pawsey/mwa/singularity/dspsr/dspsr.sif dspsr -c 1 -D 0 -b 1024 -turns 1 -U 600 -A C001_1320499816_00:34:08.87_-07:21:53.41_ch109-132_0001.fits
$ singularity run -B ~/.Xauthority /pawsey/mwa/singularity/psrchive_tempo2/psrchive_tempo2.sif pav -G -f 4 2021-11-09-13\:29\:57.ar


Offline FinePFB

$ cd /astro/mwavcs/asvo/252057
$ wget -O 1320499816.fits http://ws.mwatelescope.org/metadata/fits?obs_id=1320499816
$ cd /astro/mwavcs/vcs/1320499816/combined
$ for freq in {109..132}
do
    sed "s/FREQ/${freq}/" < make_legacy.sbatch > make_legacy_ch${freq}.sbatch
    sbatch make_legacy_ch${freq}.sbatch
done
make_legacy.sbatch 
#!/bin/bash -l

#SBATCH --nodes=1
#SBATCH --cpus-per-task=1
#SBATCH --ntasks-per-node=1
#SBATCH --mem=370gb
#SBATCH --partition=gpuq
#SBATCH --gres=gpu:1
#SBATCH --time=01:00:00
#SBATCH --account=mwavcs
#SBATCH --export=NONE

module use /pawsey/mwa/software/python3/modulefiles
module load vcsbeam
module load openmpi-ucx-gpu

DATAPATH=/astro/mwavcs/asvo/252057

srun -N 1 -n 1 fine_pfb_offline -m $DATAPATH/1320499816.metafits -b 1320499816 -T 600 -f FREQ -d $DATAPATH

Incoherent Beam

(Currently not tested since v2.18. Results not guaranteed.)

$ cd /astro/mwavcs/vcs/1320499816/vcsbeam/incoh
$ sbatch vcsbeam_incoh.sbatch
/astro/mwavcs/vcs/1318345216/vcsbeam/incoh/vcsbeam_incoh.sbatch 
#!/bin/bash -l

#SBATCH --nodes=24
#SBATCH --ntasks-per-node=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=370gb
#SBATCH --partition=gpuq
#SBATCH --gres=gpu:1
#SBATCH --time=00:30:00
#SBATCH --account=mwavcs
#SBATCH --nice=0
#SBATCH --export=NONE
 
module use /pawsey/mwa/software/python3/modulefiles
module load vcsbeam
 
srun -N 24 -n 24 make_mwa_incoh_beam \
    -m /astro/mwavcs/vcs/1320499816/vcsbeam/1320499816_legacy.metafits \
    -b 1320499816 \
    -T 600 \
    -f 109 \
    -d /astro/mwavcs/vcs/1320499816/combined \
    -o incoh

After beamforming finishes, fold on B0031-07:

$ module load singularity
$ /pawsey/mwa/singularity/presto/presto.sif prepfold -psr J0034-0721 -nosearch -nodmsearch -start 0.01 incoh*.fits

  • No labels