Pulsar tools in Di FX Adam Deller ASTRON

Outline l l l What mpifxcorr does in “pulsar” mode How to configure pulsar

Di. FX’s pulsar toolkit l Di. FX can: l l l Apply incoherent dedispersion

Incoherent dedispersion l Calculate and correct for a delay in each frequency channel based

Incoherent dedispersion l Calculate the delay in each frequency channel based on cold plasma

Incoherent dedispersion l Convert that delay to a phase offset based on the pulsar

Incoherent dedispersion l Combine with a pulse phase prediction polynomial (“polyco”) to calculate the

Incoherent dedispersion l Bin 0: 0 -0. 2 Bin 1: 0. 2 -0. 8

Incoherent dedispersion l Delays are independent of antenna, because all antennas are delayed to

Location l l Occurs in Core, at the accumulate stage If weighting+summing bins, this

How to specify? l vex 2 difx: put “binconfig = ” line in for

The binconfig file NUM POLYCO FILES: 1 POLYCO FILE 0: J 1650 -1654. bd

The polyco file l Generate using tempo (beyond scope here) 1650 -1654 26 -MAR-12

Post-correlation l l If scrunching, simply run difx 2 fits as normal, you get

Questions? Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Slides: 20

Download presentation

Pulsar tools in Di. FX Adam Deller ASTRON 6 th Di. FX workshop, CSIRO ATNF, Sydney AUS

Outline l l l What mpifxcorr does in “pulsar” mode How to configure pulsar operations Useful scripts and tools available for setting pulsar phase bins Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Di. FX’s pulsar toolkit l Di. FX can: l l l Apply incoherent dedispersion to visibilities Accumulate visibilities into an arbitrary number of pulse bins of arbitrary width Optionally, weight and sum the pulse bins (usually used for gating) Build a 1 separate FITS file per phase bin (if not summing bins) NOTE: only cross-correlations are binned! Autocorrelations are not (but see profile mode later) Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Calculate and correct for a delay in each frequency channel based on cold plasma dispersion Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Calculate the delay in each frequency channel based on cold plasma dispersion and provided DM 1400 MHz Subband X, time Y 85 ms 1416 MHz 83 ms DM = 40 period = 2 ms Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Convert that delay to a phase offset based on the pulsar period 1400 MHz 85 ms phase offset = 0. 5 Subband X, time Y 84 ms phase offset = 0. 0 1416 MHz 83 ms phase offset = 0. 5 DM = 40 period = 2 ms Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Combine with a pulse phase prediction polynomial (“polyco”) to calculate the pulse phase in every channel at time Y (as an example here = 0. 25) 1400 MHz 85 ms phase offset = 0. 5 + phase at infinite frequency from polyco = total phase = 0. 75 Adam Deller Subband X, time Y 84 ms phase offset = 0. 0 + phase at infinite frequency from polyco = total phase = 0. 25 1416 MHz DM = 40 period = 2 ms 83 ms phase offset = 0. 5 + phase at infinite freq. = total phase = 0. 75 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Bin 0: 0 -0. 2 Bin 1: 0. 2 -0. 8 Bin 2: 0. 8 -1. 0 Frequency channel by frequency channel, compare phase with provided bin edges and allocate each visibility to the appropriate bin 1400 MHz Subband X, time Y 1 1 1 1 0 0 85 ms phase offset = 0. 5 + phase at infinite frequency from polyco = total phase = 0. 75 Adam Deller 84 ms phase offset = 0. 0 + phase at infinite frequency from polyco = total phase = 0. 25 1416 MHz 0 2 2 2 1 DM = 40 period = 2 ms 83 ms phase offset = 0. 5 + phase at infinite freq. = total phase = 0. 75 6 th Di. FX workshop, CSIRO ATNF

Incoherent dedispersion l Delays are independent of antenna, because all antennas are delayed to the same reference point for correlation (the geocenter) l Polyco must be made at the location “geocenter”! Usually site code 0 in tempo. Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Location l l Occurs in Core, at the accumulate stage If weighting+summing bins, this is performed at the end of a subint. (minimizes the data return to Fx. Manager) Adam Deller 6 th Di. FX workshop, CSIRO ATNF

How to specify? l vex 2 difx: put “binconfig = ” line in for a setup SETUP gate. J 1543 -0620 { t. Int = 2. 0 n. Chan = 32 do. Polar = false n. FFTChan = 128 subint. NS = 80000000 max. NSBetween. ACAvg = 1000000 stride. Length = 16 bin. Config = bd 152 rd. J 1543 -0620. binconfig } Adam Deller 6 th Di. FX workshop, CSIRO ATNF

The binconfig file NUM POLYCO FILES: 1 POLYCO FILE 0: J 1650 -1654. bd 152 s 0. polyco NUM PULSAR BINS: 2 SCRUNCH OUTPUT: TRUE BIN PHASE END 0: 0. 200000 BIN WEIGHT 0: 0. 0 BIN PHASE END 1: 0. 220000 BIN WEIGHT 1: 1. 0 Note: bins wrap around phase 1. 0/0. 0! SCRUNCH = TRUE; so adding the weighted bins This is a simple gate (on/off) SCRUNCH = FALSE would save bins separately POLYCO points to the phase prediction file Adam Deller 6 th Di. FX workshop, CSIRO ATNF

The polyco file l Generate using tempo (beyond scope here) 1650 -1654 26 -MAR-12 83000. 00 56012. 35416666660 43. 181643 -0. 922 -9. 791 2190534. 037742 0. 571574515888 24 720 12 1400. 000 0. 27151830610250283 D-09 0. 31600661389421161 D-02 -0. 79172395063399383 D-08 -0. 76435402690880627 D-13 0. 18973528859963467 D-17 0. 60967896576050514 D-20 -0. 40545690801398583 D-22 -0. 14857173038742444 D-24 0. 37620021307428399 D-27 0. 14111375354325646 D-29 -0. 11811239363850021 D-32 -0. 45609416626443307 D-35 1650 -1654 26 -MAR-12 203000. 00 56012. 85416666660 43. 181643 -0. 918 -9. 745 2215228. 327944 0. 571574515888 24 720 12 1400. 000 -0. 24833931330618675 D-09 0. 31485469662028221 D-02 -0. 80814027761202229 D-08 -0. 75698057683234006 D-13 0. 22121795226919268 D-17 -0. 79089737421091686 D-21 -0. 43291672066653833 D-22 0. 21487126523725907 D-25 0. 37519768978778538 D-27 -0. 66910007411822625 D-31 -0. 11151335487911077 D-32 -0. 23552517403119568 D-36 Polynomial coefficients Phase at reference time Adam Deller Reference frequency 6 th Di. FX workshop, CSIRO ATNF

Post-correlation l l If scrunching, simply run difx 2 fits as normal, you get 1 FITS file out If saving the separate bins, must run difx 2 fits N times for N separate bins (using the -B <bin number> flag) l Generates N different FITS files Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Useful tools #1: profile mode l If your pulsar ephemeris is not extremely accurate - not good enough to predict absolute phase - you can use “profile” mode of mpifxcorr as a precursor to a real correlation l l Uses binning mode, not scrunching, and makes “cross-correlations” which are actually autocorrelations Divide the pulse period up into a number of bins, and make a pulse profile from the VLBI data itself Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Useful tools #1: profile mode. v 2 d file: vex = bd 152 rd. vex. obs antennas = SC, HN, NL, FD, LA, PT, KP, OV, BR, MK mode = profile Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Useful tools #1: profile mode Then run difx 2 profile on the Di. FX output, which sums the (faked) autocorrelations over all antennas for Bin# Weight 0 0. 0182229691407935 each bin 1 0. 0787451798095526 l Will generate a 2 0. 0223606968053615 “profile” text file 3 0. 0142714711684412 like this the one … on the right 41 -0. 0481090874037728 0. 0164508430016418 l Can be inspected 42 43 0. 640543867722647 with e. g. gnuplot 44 0. 359456132277354 45 0. 114425502494899 46 -4. 06530828972069 e-05 47 0. 06748867444299 48 -0. 0539940377225829 49 0. 0367702126740731 Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Useful tools #2: profile 2 binconfig l If you have a profile (either from Di. FX profile mode or [even better] elsewhere), you can convert this automatically to a Di. FX binconfig file: > profile 2 binconfig. py --profile=J 1234+5678. txt --polyco=J 1234+5678. polyco --nonormalise -n 24 --binconfigfile=J 1234+5678. binconfig [--lineskip=2 --profilecolumn=3 --dontzeronoise] Adam Deller 6 th Di. FX workshop, CSIRO ATNF

Questions? Adam Deller 6 th Di. FX workshop, CSIRO ATNF