News from
Bordeaux, with the
aim of avoiding duplication of effort and promoting shared efforts
17 October 2006 (D. Smith)
Likelihood Damien Parent added gtlikelihood
to his DC2 “loop over all pulsars” script and quickly found, like many before
him, that gtlikelihood is not plug n’ play. Thierry is helping him
master the tool. Tibolla’s DC2 and Tibaldi’s 16 September presentations have
been quite useful. Damien asked Max whether Damien should add a power law with
exponential cut-off to the list of gtlikelihood functions. Damien aims
to reconstruct the spectra for as many of the 44 DC2 pulsars that he detected
as possible, and to provide us with a “sensitivity for spectral measurements”
number along the way. Jean Ballet
answers during the VRVS meeting: “there exists a likelihood model
called ExpCutoff. It is
not very
convenient but it does what you want. Set parameter Ebreak to 1, P2 and P3 to
0. P1 is the cutoff energy. Examples are given in Luis Reyes' presentation at
DC2 closeout. I can send you an example XML file”.
D4.fits
Lucas Guillemot has read in the 1647 pulsars from ATNF and made a
D4.fits with them all. For the 255 gamma candidate pulsars, he calculates MJD
beyond the end of ephemeredes validity when the f, fdot uncertainties lead to
an e.g. 10 mP error on the number of turns(“dN”). The goal is to flag pulsars whose ephemeredes
are “spoiling”. He has plots showing that Roger’s VperY and this dN are complementary, as well they should be. He’ll try to post to
confluence before the next pulsar meeting. Denis Dumora has tutored him in
Tempo so that he better understands the variables he’s loading into D4.
Radio
Timing with Nançay They’re continuing – of the 141 pulsars
listed in the (accepted) proposal to Nançay, they are now tracking over 80.
Ismael sent us the names, dates, and signal-to-noise ratios for 5913
observations this morning. Lucas helped me digest them. Here is a tally:
Other news: 5
November is the deadline for us to re-ask for time for the next 6 months.
Absolute
Timing The NASA project office has approved our
test. I will go to Spectrum on 5 November, for a week, to set up the gear, then
return in mid-December for the actual test. Denis, who has done all the prep
work with me, will most likely not travel because non-US citizens are not
allowed at Spectrum without great trouble. The current weak point in our plans
is the GPS antenna+cable type available at Spectrum compared to those used by
the Bordeaux electronics. Neil J and Eric G are being proactive. Details at https://confluence.slac.stanford.edu/display/CAL/Event+Timestamps
On a separate
topic – Anders Borgland wrote code to take LsfContext variables from digi, calculate
our best time estimate, and load them into the Merit files so that they will
then show up in the .fits files. His code has not yet been propagated to a
working version of Engineering Release and so this calculation is still not
performed. Furthermore, I found an error in his algorithm last week: he had
followed a comment in a digiRootData include file which says that the GEM
scaler sets TimeTicks to zero for every new PPS ; however a scan of the
SvacTuple shows that it is not in fact zeroed. This leads to errors of more
than a second for many events.
Service
Challenge 1) Thierry has learned how to apply cuts to
a Merit root file and then generate a .fits file, so as to be able to run the
Science gt Tools (e.g. gtbary) on files to which all the background
rejection / gamma selection cuts have not been applied. The idea we’re pursuing
is to compare cut-by-cut efficiencies predicted by Monte Carlo with those seen
in the real data. We aspire to do this with the E&LO Vela data ; we’d like
such simulated data (Thierry is learning how to do it himself in the meantime).
In the meantime, we’ll be practicing with the DC2 data, using the Merit files
from the SLAC servers. 2) Suppose that
absolute timestamps were a little wrong once we get on orbit – would we
recognize the symptoms quickly? Would we have debugging tools? A constant time
or satellite position offset or jitter ; a bug in the barycentering code… Being
able to generate our own .fits files from Merit files means that we can fudge
the dates and explore the symptoms. Thompson points out during
the VRVS meeting that the absolute
phase study that Denis presented during DC2 is already a piece of the answer.
Posters
We submitted one to Texas in Australia, that Eric Grove will present for
us. It’s on pulsar candidate selection, and radio timing efforts. Details at https://confluence.slac.stanford.edu/display/SCIGRPS/Texas+in+Australia%2C+11-15+December+2006%2C+Melbourne
We’d further like
to submit a very similar one to the AAS meeting in Seattle, but the rules are
restrictive. If you know anyone who is going, would like to be our first
author, but doesn’t already have a presentation, please volunteer that person
for us.
Else, we’re
aiming for 2 or 3 posters on the above topics for the Glast Symposium.