localizing gamma ray bursts via interplanetary-spacecraft

Craig Markwardt wrote in message ...
First of all I am glad that you are now supplying numbers and
percentages as its easier if we both refer to gcn data.
You have inspired me to do more detailed data breakdown and I have
done so by starting in Jan 2001 when HETE comes online and counted all
gcn postings and broken them into more datasets like IPN only,
IPN/HETE, HETE only etc..
And yes I find the data does not give the same poor 5-20 pere cent
ratio over the three years that I initially claimed for the last 9
months period but my following argument still works I believe as a
more detailed analysis as you prefer seems to unearth more new
evidence against IPN as it does support it.
Furthermore I still stand by my initial claim that in the last 9
months IPN has a very low about 10-20 per cent success ratio.Lower
than the 3 year average.My count gives11 searchesIPN/HETE and 2
counterparts found.And one of those two counterparts..grb030131 seems
to my reading of the relevent gcns that the counterpart was found
outside the IPN box but inside IBASA error box?
(Incidentally by `IPN-only` I mean a grb where a search is done where
there is only a IPN annulus box but no available HETE/IBIS/Bepposax
error box like grb020525)

Your supposition that the speed of light is varies is not
substantiated by observation. [refs. 1-4]

Your analysis appears to be highly suspect. From my own analysis, I
conclude that there are
IPN Solutions of any type 300 (since GCN 100)
IPN Box solutions 108
IPN Small box solutions 29 (50 sq. arcmin., night sky)
Optical / Radio searchs 21
Counterparts found 8

So, if we judge success by number of counterparts found per search,
it's approximately 8 / 21 = 38 %, with a Poissonian statistical
uncertainty given by sqrt(8) / 21, or
success ratio of = 38 +/- 13 %

I can agree roughly with this figure although I got 20 IPN searches
with 9 counterparts found since Jan 2001 when HETE came into service.
But I must add that of those 9 I counted,.. 2 of them(030131,011121)
seemed to have the counterpart observed *outside * the IPN error box
but inside the HETE/IBIS box. In other words should that number be...
22 IPN radio optical searches
7counterparts found (within the IPN error box)

Thats a 31 % success ratio for IPN if one discounts 030131 and 011121.
You probably understand the RA dec etc better than I do. But it does
seem to me that the counterparts were observed outside the IPN error
boxes in those cases.


(I admit I dont know enough about RA dec etc . RA is given in degrees
and sometimes in h,m,s. For instance 202.037 in one post becomes 13h
27m 56sin another whch makes it hard for me to to decipher .But if
you read all the gcn`s for these two bursts there are question marks
over these as IPN successes)

These are all GRBs where an independent IPN box solution existed,
regardless of any other localization (GRBs 000630, 000911, 000926,
001007, 020813, 990506, 991208, 991216). The constraints on night sky
and box size are relevant since ground observatories cannot observe
during the daytime, and most do not have large fields of view.

The same search in the HETE data shows:
HETE-2 localizations since launch 36
Optical / Radio searches 33
Counterparts found 14

Similar success ratio = 42 +/- 11 %

(GRBs 000323 000408 000429 000607 000616 000623 000811 000812 000820
000830 000911 000925 001004 001007).

Im not too sure about this figure above either . I have the following
results which are different from yours..
39 HETE localizations with radio optical etc followups/searches
23 counterparts found
59% success ratio for HETE


GRB`s where counterparts are found

010921,011130,011212,020124,020127,020305,020317,0 20331,020531,020812,
020903,021004,021211,020813,030115,030226,030323,0 30324,030328,030329,030418,030429,030528

Thus, to within the statistical limits, the fractional success rates
(as defined) for both IPN and HETE are consistent with each other.
HETE obviously has far more follow-up observations, since they have a
dedicated and coordinated follow-up observation team, while IPN does
not. Of course, the analysis also ignores the role that IPN had in
improving the error boxes of the HETE positions, which is substantial.
This improves the HETE results at the expense of the IPN.

AS you see above I contest and revise your figures to a 60% success
ratio for HETE as compared to a 30 % success ratio for IPN. Unless
you can show that 030131,011121 are within the IPN error box.

The `IPN only` success ratio over the last 3 years is 6 searches and
0 counterparts found ..ie a 0 per cent success rate. This compares
unfavourably with the HETE only ratio which is 24 searches and 17
counterparts found or a 70 % success ratio.
And seeing as you insist that the size of error box should be taken
into account to explain this low IPN only success ratio then by the
same token one must then say that where IPN has constrained or reduced
the HETE error box by a percentage then the IPN /HETE success ratio
should actually be better by the same percentage then the HETE only
ratio.
But it isnt! Its worse . There were 13 searches on IPN/HETE boxes with
only 6 counterparts found or a 42 % success ratio. This is
unfavourable compared to the 70% HETE-only rate but even more so if
you use your `area` parameters . In other words if the IPN annulus
reduces on average the HETE box by 50 % then the IPN HETE success
ratio should be double the percentage of the HETE only ratio. Instead
its only 1/2.
Furthermore if my model is correct when it predicts that if the IPN
annulus is incorrect it should lead to a reduction in the HETE success
rate by the same average IPN percentage reduction of the HETE error
box. So if IPN HETE has astronomers looking in only 50 per cent of the
HETE error boxes then my model predicts that should lead to a 50 per
cent reduction in IPN HETE successes(if my claim that c was variable
is right). And thats what the figures and data show.
HETE is 70 % and HETE IPN is 42 % Thats almost 1/2 for IPN/HETE when
it should be double the HETE only ratio and although I have only
guessed the average reduction in HETE error box by IPN I imagine my 50
per cent average isnt far off the mark
Also, it ignores that there are several GRBs which are optically and
radio "dark" and yet have X-ray afterglows localized to extremely high
precision with the Chandra X-ray Observatory, which would tend to
improve the success rates of both kinds of solutions.

Your claims are unsubstantiated. You ignored the fact that most of
the IPN solutions went unobserved by follow-up teams, and therefore
you woefully underestimated the success rate of IPN.


In addition, you still maintain that somehow the IPN solutions are
being manipulated, and yet every IPN solution that I am aware of to
date, has been consistent with a confirmed counterpart. You have not
provided a mechanism by which the solution could be manipulated.
Indeed, there is none, since it involves simple geometry and geometric
viewing constraints.


I am not privy to the IPN methods but I think that a laymans general
summary would be that the IPN staff take the different time/flux
lightcurves from the different detectors in each satelite. They then
have to approximate a rough trigger and overall peak shape and
calculate a time of arrival trigger point in UT for each satelitte for
that specific burst. THey do this for all available satlitees whether
its one two or three etc..
They then would use these calculated time of arrivals lets say 3 from
NEAR INTEGRAL and ULYSES. As each satelite is a rough known distance
and location from earth they then can calculate annuli rings sometimes
double rings based on the assumption that c is constant and the
different time of arrivals can give rough directions in `rings`
My contention is that where the error box is known from HETE or
Integral in most cases the rings do not overlap and the IPN staff
assume this discrepancy is due to innacurate data or maybe to a
misinterpretaion of trigger arrival times. If you have seen different
discriminator channel lightcurves it is obviuos how matching
lightcurves from different sources is a approximate science. These
discrepencies are admitted even in the Laros papers and in the IPN
website pages
I then say that the process by which they `manipulate` the results is
that each arrival time is tweaked until it gives a ring that overlaps
the known error box. Another tweaking mechanism would be to widen the
margin of error on the annuli ring.
To back up my claim I point to the massive discrepency between the 100
per cent success rate of localizing the error boxes as compared to the
0 per cent success rate of IPN annuli where no error boxes are
available to conform the annuli to.
Why is it that IPN has a 100 per cent success ratio in finding HETE
IBIS error boxes? Even if c were constant there would be some misses
and a maybe 80 % success rate. The 100 per cent ratio in question is
too good to be true especially in light of the poor IPN/HETE and IPN
only success ratios. And most damning in my opinion is the fact that
when IPN annuli reduce the area of HETE error boxes, instead of giving
a proportionate improvement to the success rate the rate actually
drops by almost 1/2. That shouldnt happen if IPN annuli were correct.
And as I mention earlier that drop is roughly proportional to the
reduced area of the new HETE/IPN error box implying that if the HETE
error box is correct then the HETE/IPN overlap is random and
indicative of annuli that are incorrect. I contend that one would get
the same success rate as HETE/IPN hybred overlap if one were to
randomly search in only `part` of the HETE error boxes total areas.
And if that `part` were the same as the average overall IPN reduction
of HETE error boxes in IPN/HETE localizations. (I guess roughly that
the average IPN reduction of HETE error box area by overlap MUST be
proportional to the 42% to 70% per cent ratio in success I mention
earlier between IPN/Hete and HETE only success ratios.It would be
interesting to see what IS the average IPN reduction in HETE error box
area size? My rough calculation using the gcn`s I where percentages or
size reductions are available is that it averages at about 40-60 per
cent )
Sean