Download Cygnus X3 - IceCube Neutrino Observatory

AstroPlasmas group
June 14, 2010
Mike Baker
I have made several different analyses looking for timing
structure for neutrino signals, I have a discussion of the
IceCube part and then a bit more about the astrophysics
in general
* Search for periodic neutrino emission from binary
systems
* Search for neutrinos in coincidence with GeV-TeV
photon flares
* Search for generic clustering in time and space of
neutrinos
Binary systems in the galaxy where one
member is a compact object haven been
posited as a source of cosmic rays. We
search for clustering as a function of orbital
period is a very similar analysis to searching
for event clustering in MJD. We performed
in AMANDA 7 yr, IC22 periodic searched
where we used the period of binary X-ray
stars and micro-quasars from stereoscopic
information and fit the phase.
We search for a best-fit
Gaussian describing the
events in phase. The
discovery potential and
AUL have the same trend
as as a function of time as
in the general
(untriggered) flare search.
I ran the analysis over the 7 sources, and these are the best-fit
parameters and pretrial p-values:
Source
Cygnus X3
SS433
Cygnus X1
LS I +61 303
GRS 1915+105
XTE J1118+480
GRO J0422+32
|
|
|
|
|
|
|
|
(Ra, Dec)
(308.107,40.958)
(287.957,4.983)
(299.591,35.202)
(40.132,61.229)
(288.798,10.946)
(169.545,48.037)
(65.428,32.907)
Period
0.199679
13.0821
5.5929
26.498
30.8
0.16993
0.21214
|pretrialP|
| 0.00186 |
| 0.3523 |
| 0.0793 |
| 0.2271 |
| 0.4266 |
| 0.2784 |
| 0.0373 |
Minimum P-value: 0.00186146 Best Source: Cygnus X3
NSrc
4.27
2.15
1.39
1.83
2.03
2.40
2.95
Gamma
3.75
2.89
1.54
3.95
3.95
2.35
3.06
mean
0.820
0.601
0.708
0.521
0.986
0.872
0.800
sigma_w
0.02
0.02
0.02
0.02
0.045
0.131
0.02
Comparing to scrambled samples, the smallest p-value from
all 7 objects is less than that found in data 1.8% of the time.
I think the result is interesting, but it is still compatable with
background.
Here is how the events look in phase for Cygnus X3 – the three
highest-weight events come in 1/20 of the period, which corresponds
to a 15-minute window of the 4.8 hour orbit.
I took the phase=0 information from the Fermi paper from Science,
and it looks like the maximum period is near 0.82. Phi=0 is at
superior conjunction (x-ray minimum) and phi=0.5 is at inferior
conjunction
One interesting thing
about Cygnus X3
during the IC-40
period (April 2008 –
May 2009) is that
there was a series
radio flares, with
accompanying high
states in GeV gamma
rays.
There is another
analysis geared
toward lower
energies looking into
neutrino coincidence
also.
“Saxophone” plot: relation
between ratio and soft xray fluxes varies
depending on the flaring
state.
Right are the periodic flux
information from Fermi (top)
and ASM (bottom). Fermi
sees periodicity only when
the source is in a high GeV
state, we haven't looked at
the raw times of our events
yet.
I also managed to find
a paper from the 1985
International Cosmic
Ray Conference with a
plot of muons seen
from a 3 degree cone
from the direction of
Cygnus X-3.
http://adsabs.harvard.edu/full/1985ICRC....9..445A
MWL Triggered Flare Search
Blazars are candidate sources for UHECRs, and hence neutrinos,
and exhibit varibility in HE photon wavebands on the order of days.
We use this MWL information on variability to motivate a cut in time
for a point-source search.
Two kinds of alerts, one from
sources providing
lightcurves with sporadic
data.
1ES1218+304 has
a strong flare seen
with VERITAS →
Markarian 501 also
has a short flare
← seen with VERITAS
The other kind of alert we have is from satellite sources
with comprehensive coverage in a waveband
Start out with a Maximum Likelihood Blocks reconstruction of
the Fermi LAT lightcurve.
Instead of making a guess of what the best cut on the flux is,
we maximixe with respect to the threshold, to find the best
answer to the question:
”At what flux of the lightcurve does neutrino emission turn on?”
threshold
Resulting Pdf
Here is the discovery potential for the lightcurve of PKS 1510-089.
I compare several methods, all using the lightcurve:
In blue is the hypothetical case where we know the best cut on
threshold and any lag.
In red is the case used in the analysis, where we fit the threshold
and allow a short lag as a method of interpolating the lightcurve.
I also tested a case where we fit the threshold and allow a search
for a long lag between photons and neutrinos, this is shown in
black.
The most significant source
found was one high-energy
event during a GeV flare of
PKS 1502+106, but it wasn't
very strong, the overall pvalue of the analysis is 29%
We've picked out
most of the highlyvariable sources for
the triggered
analysis, but it still
could be interesting
to include weaker
flares in a stacking
analysis, like in
GRB searches
← 3C279
Fermi also finds two distinct kinds of flaring behavior: for many of the brighter
sources the lightcurve displays complicated rising and lowering, others have
relatively simple, discrete flares. This means that there very different things
going on in the jet for different flares from different sources.
PKS 1454-354 →
Fermi also has their own method of fitting flares with 3-day binned
LC, while I used 1-day binned LC.
Untriggered All-Sky Flare Search
We know of transient astrophysical phenomena with timescales of millisecondsseconds (GRB) to minutes-days (AGN flares, SN afterglow). Observations in
photons give us candidates to test for correlations in triggered searches, but
these observations are not always continuous depending on the energy
range tested.
We have an interest in performing a general search which covers many orders
of magnitude of flare duration to find occurrences undetected or undetectable
with photon emission. We fit for the best mean and sigma of a Gaussian in
time to find the strongest flare from a given source location.
Here background events are in blue,
with 5 injected signal events in red.
The height of the line corresponds
to S/B ratio of the event in the
time-integrated analysis.
We find that by looking for a clustering of events in time we can dramatically
reduce the number of events needed to see a discovery compared to a timeintegrated search, and can make a detection which is below the upper limit in
the general search.
The method's performance is similar to that of the time-integrated
analysis even for long flares, opening up the possibility of looking from
GRB timescales to steady sources. For flat emission, only 10% more
signal is needed for discovery than a PS search even when we add two
degrees of freedom to the likelihood.
AGN
~Time-Integrated
GRBs
Another possible outcome is that we see something completely
unexpected, some sort of burst of neutrinos with no counterpart
in photons.
?????????
I also had a look at what the Discovery potential and upper limit
looks like in terms of fluence. A dedicated GRB analysis which
looks at a widening time window near the burst reports the average
fluence needed from 117 bursts as a function of the time window
around the burst.
Skymap (with all fit parameters) is available for download as a root file:
http://www.icecube.wisc.edu/~mfbaker/IC40/flares_g/tAllSkyBatch_UnblindFix.root
Hottest Spot:
-log10(estp): 4.67768 at 254.75 RA and 36.25 dec.
The mean is at: 54874.7 MJD and the flare width is 15 seconds.
The spectral index fit is: 2.15 and the fit nsrc is: 2
Fifteen seconds doesn't
stand out on year-long
timescales, so we zoom
in on the flare time to
see both events.
Comparing to scrambled skymaps, the hottest spot is in the
bulk of the distribution of the pretrial p-value at the hottest spot.
The final p-value is ~58%.
The hottest flare
corresponds to two
events which are 22
seconds apart. There is
a spatial separation of
2.01 degrees between
the events.
Paraboloid sigma = 0.60 deg
Nch 65
Mue En = 3.33e4
Paraboloid sigma = 1.96 deg
Nch 17
Mue En = 2.43e3
There may be more information about the flare – IceCube has a
program with ROTSE to take optical images of close doublets of
events, and this met the trigger criteria.
Finally, here is a zoom-in of the hottest flare, with both events.
The higher-energy event is closer the best p-value.
Markarian 501 is marked as the black star.