toby burnett university of washington university of … · university of washington university of...
TRANSCRIPT
1
Toby Burnett University of Washington
University of Pavia
Physics Colloquium 13 May 2011
Launch of Fermi (née GLAST) June 11, 2008
Understanding the gamma-ray sky: 2+ years of Fermi data
U. Pavia 13 May 2011
The conception: launch and deployment of CGRO: April 5, 1991
Inspired Bill Atwood (SLAC at the time) and Peter Michelson (Stanford) to realize that there was a much better design for a gamma-ray telescope.
Start of HEP/Astrophysics, or DOE/NASA collaboration
2
EGRET
Bill Atwood Peter Michelson
U. Pavia 13 May 2011
2011 Rossi prize
The 2011 Rossi prize is awarded to Bill Atwood, Peter Michelson, and the Fermi Gamma-Ray Space Telescope/LAT team for enabling, through the development of the Large Area Telescope, new insights into neutron stars, supernova remnants, cosmic rays, binary systems, active galactic nuclei, and gamma-ray bursts.
3
Bill Atwood Peter Michelson
U. Pavia 13 May 2011
Final phase
4 U. Pavia 13 May 2011
Launch!
5 U. Pavia 13 May 2011
6
25.6 Degree Orbit
SAA
http://observatory.tamu.edu:8080/Trakker/
Circular orbit, 565 km altitude (96 min period), 25.6 deg inclination U. Pavia 13 May 2011
7
GN
HEASARC
-
-
DELTA
7920H •
White Sands
TDRSS SN
S & Ku
LAT Instrument
Science
Operations Center
(SLAC)
GBM Instrument
Operations Center
GRB
Coordinates Network
• Telemetry 1 kbps •
- •
S
Alerts
Data, Command Loads
Schedules
Schedules
Mission Operations
Center (MOC)
Fermi Science
Support Center
• m sec •
•
•
Fermi Spacecraft
Large Area Telescope
& GBM GPS
Fermi MISSION ELEMENTS
U. Pavia 13 May 2011
After 1 day First data, 25 June, L+14
½ hour, 417 photons
The very first photons
8 U. Pavia 13 May 2011
August 27: First light press release!
9 U. Pavia 13 May 2011
Observation modes
Pointing (traditional telescope) • Continuous stare, if not obscured
• Otherwise: – Follow Earth limb (CGRO strategy)
– Slew to a secondary point
– Survey mode
Survey mode – our standard – Look away from Earth, but rock
toward poles to equalize north/south
– Full sky coverage every 3 hours.
10
Exposure for one year
Maximum deviation 20%
Note that south pole is less
due to South Atlantic Anomaly
N
S
U. Pavia 13 May 2011
11
Fermi gamma-ray
U. Pavia 13 May 2011
Primack et al
No significant attenuation below ~10 GeV.
A dominant factor in EBL models is the star
formation rate -- attenuation measurements
can help distinguish models.
Photons with E>10 GeV are attenuated by the diffuse field of UV-
Optical-IR extragalactic background light (EBL)
Lower-energy gamma-rays are special
EBL over cosmological distances
is probed by gammas in the 10-100
GeV range.
In contrast, the TeV-IR attenuation
results in a flux that may be limited
to more local (or much brighter)
sources.
only e-t of the original source flux reaches us
12 U. Pavia 13 May 2011
Fermi-LAT collaboration
United States • California State University at Sonoma • University of California at Santa Cruz - Santa Cruz Institute of Particle Physics • Goddard Space Flight Center – Laboratory for High Energy Astrophysics • Naval Research Laboratory • Ohio State University • Stanford University (SLAC and HEPL/Physics) • University of Washington
France • IN2P3, CEA/Saclay
Italy • INFN, ASI
Japanese GLAST Collaboration • Hiroshima University • ISAS/JAXA, RIKEN • Tokyo Inst of Technology
Spain • ICREA and Inst de Ciencies de l’Espi
Swedish GLAST Collaboration • Kalmar University • Royal Institute of Technology (KTH) • Stockholm University
13
PI: Peter Michelson (Stanford & SLAC)
~270 Members (including ~90 Affiliated
Scientists, plus 37 Postdocs, and 48
Graduate Students)
Cooperation between NASA and DOE, with
key international contributions from
France, Italy, Japan and Sweden.
Managed at Stanford Linear Accelerator
Center (SLAC).
U. Pavia 13 May 2011
2-years of LAT data
14 U. Pavia 13 May 2011
Back to gamma rays: a journey from very close to very far away
• The closest source:
15
e+
e–
Incoming positron
Annihilate here (micrometeorite shield) to two gammas)
U. Pavia 13 May 2011
Our angular resolution, or PSF
16
3 Orders of magnitude
Dashed line is a measurement using the data. Current public representation is wrong by x2 at high energies. This misunderstanding plays a role in at least two external “discoveries”
U. Pavia 13 May 2011
x30
N
S
W E
Looking down at the brightest source
17
At 650 km, The
Earth covers
40% of the sky
Earth-based coordinate system
Horizontal
pointing
Zenith angle
Limb is forward scattering from top of atmosphere: fits well to cosmic ray spectrum
nadir limb
U. Pavia 13 May 2011
About representing the data
• Pixels and projections: How to represent the entire sky? – Tesselation: the HEALpix scheme
– Fermi issues: • the resolution depends on energy! (next slide on PSF)
The UW-invented sparse representation allows compact storage of data binned in energy and angle
• Large dynamic range: how to show contrast, especially with diffuse background: we adopt a scheme to show photon density, using PSF weighting
• Pseudo-color schemes:
• Convert a number to a color. But what are the colors supposed to convey? – Scale function (linear, sqrt, log)
Offset, bias in mapping important
18
B
SLS
heat
grey U. Pavia 13 May 2011
How are they generated?
19
Acceleration (shock waves)
Synchroton Inverse compton
Magnetic field
Radiation field
e-
p
ISM: H I, H II, H2, dust
Hadronic interaction
0
Dark Matter
Dark matter annihilation
? U. Pavia 13 May 2011
WMAP and Fermi
20 U. Pavia 13 May 2011
Goal: account for every photon
• ~Steady Sources of photons – Point sources
• Pulsars, including binaries
• Galaxies, mostly AGN – AGN probes B, photons
• Unknown
– Galactic diffuse
– Isotropic diffuse • Unresolved point sources
• Proton background
• Unknown
• Transients – GRB
– Nova
21 U. Pavia 13 May 2011
Making the 2FGL catalog
U. Pavia 13 May 2011 22
Understanding the sky: extract a list of sources for the 2FGL catalog
Data: 28 M
Exposure: 52 Ms
Two years (excluding 3 GRBs) “Pass7 processing” [1FGL: 11 months]
2FGL Table 1888 entries [1FGL: 1451]
23
Light curves, SED plots, associations
+ U. Pavia 13 May 2011
Bright sources at high latitudes are easy
U. Pavia 13 May 2011 24
Strong source high latitude example: SED plots
Circles are 3, 100 MeV PSF (‘front’ section) (varies by a factor of 30 with energy!)
Use log parabola if better fit
Pulsars fit with exponential cutoff
25
Power law
Measuring point source properties: maximize likelihood
Model of the sky must account for all photons PSF Aeff Galactic, isotropic diffuse including CR 1/8 degree grid, pixels centered on plane Earth limb
An important issue: how to measure significance? Test Statistic: TS=-2log(Lfit/Lnull)
We conservatively choose only sources with TS>25.
26
TS=14 TS=32 TS=1000
null
max
TS/2
flux
Log(L
)
Detecting sources
Initial list: use 1FGL supplemented with: PGWave, MRfilter: image based
MST (minimal spanning tree): pattern of high energy photons
Likelihood, using current model Test each of 3.1 M 0.1 deg pixels with a trial point source,
(index 2), record the corresponding TS.
Look for clusters with max TS>10, use as seed for new iteration
Efficiency checks: Test with Monte Carlo – generated sources
Recheck by applying PGWave
27
The 2FGL process: Stage I
Use the pointlike tool applied between 100 MeV and 300 GeV. Initial list: use 1FGL as seeds supplemented with:
PGWave, MRfilter: image based MST (minimal spanning tree): pattern of high energy photons
Create model with point and extended sources, galactic, isotropic and limb diffuse Require exponential cutoff for all LAT pulsars, use log parabola if
improves fit Fit the entire sky including updating positions (procedure detailed
below) Create 'residual TS' map:
TS for new point source (index 2) at each of 3.1M 0.1 pixels Look for clusters with max TS>10, use as seeds in new model
Provide all with TS>10 to stage II
28
CenA Lobes
MSH 15-52
Vela X
IC 443
LMC
SMC W28 W30 HESS J1825-137
W44 W51C
Cygnus Loop
Extended source templates
Details about the sky model
Tessellate sky using HEALPix: 1728 regions
Each ~5 square pixel defines:
Center of circular regions for:
data (5 deg)
sources (10 deg)
sources inside are varied; those outside fixed to results of previous iteration
Note ~x3 overlap of data: not independent
Diffuse component normalizations free
Iteration procedure:
Each region fit (full likelihood maximized) independently
Each fit remeasures point source positions: Apply updates between cycles.
Check changes in log(L): iterate until none changes by more than 10 (8-10 iterations required)
30
Colors: HEALPix index
Localization
Basic principle: the likelihood function, as a function of the position of a source, is an estimator of the position, with the curvature defining the resolution.
31
Demo with 3C 273.
PSF: 3 @ 100 MeV (front)
PSF: 0.1 for E>10 GeV (front)
Error ellipse defined by 95% contour (2.45). Plot shows contours, and results of fit to quadratic surface
‘TS plot’
galactic 67.3%
isotropic 23.7%
limb 0.5%
sources 8.5%
2FGL skymodel
Stage I Summary
32
Example consistency check: all photons in 5 radius circle (Approx. 12 d.o.f.)
Contributions for all energies, full sky
Ch
i squ
ared
Consistency mostly good
Limb distribution
Free parameters
Type Number
Spectra 7603
Diffuse normalization 3456
Location ( 2 per TS>16 source)
5096
It is not all so rosy…
Most sources apparently associated with diffuse structures probably result from inadequate representation of the diffuse itself
33
Galactic center is complicated! See talk by T. Porter
Orion molecular cloud: poster by S. Digel & F. Giordano
Sources: TS>10 seeds for 2FGL
Sources: did we miss any?
34
TS residual map, threshold=10, ~3 (catalog: 25)
U. Pavia 13 May 2011
2FGL process: Stage II
Use the standard tool gtlike in binned mode Accept all 3499 TS>10 sources from stage I Similar iterative scheme, but with 933 overlapping
regions chosen to equalize number of sources Refit spectra and diffuse normalization Fit pulsars with exponential cutoff
Everything else with power law: try log parabola Retain all with gtlike TS>25 when generating model
The pointlike fit and TS generally agree well, but some scatter for low values.
Generate: Table spectral plots light curves associations
35
TS comparison
pointlike
gtlik
e
Discussion
277 1FGL sources are not represented. Recall that they were only used as seeds at the start of the new process Some reasons:
New requirements for localization
Extended sources were represented by more than one point source
Improved galactic diffuse model
There, but not significant enough (flared during first 11 months)
Dominant effect is new fitting procedure: application of current procedure (especially now using binned vs. unbinned gtlike) to 1FGL accounts for nearly all.
The 5 LAT pulsars that are not (DC) detected are put in ‘by hand’
36
2FGL Associations
37
Classifications
38
Type Number Percentage of total
Active Galactic Nuclei 832 44%
Candidate Active Galactic
Nuclei
268 14%
Unassociated 594 32%
Pulsars (pulsed emission) 86 5%
Pulsars (no pulsations yet) 26 1%
Supernova
Remnants/Pulsar Wind
Nebulae
60 3%
Globular Clusters 11 < 1%
Other Galaxies 7 < 1%
Binary systems 4 < 1%
TOTAL 1888 100%
Very Preliminary - Work Still In Progress
Pulsars
39
radio emission cone
-ray
emission
fan beam
Discovered 1967 (Nobel prize 1974) Vela discovered 1968
U. Pavia 13 May 2011
88 gamma-ray pulsars
U. Pavia 13 May 2011 40
Pulsars seen from above
U. Pavia 13 May 2011 41
Vela: published results with first 2.5 months
42
Remarkably
sharp peaks;
features to
~0.3ms.
Turns nearly
completely off
between the
double pulses.
•<2.8% of phase-
averaged pulsed
emission, 95%
confidence
•Stringent limits or
measurement will be
available with more
livetime
Radio pulse
89 ms
U. Pavia 13 May 2011
Vela: phase-averaged SED
43
( / )
0( )b
cE EN E N E e
0.051.51
0.04
2.9 0.1 GeVcE
Consistent with b=1
(simple exponential)
b=2 (super-exponential)
rejected at 16.5
No evidence for magnetic
pair attenuation:
Near-surface emission
ruled out
U. Pavia 13 May 2011
CTA-1: Discovery of the first gamma-ray-only Pulsar with “blind search” technique
44
Quick discovery enabled by new
blind-search technique (Atwood et al ApJ (2009))
P ~ 317 ms Pdot ~ 3.6E-13
• Spin-down luminosity ~1036 erg s-1, sufficient to supply the PWN with magnetic fields and energetic electrons.
• The γ-ray flux from the CTA 1 pulsar corresponds to about 1-10% of Erot (depending on beam geometry)
1420 MHz Radio Map:
Pineault et al., A&A 324, 1152 (1997)
27 total discovered so far U. Pavia 13 May 2011
From our pulsar catalog paper
45 U. Pavia 13 May 2011
Millisecond pulsars and Fermi
46 U. Pavia 13 May 2011
See http://nanograv.org 47 U. Pavia 13 May 2011
Looking for MSP’s: the effect of Fermi
48 U. Pavia 13 May 2011
Supernova Remnants: Origin of Cosmic Rays?
49
“How fitting it is that Fermi
seems to be confirming the bold idea advanced over 60 years ago by the scientist after whom it was named,” Roger Blandford,
U. Pavia 13 May 2011
Nearby galaxies (including our own!)
U. Pavia 13 May 2011 50
Our Nearest Non-blazar AGN
2011 May 10 Fermi Symposium - Cheung
M. Su’s talk yesterday
Our Nearest Non-blazar AGN
2011 May 10 Fermi Symposium - Cheung
53
Diffuse Emissions in Other Galaxies
Test Statistic (TS) = -2 log(Lsource - Lno source)
0.68, 0.95, 0.99 confidence level localization contours
Appear as LAT point sources, starburst regions unresolved
Galactic diffuse, isotropic diffuse, and point sources subtracted
Fermi LAT (>200 MeV) Fermi LAT (>200 MeV)
6o x 6o region of the sky 6o x 6o region of the sky
U. Pavia 13 May 2011
Angle on Classification of Radio Galaxies
Radio Galaxies, Lobe-dominated Sources Blazars
2011 May 10 Fermi Symposium - Cheung
from Urry & Padovani (1995)
55
Blazars
(from Sikora, Begelman, and Rees (1994))
3C273 Identified as distant in 1962 by Greenstein and Schmidt (added quasar to the language)
U. Pavia 13 May 2011
Blazars are variable on monthly scales
56
PKS 1502+106
3C273
CTA 1
PKS 2155-304
3C454.3
Vela
BL Lac Crab
U. Pavia 13 May 2011
3C454.3 flare!
57
• Well-known radio source, quasar
at z = 0.859;
also detected by EGRET, AGILE
• Not a simple power law
• Can describe as a broken
power law with a break, 1 ~
2.3 to 2 ~3.5 at Ebr ~ 2 GeV
•Origin of the break?
U. Pavia 13 May 2011
Multi-wavelength: where we fit in
58
radio IR
UV X-ray
Fermi
A previous multiwavelength
radio IR
UV X-ray
Fermi
U. Pavia 13 May 2011
59
PKS 1502+106 • z=1.84 • Extremely rapid flare just as we started!
U. Pavia 13 May 2011
PKS 2155-304: 11-day multiwavelength campaign
• BL Lac @ z=0.116
60
ATOM RXTE
Swift
Fermi
HESS
EGRET
U. Pavia 13 May 2011
The galactic diffuse emission
61
Used for 1FGL: the “gll_iem_v02” version
Current one includes the “lobes”
Largest component: H I Contribution by ring
62
Looking away from the center
63 U. Pavia 13 May 2011
Isotropic diffuse
64 U. Pavia 13 May 2011
(Not really flat: have to take proton contamination into account)
LAT Isotropic Diffuse Flux
|b| > 10º
extragalactic diffuse
CR background
LAT
arXiv: 1002.3603
65 U. Pavia 13 May 2011
Not just unresolved blazars!
66
See arXiv:1003.0895
U. Pavia 13 May 2011
Explosions/Flares: very near and very far
67 U. Pavia 13 May 2011
V407 Cyg: nova on March 11, 2010
68 U. Pavia 13 May 2011
V407 Cyg: in our 2-year catalog
69 U. Pavia 13 May 2011
High Energy Activity from the Crab
No corresponding flare in X-rays with
INTEGRAL (Atel # 2856), Swift (Atel #
2858, 2866), or RXTE (Atel # 2872) or
NIR (Atel #2867). No evidence for active
AGN near Crab (Swift, Atel # 2868). U. Pavia 13 May 2011 70
Crab Flares (>100 MeV)
U. Pavia 13 May 2011 71
Fermi-GBM
72
Charles Meegan (PI)
Jochen Greiner (Co-PI) U. Pavia 13 May 2011
GBM Trigger Rate (weekly)
73
129 GBM detected GRBs
• 2 SGRs (SGR 0501+4516, SGR 1806-20)
• 1 AXP (AXP 1E1547.0-5408)
• >5 TGFs (Terrestrial Gamma-ray Flashes )
U. Pavia 13 May 2011
GRB080916C (AKA collaboration meeting burst)
74
8 keV – 260 keV
260 keV – 5 MeV
LAT raw
LAT > 100 MeV
LAT > 1 GeV
T0
• The first low-energy peak is not observed at LAT energies
• 14 events above 1 GeV
• The bulk of the emission of the 2nd peak is moving toward later times as the energy increases
• Clear signature of spectral evolution
• new era of GeV GRB lightcurves!
GROND optical follow up [GCN 8257, 8272] Faint (21.7 mag at T
0+32h) and
fading (T0+3.3d) source
RA = 119.8472˚, Dec = −56.6383˚ (±0.5” at 68% C.L.)
Photometric redshift:
z=4.2 +/- 0.3
U. Pavia 13 May 2011
Some implications of high-energy photons at z=4.35
Bulk Lorentz factor: =608, 887 for time bins b and d
Problem for predictions of EBL absorption?
13.2 GeV observed within 16.5 s
Quantum gravity limit: MQG > 1.3 x 1018 GeV
75 U. Pavia 13 May 2011
Fermi GRBs
GBM: ~250 GRB/yr LAT: ~9 GRB/yr
76 U. Pavia 13 May 2011
THE DATA ARE PUBLIC
77 U. Pavia 13 May 2011
Did Hooper and
Goodenough discover
DM in our data?
78
"When I look at this data, it lines up perfectly," he says. "It quacks like a duck."
U. Pavia 13 May 2011
79
And did Ando and Kusenko discover “primordial magnetic fields” by detecting halos around AGN’s that we missed?
U. Pavia 13 May 2011
Summary • Many discoveries
• But: After two years, we are still trying to understand both the performance of the detector, and subtleties in the sky!
• Dark matter still the holy grail
80 U. Pavia 13 May 2011
The UW group: Key ideas
Basic object model for sources, all-sky
Data representation: variable size HEALPix, stored as sparse list of pixels containing photons. Ideal for all-sky analysis: one compression step good for
entire sky: for 2 years, 3.4 GB (scales with time) 41 MB (~constant), factor of 83.
But needs extraction of subsets for timing
Use for display, easy to weight photons
Likelihood analysis optimized for speed, without loss of precision vs. unbinned technique. Use likelihood technique for source finding: According to
Cramer-Rao thm, the best way!
Allows quick localization
81 U. Pavia 13 May 2011
Multi-wavelength context
82
High energy photons often produced in a different physical process to the lower energy emission -> Independent handle on the physical conditions.
Gamma-rays are produced in non-thermal processes
Known classes of gamma-ray sources are multiwavelength objects seen across much of the spectrum
Fermi LAT
High energy gamma-rays explore nature’s accelerators - “Where the energetic things are”
natural connections to UHE cosmic-ray and neutrino astrophysics
U. Pavia 13 May 2011