Observations of Gamma-Ray Bursts with the Fermi-Large

Area Telescope

Vlasios Vasileiou

Laboratoire Univers et Particules de Montpellier

(CNRS/IN2P3 & Universite Montpellier 2)

for the Fermi LAT and GBM collaborations

2

Overview

●Gamma-Ray Bursts●The Fermi Gamma Ray Space Telescope●Scientific highlights from Fermi-LAT GRB observations●Systematic Studies

● Of LAT-detected GRBs: “The first Fermi-LAT GRB catalog”● Of non LAT-detected GRBs

●Non-GRB Science performed with GRBs and Fermi● Constraining Lorentz-Invariance Violation● Constraining the opacity of the Universe to high-energy

gamma rays.

Gamma Ray Bursts

Gamma-Ray Bursts

● Bright flashes of gamma rays ~ 1/day

– Brightest events in the gamma-ray sky.● Discovered accidentally by the U.S. Military

Vela Satellites in 1967 – announced in 1973.● Have cosmological distances (detected up to

redshift of 8.2).● Are distributed uniformly in the sky.

2704 BATSE Bursts

First detected GRB

Klebesadel et al. 1973

Prompt and Extended Emission

➢Prompt spiky emission➢Primarily observed in kev — MeV energies➢No preferred pattern in the lightcurves

➢Non-thermal spectra➢Typically follow empirical “Band” function

GRB 090916c, Science, Volume 323, Is, 5922, pp. 1688 - (2009)

➢Emission followed by a smooth afterglow➢Observed in X-rays, visible, IR, optical, and MeV/GeV➢Exponential decrease in intensity t-1, t-2

GRB090510 – De. Pasquale et al. ApJL 709 (2010) L146-L151

• Measured fluences in the 10-8 – 10-3 erg/cm2 range. • Imply an isotropic energy release E

iso~1051 - 1054 erg

• Comparable to the rest mass of the Sun (2x1054 erg)• Jetted (collimated) emission relaxes the energy output down

to ~ 1051 erg

Horvath 2002

Hardness Ratio vs Duration

Fluence vs Duration

Duration Distribution

Credit: Pete Woods, UAH & NASA/Marshall.

GRB Progenitors

● Plots imply the existence of two distinct populations ● Separation supported by other observables, e.g. galaxy types,

redshift distributions.● Short-hard GRBs → believed to be created by mergers of compact

binary systems such as NS-NS or BS-BH.● Long-soft GRBs → believed to be created by explosion of massive

rapidly-rotating stars.● Both systems end up in a massive rapidly-rotating black hole – torus

system that emits radiation from two relativistic jets.

The Fermi Gamma­Ray Space Telescope

LAT Collaboration~400 members

France IN2P3/LUPM Montpellier IN2P3/LLR Ecole Polytechnique IN2P3/CENBG Bordeaux CEA/Saclay CESR Toulouse

Germany MPI fuer extraterrestr. Physik

Italy INFN Bari, Perugia, Pisa, Rome, Trieste, Udine ASI INAF-IASF

Japan Hiroshima University ISAS/JAXA Tokyo Institute of Technology

Spain IEEC-CISC, Barcelona

Sweden Royal Institute of Technology (KTH) Stockholm University

United States Stanford University (SLAC and HEPL/Physics) UC Santa Cruz Goddard Space Flight Center Naval Research Laboratory Sonoma State University Ohio State University University of Washington University of Denver Purdue University – Calumet Yahoo Inc.

The Fermi Gamma-Ray Space Telescope

GBM CollaborationUSA

University of Alabama in HuntsvilleNASA Marshall Space Flight CenterLos Alamos National Laboratory

Germany MPI fuer extraterrestrische Physik

● Launched August 2008● Multi-national Collaboration● Carries two instruments on board

– Large Area Telescope (LAT)– Gamma-Ray Burst Monitor (GBM)

The Gamma-Ray Burst Monitor (GBM)

● 12 x Sodium Iodide (NaI)− 8keV – 1MeV− Used for burst triggering and localization− Source direction inferred from the

relative rates in each of the hit NaI detectors.

● 2 x Bismuth Germanate (BGO)− 150keV – 40MeV − Provide spectral overlap with the LAT

● Detector arrangement provides coverage to the whole unocculted by the earth sky (8sr).

● Together the GBM detectors provide broad 5-decade-in-energy spectral coverage of GRBs.

The Large Area Telescope

Pair-conversion gamma-ray detector● Tracker

➢ Measures the direction (and energy), primary particle ID.● Imaging Calorimeter

➢ Measures the primary's energy➢ Images the shower (helps with energy/direction rec.)

● Segmented anti-Coincidence Shield➢ Identifies background of charged Cosmic Rays

● Performance✔ 20MeV – >300GeV✔ Wide field of view (2.4sr at 1GeV)✔ Full sky coverage every ~3h (2 orbits)✔ Large effective area (8000 cm2 at 1GeV on axis)✔ Good angular resolution (~0.2o at 1GeV)

GBM + LAT

● GBM+LAT=Wide field-of-view coverage of the sky in a broad 7-decade-in-energy spectral range

● Usual method of operation:

– GBM triggers

• Gives few-degree-accurate localization for LAT to observe

• If event bright or hard → also requests repoint of Fermi → allows for on-axis sensitive observations by the LAT

– If LAT detects at E>100MeV

• publishes more accurate ~0.3-1o (sys+stat) localization through GCN

• Swift & ground-based telescopes observe → hopefully detect an afterglow → hopefully a redshift

• In addition, the LAT continuously performs blind searches for GRBs both on board and on ground.

Automatic Repointing Towards GRB090902B

● Red cross → GRB090902B● Red/Green lines → The LAT ● White points → Detected events● White circle → LAT Field of View● Dark gray region → Earth's shadow● Yellow dot →Sun

The Sun

GRB090902B

LAT's FOV

The Earth's Shadow

The LAT

Events

Automatic Repointing Towards GRB090902B

● Red cross → GRB090902B● Red/Green lines → The LAT ● White points → Detected events● White circle → LAT Field of View● Dark gray region → Earth's shadow● Yellow dot →Sun

Detections as of 090904

Fermi GRB detection statistics

• The GBM detects ~250 GRBs/year– ~half in the LAT FoV

• The LAT detects ~10 GRBs/year (35 total)– ~8% of GBM GRB in LAT's FOV observed– 26 bursts above 100 MeV →

• ~1/2 with more accurate followup localisations by Swift and ground-based observatories– Swift XRT/UVOT, GROND, Gemini-S, Gemini-N, VLT

• most of the bursts in this half resulting to a redshift measurement – Largest z: 4.35 for GRB 080916C, smallest z: 0.74 for GRB090328

• 3 joint Swift/GBM/LAT prompt detections to date

Map as of April 6th 2010

Scientific Highlights from Fermi­LAT GRB 

Observations

?GBM NaI

LAT – decay as t-1.2+-0.2

GRB 080916C – our first bright GRB

GBM ← | → LAT

GBM BGO 260keV-5MeV

LAT E>100MeV

GBM BGO 260keV-5MeV

GBM NaI 8keV-260keV

LAT E>1GeV

Delayed emission appears as a spectral hardening

Abdo et al. Science 2009, 323, 1688

LAT emission starts delayed and persists longer (up to 1.4ks) with respect to GBM emission.

Highest-energy photon detected: 13GeV at 16.5s~70GeV in the GRB frame (z=4.35)➔Constrains dependence of c on E

γ

Minimum bulk Lorentz Factor (γγ opacity arguments): Γ

min=887+-21 and 608+-15 for bins b and d

Minimum emission radius: R

min≈Γ

min2cΔt/(1+z)=9x1055 cm for bin b.

Brightest Eiso

ever measured:

4.3x1054erg (20keV–2MeV) and 9x1055 (10keV-10GeV)➔implies a very narrowly-collimated jet.

?

?

Additional power-law component First time detected in a Short GRB Starts delayed in the prompt and persists up to ~200s Dominates at E>100MeV and E<20keV

Highest Epeak

=~4MeV ever for a time-integrated spectrum.

Highest-E photon detected ~31GeV at 0.83s post-trigger Sets Γ

min=1200 – 1000 interval c

Sets strongest constraints on dependence of c on Eγ

GRB090510 – our bright short GRB

De Pasquale et al. 2010Ackermann et al. 2010 ApJ 716

LAT E>1GeV

LAT E>100MeV

BGO 260keV-5MeV

NaI 8keV-260keV

GRB090510 – Prompt Emission Models

● Non-thermal leptonic emission – Synchrotron and Self-Compton Synchrotron (SSC)✗ If magnetic field strong → SSC component too weak to explain LAT observations– If magnetic field weak SSC component strong→

✗ Strong internal absorption secondary MeV emission ~simultaneous with keV → →disagrees with observed 0.2s delay.

✗ Model also faces synchrotron “line-of-death problem” (Band =-0.5+-0.07 > -2/3)α

● Forward-shock from the early afterglow – Synchrotron emission – Peak of the LAT emission == deceleration time of the relativistic blast wave.

✗ Model requires too high (~1) radiative efficiency for synchrotron emission– Models fail to explain the lower-energy (E<20keV) extension of the PL

● Hadronic models – Photohadronic and proton/ion synchrotron processes induce EM cascades secondary e→ -e+

pairs emit through synchrotron and Inverse Compton✗ Large makes photopion efficiency low Γ → requires too large energy release

✗ Stronger magnetic field higher photopion efficiency → → predicts softer LAT spectrum✗ Proton synchrotron in strong magnetic field

• Delayed onset == time to accelerate, accumulate, and cool the ultrarelativistic p.✗ Requires a very collimated beam (=~1o)→ such strong beaming not found in S.GRBs

See our GRB090510 paper (Ackermann et al. ApJ 2010 716) for more discussion and references.

GRB 090902B – Highest Energy ever detected

● Highest-energy photon detected ever from a GRB:

– 33.4 GeV at 82s from z=1.822

– 94 GeV in the GRB frame!● Also this is the second GRB after 090510 (a

short GRB) in which the PL extended at lower energies.

Abdo et al. 2009, ApJ 706L, 38A

GRB 090926A – The first observed cut-off PL

• Extra power-law:• Starts delayed and persists at longer times (5ks).• First time ever a cutoff on the extra PL observed.

• Significant at bin c – sharp spike• Marginally significant at bin d• Permits direct measurement of Γ=~200–700

• Sharp spike at bin c• It peaks at all energy ranges synchronized (<50ms)

and with similar widths → Implies PL and Band related; (co-located or otherwise causally correlated) ?

LAT E>1GeV

LAT E>100MeV

LAT All events

BGO 260keV-5MeV

NaI 14.3keV-260keV

NaI 8keV-14.3keV

Ackermann et al. 2009

?

•Forward Shock (FS) as the jet propagates in the external medium• Onset time == time required for forward-shock to sweep up enough material and brighten• Hard to explain rapid HE variability observed in some bursts (e.g. GRB090926A)• Requires large Γ (larger than that of GRB090926A) or a dense circumburst medium✗ Synchrotron? → cannot explain correlated LCurves (e.g. GRB090926A)• IC of Band photons by HE electrons at the FS? → possible & can explain correlated LCurves

• Hadronic models (pair cascades, proton synchrotron)• Late onset == time to accelerate protons & develop cascades• Proton synchrotron radiation (requires large B-fields)• Synchrotron emission from secondary e± pairs produced via photo-hadron interactions

• Can naturally explain the low energy extension of the PL✗ Scenarios require substantially more energy (1-3 orders of magnitude) than observed✗ Hard to produce correlated variability at low- and high-energies (e.g. spikes of GRB 090926A)

•Leptonic models (inverse-Compton or SSC)✗ Hard to produce a delayed onset longer than spike widths✗ Hard to produce a low-energy (<50 keV) power-law excess (as in GRB 090510, 090902B)✗ Hard to account for the different Band α and of the HE component spectral index.• But photospheric emission models could explain these properties

● SSC during late internal shocks● Thermal photosphere made by the powerful relativistic wind● Magnetic reconnection in Poynting-flux dominated outflows

● These are just some of the models.. for more see discussions in our papers

More models for the extra component

The First LAT Catalog of Gamma­Ray Bursts

The Fermi-LAT GRB Catalog

● First systematic study of GRB properties at high (E>20MeV) energies. ● Covers a 3 year period starting from August 2008 (32 detections)➢ Will include tabulated data describing important GRB parameters

– Usual GRB properties:

• Duration, average flux, peak flux, time of the peak flux, fluence

– High-energy extended-emission parameters:

• Temporal decay slope, spectral evolution, start/end time

– Prompt emission parameters:

• Delayed onset of the LAT emission, spectral evolution & components➢ Includes discussions on the unique properties of individual bursts (extra spectral

components, HE spectral cut-offs, analysis caveats).➢ Includes details on the tools and methods involved in the analysis. ➢ To be submitted soon

Extended Emission

PRELIMINARY

GRB090510

[s]

GRB090902B

[s]

GRB080916c

[s]

● Flux decays as a power law in time.– Power-law temporal decay: f(t)µ ta with a ~ -1 – -2 – Radiative or adiabatic fireball (Ghisellini et al. 2009)– No obvious breaks or other features.

GBM Duration

Extended Emission - Spectra

● No obvious spectral-evolution pattern.

● Spectral index typically averages around -2

GRB090902B[s]

GRB080916c[s]

GRB090510

[s]

Extended vs Prompt Emission Spectral Index

●ΓEXT

→ Average Spectral index of

extended emission●β → Spectral index of Band function

in the prompt phase ●Prompt and extended phase spectra

not correlated

βα

Energetics

● Long GRBs (filled dots): LAT fluence typically between 10% and 1% of GBM fluence● Short GRBs (open dots): LAT>GBM fluence● We detected 4 exceptionally bright bursts: GRB 080916c, 090510, 090902B, 090926A● They do not appear bright because they are systematically closer to us.

LAT=GBM

LAT=0.01xGBM

Highest-Energy Photon Detected

●LAT-detected emission frequently reaches several-tens-of-GeV energies (in the GRB frame).

●Good signs for the detection prospects of VHE observatories.

Systematic Study of non LAT­detected GRBs

Explaining the absence of LAT detections

● ~Half of GBM GRBs happen in the LAT FOV, however only ~10% are detected at E>100MeV.

– We investigated why we didn't detect the rest 90%. ● 1st step: Estimated fraction of GRBs that should had been detected.

– Calculated LAT (0.1-10GeV) ULs (over GBM duration) and compare with predictions from extrapolation of the GBM-fit (9keV–40MeV).

– Sample: GBM bursts that occurred in the LAT FOV with no LAT detection (161) & Band function is the preferred model (-20%): Total 126 bursts.

– Result: Number of GRBs with predicted flux>LAT UL (i.e. should have been detected) is ~50%

PRELIMIN

ARY

2nd step – Examine “bright BGO sample”

● Sample: 30 GBM bursts with Δβ<0.5, rate>75cts/s in the BGO, and no LAT detection

● Similar to the larger sample: rate of GRBs with higher predicted LAT flux ~50%.

– Bright sample representative of the parent sample for purposes of this work

● Performed joint GBM+LAT spectral fits

– HE Spectral index β becomes considerably softer in the joint fits → fraction of detectable but not detected bursts down to ~23%. Not detectable by LAT

Detectable by LAT

3rd step: Repeat joint fits with modified spectral model

● Added spectral softening in the model between BGO & LAT (exponential cutoff or a step function at 50 MeV) and repeated joint fits.

● The extra softening significantly improved the fit in 6/30 (20%) of these bursts → they require some form of spectral softening at tens of MeV energies.

● Rest 80% of the bursts consistent with just a softer β.● Note: Softening vs cutoff ↔ constant vs variable Γ?

GBM only Joint GBM-LAT Joint+step function at 50MeV

LAT non-detections – ULs on Γ

● Assuming that the spectral cutoffs in these 6 GRBs are because of internal opacity effects, we can set ULs on the bulk Lorentz Factors of their jets.

● We only know the redshift for 091127 so we set Γmax

(z) for the rest.

● Results to be published soon.

Standard LAT data

LLELLE

BGO

NaI

Spectral cutoffs & the LLE Event Class

● Standard LAT event selections (“Transient” class) run out of effective area at E<100MeV.

● “LAT Low Energy” (LLE) event selection → Very relaxed set of cuts → plenty of statistics in the tens-of-MeV-energy gap to probe GRB spectral cutoffs.

• See plots for application on GRB110328

LLEP6_V3_Transient

100MeV

Effective Area for GRB110328

3a. Constraining Lorentz­Invariance Violation

•There is a fundamental scale (the Planck scale λPl≈10-35 m) at which quantum gravity

(QG) effects are expected to strongly affect the nature of space-time.

• Lorentz symmetry implies a scale-free space-time (all scales are equivalent) → QG effects may cause violations of Lorentz Invariance (LIV) → speed of light in vacuum may acquire a dependence on its energy → υγ(Eγ

)≠c.

•The Lorentz-Invariance violating terms are typically expanded using a series of powers of the photon energy E

γ over the Quantum Gravity mass M

QG:

where sn={-1,0,+1} is a model-dependent factor.

•The Quantum-Gravity Mass MQG

• Sets the energy (mass) scale at which QG effects become important. • Is expected to be of the order of the Planck Mass and most likely smaller than it

Lorentz-Invariance Violation

• Since , the sum is dominated by the lowest-order term (n) with sn

0≠ ,

usually n=1 or 2 (“linear” and “quadratic” LIV respectively):

,where sn=+1 or -1 for subluminal and superluminal speeds respectively.

• There are many models that allow such LIV violations, and some others that actually require them (e.g. stringy-foam model J. Ellis et al. 2008).

• If the speed of light depends on its energy, then two photons with energies Eh>El emitted

together will arrive at different times. For sn=+1 (speed retardation):

• We want to constraint LIV Set lower limits on M→QG,n

➢We accomplish that by setting upper limits on the time delay t between photons of different Δenergies.

Lorentz-Invariance Violation

LAT All events

LAT >100MeV

LAT >1GeV

GBM NaI8-260keV

GBM BGO 0.26-5MeV

LAT – E vs T

We set upper limits on the delay Δt by associating the 31GeV photon with a lower-energy emission interval.

The starting time of that interval sets an upper limit on the time delay Δt

Most conservative case: 31GeV photon was not emitted before the start of the GRB:

Δt≤860ms ↔ MQG,1

≥1.19MPl

Photon was emitted some time after the start of the main <MeV emission:

Δt≤300ms ↔ MQG,1

≥3.42MPl

Photon was emitted some time after the start

of the >MeV emission:Δt≤178ms ↔ M

QG,1≥5.72M

Pl

Photon was emitted some time after the start of the >1GeV emission:

Δt≤99ms ↔ MQG,1

≥10.0MPl

GRB 090510

Method #2 – Dispersion Cancellation

Any energy-dependent time delays in our data would deform the high-energy peaks in the LAT light curve.

We can search for the spectral-lag value that cancels any such dispersions and maximizes the sharpness of the lightcurve.

A non-zero spectral-lag value would be a result of LIV and/or intrinsic to the GRB.

A simulated GRB light curve with a 20ms/GeV spectral lag.

The same light curve after applying an opposite lag (peaks now maximally sharp)

*Scargle J. D. et al. astro-ph/0610571v2

Searched for spectral lags using all the LAT detected events (35MeV-31GeV). The curve shows a measure of the sharpness of the light curve (Shannon information)

versus the trial spectral lag.

The solid vertical line denotes the minimum of the curve, which is our effective spectral-lag measurement.

The containment interval denoted by the vertical dashed lines is an approximate error region, but does not reflect statistical uncertainties.

Finding the spectral lag

Our effective spectral-lag measurement:

➔ The lightcurve was already maximally sharp.

✔ Similar results were obtained after small changes to the upper energy limit and the time interval of the used dataset.

Estimating the Statistical Error

We applied the same method on randomized datasets (shuffled the times between events) to measure the uncertainty of the measured spectral-lag value.

– 99% of the times the randomized data sets corresponded to a spectral lag smaller than ±30ms/GeV (90% of the times in ±10ms/GeV).

Combined result: symmetric upper limit on the spectral lag coefficient:

|Δt/ΔΕ|<30ms/GeV ↔ MQG,1

>1.22MPl

(99% C.L.) on possible linear (n=1) dispersion of either sign (sn=±1).

Limit almost the same as the most conservative limit of the previous method.

Distribution of the best trial-spectral lag values in 100 randomized datasets.

Upper Limits Table

● We constrained small changes in the speed of light caused by linear and quadratic perturbations in (Eγ/MQG).

● Using two independent techniques, we have placed strong limits on linear perturbations for both super- and sub-luminal speeds that were all higher than the Planck Mass.

● Our results support Lorentz invariance and disfavor models in which a quantum nature of space-time alters the speed of light, giving it a linear dependence on photon energy.

● More in our paper Abdo et al. Nature 2009, 462, 331A

3b. Constraining the Opacity of the Universe to High­

Energy Gamma Rays

The Extra-Galactic Background Light

● Accumulation of all energy releases in the form of electromagnetic radiation.

● Includes everything but CMB and the local foreground emissions (Milky Way, Solar System, etc.).

● Opacity effect: E>GeV Gamma-rays from extragalactic sources interact with it through γγ → e-e+

● Why is it important?● Contains information about the evolution

of matter in the universe: SFR, dust extinction, light absorption and re-emission by dust, etc.

● Its knowledge is necessary to infer the actual spectra of extragalactic gamma-ray sources.

● Observations of spectra that show no signs of absorption and that extend to >10 GeV energies from extra-galactic sources can set upper limits on the opacity of the universe or equivalently on the density of the EBL.

GRB Observations and the EBL

1. Assume intrinsic spectrum extends “as is” (with no extra curvature, breaks, etc.) from unabsorbed-by-the-EBL energies (say under ~10 GeV) to higher energies.

2. Calculate probability of this assumed intrinsic spectral model giving a detected photon of energy E≥E

max (for our actual observation of the source).

● Stecker et al. ('06) Baseline and Fast Evolution models predict too much opacity → probability for E≥E

max applied on our GRB090902B and 080916C

observations too low. ● These results are part of a more comprehensive

paper (Abdo et al. 2010ApJ...723.1082A) that uses multiple methods on multiple source types (blazars and GRBs).

● Overall results significantly (>11σ) reject these two EBL models.

Application to the Stecker et al. Baseline model. The Fast Evolution model predicts an even higher opacity

Area E>Emax

=2x10-4

Conclusion

● The Fermi LAT and GBM allowed us to – detect the keV-MeV-GeV emission from a large sample of bursts and

systematically characterize it,– explore the relation between the high and low energy emissions,– constrain current theoretical models on GRBs and guide future research, and– use GRBs as probes to explore other non-GRB sciences such as particle

physics and cosmology.

• The LAT observations during these first three years have spurred the development of numerous theories and models for GRB high-energy emissions.

• Now in our next three years, we have to find which ones are correct!

Thank you

Backup

Beaming corrections to emitted energy

● There are many reasons to believe that GRB emission is beamed (relativistic beaming, GRB emission mechanism)

● Beaming angle can be measured by breaks in the afterglow lightcurves

● After correcting for the case of a beamed geometry, isotropic energy released ~5*1050erg

● GRB emission now comparable with the emission from supernovae

D. A. Frail. Astro-ph/0311301

e+ e–

γ

The Large Area Telescope

• Precision Si-strip Tracker– 18 XY tracking planes– Single-sided silicon strip detectors (228 µm

pitch), 880,000 channels– Tungsten foil converters (1.5 X0)– Measures the photon direction; gamma ID

• Hodoscopic CsI Calorimeter– Array of 1536 CsI(Tl) crystals in 8 layers– 3072 spectroscopy chans (8.5 X0)– Hodoscopic array supports bkg rejection and

shower leakage correction– Measures the photon energy; images the

shower

• Segmented Anticoincidence Detector– 89 plastic scintillator tiles– Rejects background of charged cosmic rays;

segmentation minimizes self-veto effects at high energy

• Electronics System– Includes flexible, robust hardware trigger and

software filters

Sub-systems work together to identify and measure the flux of cosmic gamma Sub-systems work together to identify and measure the flux of cosmic gamma rays with energy between 20 MeV and 300 GeVrays with energy between 20 MeV and 300 GeV

Calorimeter

Tracker

ACD [surrounds 4x4 array of TKR towers]

High Energy Emission from GRBs

● SMM: detected GRBs in the 0.3-9MeV range

– 60% had significant emission above 1MeV

● EGRET: 0.03-30GeV range

– Detected photons above 100MeV from 4 GRBs

– GRB940217: 2 photons at ~3GeV, 1 photon at 18GeV 90 mins after the prompt emission

● Combined BATSE and EGRET data from GRB941017

● A distinct high energy component extending to at least 200MeV with no sign of a cutoff.

Gonzalez, et al., Nature 424, 847 (2003).

Duration Estimation

● GRB T90s are calculated based on the time development of the cumulative background-subtracted lightcurve.

● In low statistics lightcurves (as in the LAT) → individual fluctuations can introduce uncertainties in the choice of the plateau and can also “drive” the final T05/T95.

● To characterize these fluctuations we perform duration estimations on simulated lightcurves that are statistically compatible with the actual detected lightcurve.

● The final result comes from the median and +-1σ quantiles of the simulated T05/T95/T90 distributions.

● Method under development and verification.

● Improvements include removing the effects of variable exposure observations.

GRB090328

Preliminary

Method #1

Associations with individual spikes constrain both positive and negative time delays (sn=±1)

Such associations are not as secure → used as intuition builders (what we could do)

31GeV Photon lies at the center of a 20ms-wide pulse. We constrain both a positive and a negative time delay:

|Δt|<10ms↔ MQG,1>102MPl

750MeV photon & precursor. We place one more limit on a negative time delay:

|Δt|<19ms↔ MQG,1>1.33MPl

Absorption by the EBL