Particle Physics with LOFAR Stijn Buitink
 Vrije Universiteit Brussel LOFAR Data School 21-11-2014

many thanks to Pim Schellart, Anna Nelles, Arthur Corstanje for plots, pictures, etc.!

Content •

Particle physics with a radio telescope?

•

Working with data from the Buffer Boards

•

Analysis techniques for timeseries data

•

Introduction to LOFAR cosmic-ray analysis
 …and some other cool stuff!

LHC collision energy: 14.000 GeV

Particle acceleration in astrophysical shocks gamma rays cosmic rays

supernovae, AGNs, GRBs, ...

Earth’s atmosphere

source

neutrinos

Cosmic energy scale Detected: Cosmic rays up to 1020 eV
 Neutrinos at 1014 -1015 eV Gamma rays up to 1014 eV

solar flares

1010 eV

LHC

1012 eV

AGN

Supernovae

1015 eV

GRB

or...?

1020 eV

Cosmic-ray air showers cosmic rays

Air Showers: Cascades of secondary particles
 hadrons, muons, electrons, positrons, photons, etc.

deflection of electrons/positrons in magnetic field
 transverse current produces radiation

ultra short pulse (10-100 ns) can be 
 detected with radio antennas

Air showers detection with LOFAR van Haarlem et al. : LOFAR: The Low-Frequency Array

Goal:
 understand origin of CRs around 1017 eV
 disentangle Galactic & 
 extragalactic component

Fig. 1. Aerial photograph of the Superterp, the heart of the LOFAR core, from August 2011. The large circular island encompasses the six core stations that make up the Superterp. Three additional LOFAR core stations are visible in the upper right and lower left of the image. Each of these core stations includes a field of 96 low-band antennas and two sub-stations of 24 high-band antenna tiles each.

Challenge:
 ultra-short signals (10-100 ns)
 random arrival time & direction
 complicated radiation mechanism low-frequency radio domain below a few hundred MHz, representing the lowest frequency extreme of the accessible spectrum.

Since the discovery of radio emission from the Milky Way (Jansky 1933), now 80 years ago, radio astronomy has made a continuous stream of fundamental contributions to astronomy. Following the first large-sky surveys in Cambridge, yielding the 3C and 4C catalogs (Edge et al. 1959; Bennett 1962; Pilkington & Scott 1965; Gower et al. 1967) containing hundreds to thousands of radio sources, radio astronomy has blossomed. Crucial events in those early years were the identifications of the newly discovered radio sources in the optical waveband. Radio astrometric techniques, made possible through both interferometric and lunar occultation techniques, led to the systematic classification of many types of radio sources: Galactic supernova remnants (such as the Crab Nebula and Cassiopeia A), normal galaxies (M31), powerful radio galaxies (Cygnus A), and quasars (3C48 and 3C273).

During this same time period, our understanding of the physical processes responsible for the radio emission also progressed

Although the first two decades of radio astronomy were dominated by observations below a few hundred MHz, the prediction and subsequent detection of the 21cm line of hydrogen at 1420 MHz (van de Hulst 1945; Ewen & Purcell 1951), as well as the quest for higher angular resolution, shifted attention to higher frequencies. This shift toward higher frequencies was also driven in part by developments in receiver technology, interferometry, aperture synthesis, continental and intercontinental very long baseline interferometry (VLBI). Between 1970 and 2000, discoveries in radio astronomy were indeed dominated by the higher frequencies using aperture synthesis arrays in Cambridge, Westerbork, the VLA, MERLIN, ATCA and the GMRT in India as well as large monolithic dishes at Parkes, Effelsberg, Arecibo, Green Bank, Jodrell Bank, and Nanc¸ay. By the mid 1980s to early 1990s, however, several factors combined to cause a renewed interest in low-frequency radio astronomy. Scientifically, the realization that many sources have inverted radio spectra due to synchrotron self-absorption or freefree absorption as well as the detection of (ultra-) steep spectra

Transient Buffer Boards (TBBs) M. P. van Haarlem et al.: LOFAR: The LOw-Frequency ARray

Low Band Antenna Analogue signal Receiver : A/D conversion Digital Filter

Transient Buffer

Beamformer

Station Cabinet

To correlator in Groningen

High Band Antenna

Fig. 8. Schematic illustrating the signal connections at station level as well as the digital processing chain. After the beam-forming step, the signals are transferred to the correlator at the CEP facility in Groningen.

raw LOFAR data can be accessed via TBBs provide access to a snapshot of the running data-streams storage. The station LCUs run a version of Linux and are admin-bu↵ers 5 of data stored ring buffer fortheeach active antenna from theseconds HBA or LBA antennas. As depicted in Fig. 8,on a dediistered remotely over network from the LOFAR operations transient bu↵er board (TBB) is used that operates in par- center in Dwingeloo. Processes running on the LCU can include -catedraw data orEach sub-band data allel with thetimeseries normal streaming data processing. TBB can control drivers for the TBBs, RCUs, and other hardware com1 Gbyte of data for up to 8 dual-polarized antennas either ponents as well as additional computational tasks. All processes -store12 bit before or after conversion to sub-bands. This amount is suffi- running on the LCUs are initialized, monitored, and terminated -

cient to store 1.3 s of raw data allowing samples to be recorded at LOFAR’s full time resolution of 5 ns (assuming the 200-MHz sampling clock). Following successful tests for various science

by the MAC/SAS control system discussed below in Sect. 9. Computationally the LCU provides several crucial computing tasks at the station level. Chief among these are the beam-

Reading out TBBs Trigger strategies •

Manual trigger

•

External trigger

•

Local station trigger

•

Central trigger

LOFAR Radboud Array (LORA) external cosmic-ray trigger

LORA (Scintillator) High-Band Low-Band

x 20

Calibration I Time scale too short to calibrate on single source

•

Noise dominated by galactic background

•

•

Noise curve measured for specific reference antenna All antennas are calibrated relative to reference antenna

Galactic Power [a.u.]

•

X (NE-SW)

Local Sidereal Time Y (NW-SE)

Calibration II Absolute calibration strategies 1.Octocopter

2. Galactic background

3. Reference antenna on crane Absolute gain, antenna pattern, bandpass… in progress

2.3 The antenna model

The LBAs have a largemodel field of view and the antenna gain is a The antenna direction, polarisation, and frequency. Therefore, a radio puls

z antenna cannot be properly calibrated unless its direction is k direction can be found by using the differences in arrival tim ˆ en different antennas. Once the direction is known the antenna pa ˆ e to calculate the electromagnetic field of the incoming signal 1 The antenna pattern for the LBA X ✓is simulated with the WI ˆ e✓ age. This program calculates the electric fields ... inside the waves from different arrival directions, with different frequenc put voltage is then calculatedYusing an equivalent circuit that is an internal resistance equal to the antenna impedance. y The antenna responsex can be described by the Jones matri trix, that translates the field strength of the incoming wave to ✓ ◆ ✓ ◆✓ ◆ JXq JXf VX Eq = , JY q JYf VY Ef

Jones Matrix

where JXq is the complex response of the antenna and amplifi simulated with WIPL-D package wave purely polarized in the eˆ q direction. - complex response The to incoming wave WIPL-D software produces Jones matrices that are ca arrival direction angles and frequency frequencies. The matrices corres - depends on direction, polarisation, and frequencies between grid points can be found by interpola

-

RFI Cleaning & Gain calibr

Cleaning RFI

Average of multiple block FFTs
 phaseresolution
 stability of Radio Frequency e.g. 216 samples/block = 33 ms !=Use 3 kHz Interference transmitters multiple blocks to reduce noise line identification:
 - polynomial fit to baseline - phase stability

Pulse finding

Hilbert Envelope

A reliable technique is to calculate the Hilbert envelope of the tim which is defined as: q A(t) = x2 (t) + xˆ2 (t).

where x(t) ˆ is the Hilbert transform, or imaginary propagation, of th Hilbert transform defined by F (x(t))(w) ˆ = i · sgn(w) · F (x(t))(w)

where F denotes Fourier transform. 4 shows cosmic-ray Fig. 4 the Air shower radio pulse. TheFigure raw signal is plotteda in blue, and thep by an LBA. The Theroot blue curve isofthe sampled) raw signal, and line. the re mean square the(up raw signal is indicated by the dashed

A reliable technique is to calculate the Hilbert envelope of Envelope which is defined as: q A(t) = x2 (t) + xˆ2 (t).

where x(t) ˆ is the Hilbert transform, or imaginary propagatio defined by F (x(t))(w) ˆ = i · sgn(w) · F (x(t))(w)

where F denotes the Fourier transform. Figure 4 shows a cosm

Pulse arrival direction

Beamforming


Plane wave fit (using pulse arrival time)

very sensitive


less sensitive


many local minima (side-lobes)

more stable

1 1 n i=0 2 2 2 2 ˆ ˆ Q= Â (E + E E E ), n 1 i,x i,x i,y i,y n2i=0 U = Â (Ei,x Ei,y + Eˆi,x Ei,y ), nn i=0 2 1 ˆ U= Â n (E 1 i,x Ei,y + Ei,x Ei,y ), n2i=0 ˆ 9 th LOFAR V = Â (Ei,x Ei,y Ei,x Ei,y ). n n i=0 1 2 V = Â (Eˆi,x Ei,y Ei,x Ei,y ). 1n 1 2 2 2 2 n i=0 ˆ I = Â (Ei,x + Eˆi,x + E + E ), ˆi, j i,yEi, j isi,ysample i of electric field(7) where component j and E n i=0 For an elliptically polarized signal one can calculate from theit angle of semi-major axis: ˆ where E is sample i of electric field component j and E i, j i, j 1n 1 2 angle that the semi-major axis of the polarization ellipse mak 2For an2 elliptically 2 Q = Â (Ei,x + Eˆi,x Ei,y Eˆi,y ), polarized signal one can (8) calculate from the S ✓ ellipse ◆ n i=0 angle that the semi-major axis of the polarization makes 1 U 1 y = tan . n 1 ✓ ◆ 2 2 Q 1 U ˆ 1 U = Â (Ei,x Ei,y + Ei,x Ei,y ), (9) y = tan . n i=0 2 Q Additionally the degree of polarization is calculated which is tion of the power in the of polarized component of thewhich wave is de 2n 1 ˆ Additionally the degree polarization is calculated degree of polarization: V = Â (Ei,x Ei,y Ei,x Ei,y ). (10) p tion of the power in the polarized component of the wave n i=0 2 +U 2 +V 2 Q p =p 2 . 2 2 Q +UI +V mple Integrate i of electric field component j and Eˆi, j its Hilbert ptransform. Stokes parameters
 = . y polarized signal one can calculate from the Stokes parameters the I

Pulse polarization

over bins containing pulse

mi-major axis of the polarization ellipse makes with the xˆ axis 4✓ Introduction to Cosmic Ray Analysis ◆ Introduction to Cosmic Ray Analysis 1 14 U

Air shower detection with LOFAR LORA (Scintillator) High-Band Low-Band

LORA LOFAR Radboud Array scintillator detectors

trigger

TBBs low band antenna

offline analysis

2 ms read-out Pim Schellart et al., A&A 560, 98 (2013)

ature antennas grouped in rings

event display

superterp

wavefront ure as a s delays

asured

t to al pentagons: LORA scintillators

ay f Xmax ower

reconstructed core & direction

Corstanje et al. (in prep)

station outside superterp

quality of the fits to the LOFAR measurements.

Measurements

Nanosecond timing precision

or this analysis we have used air-shower mearadio wavefront = hyperbolic ements with LOFAR accumulated between June 1 and November 2013. In order to have a dense, h-quality sampling of the radio wavefront, and 0.1 degree resolution ubstantial distance range of more than ⇤ 150 m, require an air shower to be detected in at least r LOFAR core stations (each with two rings of 48 l-polarized antennas). Furthermore, the highquality data is obtained with the outer ring of -band antennas and therefore the sample is rected to this subset. This leaves a total of 165 asured air showers. Of these 165, three fail calation of time differences between stations (see t. 3.2) and one is unreliable due to thunderm conditions (see Sect. 4.6). This leaves a toof 161 high quality air shower measurements for analysis. All measured air showers are processed by the ndard cosmic-ray reconstruction software as debed in [12]. (a) Hyperbolic fit Arthur Corstanje et al., Astropart. Phys. 61 22 (2015) Pulse arrival times & uncertainties

(a) Near

Understanding the polarization

vxvxB PRIMARY geomagnetic vxB

vxvxB SECONDARY charge excess vxB

Figure 6. Polarization footprint of a single cosmic ray air-shower, as recorded with th band antennas, projected into the showerPim plane. Each arrow represents from Schellart et al., JCAP 10the 14 signal (2014) Interference: emission pattern = asymmetric The direction of the arrow is defined by the polarization angle ⇤ with the ˆ e⇤v B⇤ axis a

ID 86129434

10-90 MHz

zenith 31 deg 336 antennas χ2 / ndf = 1.02

SB et al., Phys Rev D 90 082003 (2014)

ID 98345942

HBA 110-240 MHz

zenith 43 deg 231 antennas χ2 / ndf = 1.9

• •

High band: Cherenkov rings Harder to analyse due to tile beamforming

Anna Nelles et al., submitted to Astropart. Phys.

CR mass composition
 proton penetrate deeper than iron nuclei

Fe

p

•

In its first years of operation LOFAR has “solved” the radiation mechanism

•

Now: astrophysics!

•

Comparison to simulation gives mass composition of CR flux

•

Changes in mass as a function of energy hint at different source component

•

Searching for the transition Galactic/extragalactic sources

What else can we see…?

RS503

RS106

CS002

RS503

CS002

RS106

RS205

RS208

SB, W.Frieswijk, S.ter Veen

Trigger: manual

future: local station trigger?

1020 - 10?? eV: Moon = 107 km2 detector area

WSRT best sensitivity now: Buitink et al. A&A 521, 47(2010)

CR/neutrino

LOFAR radio flash ns scale!

NuMoon: New Observation Mode

119

of false detection events.

120

2.1. Data flow

Station 1

ADC data

TBB

best limit @ WSRT offline analysis

PPF

ADC data PPF

Station 2

ADC data

TBB

TBB

Station Beamformer

ADC data PPF

PPF

Ionospheric de-dispersion & Tied Array Beaming

Storage Device 244 Subbands

Collecting (max. 50) Beams PPF Inversion

Triggering

TBB

Station Beamformer

Collecting station subbands

Storage Device

~ (50)

PPF Inversion

PPF Inversion

new challenge: real-time analysis! signal synthesis + trigger decision + communication

Triggering

Triggering

within 5 seconds

Anti – Coincidence

CEP

TBBs & you ? •

Transient Buffer Boards (TBBs) store 5 seconds of raw timeseries data

•

Many triggering strategies possible

•

Full-sky-all-the-time when running in background

•

Cosmic rays, neutrinos from the Moon, lightning, fast radio bursts, … Thanks for your attention!