1. HOW TO FIND MERGING CLUSTERS AND THE FRONTS INSIDE THEM
FABIO GASTALDELLO INAF-IASF MILAN
G. CAPURRI, A. BOTTEON, M. ROSSETTI, G. PANTIRI, M. DELLA TORRE, G. FERIOLI, G. BRUNETTI, R. KALE, D. DALLA CASA,D. ECKERT, S. MOLENDI, S. GHIZZARDI, S. DE GRANDI, S. ETTORI

2. OUTLINE
DYNAMICAL STATE AND COOL CORE STATE OF A PLANCK SELECTED SAMPLE AND LESSONS LEARNED ABOUT SZ vs X-RAY SELECTION (M. ROSSETTI, G. FERIOLI, G.PANTIRI, M. DELLA TORRE)
SHOCK FRONTS AND PARTICLE ACCELERATION IN RELICS IN A115 AND EL GORDO (A. BOTTEON)
APPLICATION OF EDGE DETECTION TECHNIQUES TO WELL KNWON MERGING CLUSTERS TO SEARCH FOR SURFACE BRIGHTNESS DISCONTINUITIES, i.e. COLD FRONTS, SHOCK FRONTS (G. CAPURRI)

3. PROPERTIES OF PLANCK CLUSTERS
“The majority of newly discovered Planck clusters show evidence for significant morphological disturbances”
(Planck Collaboration 2011, Planck Early results IX)
Are Planck-selected objects more dynamically disturbed than expected?
Compare a well-defined Planck subsample with X-ray selected samples
)M.ROSSETTI, FG, G. FERIOLI et al (2016) MNRAS 457, 4515

4. DYNAMICAL STATE OF PLANCK CLUSTERS
Abell 754
Abell 496
Offset between X-ray peak and BCG* position as a dynamical indicator (Hudson et al 2010, Sanderson et al 2009, Mann & Ebeling 12 )
*BCG= Brightest Cluster Galaxy
DISTURBED
Large offset
RELAXED
Small offset
)M.ROSSETTI, FG, G. FERIOLI et al (2016) MNRAS 457, 4515

5. DYNAMICAL STATE OF PLANCK CLUSTERS
Measured on almost complete sample (128/132) of Planck high S/N clusters with public X-ray (Chandra or XMM) observations and BCG identification (literature + archival analysis)
DYNAMICAL STATE OF PLANCK CLUSTERS
)M.ROSSETTI, FG, G. FERIOLI et al (2016) MNRAS 457, 4515

6. PLANCK SZ vs. X-RAY SAMPLES
Literature information on the BCG – Xray peak offset available for many samples, often with heterogeneous selection.
We compared only with purely X-ray selected samples
ME-MACS (Mann & Ebeling 2012):
108, most massive high-z (>0.15) objects in RASS data
HIFLUGCS (Zhang+, 2011):
62, Brightest X-ray clusters, local, low mass objects
REXCESS (Haarsma+2010):
30, intermediate mass and z
)M.ROSSETTI, FG, G. FERIOLI et al (2016) MNRAS 457, 4515

7. PLANCK SZ vs. X-RAY SAMPLES
Relaxed
Disturbed
)M.ROSSETTI, FG, G. FERIOLI et al (2016) MNRAS 457, 4515

8. PLANCK SZ vs. X-RAY SAMPLES
Relaxed
Disturbed
Kolmogorov-Smirnov
test
KS Statistic=0.228
Null hypothesis probability=
0.4%
Relaxed fraction
Planck: ±4%
ME-MACS: 73±4%
(HIFLUGCS: 74±5%, REXCESS 77±7%)
Fewer relaxed objects in Planck than in (all) X-ray samples

9. COOL CORE BIAS
Relaxed clusters usually feature a centrally peaked density profile, causing a prominent surface brightness peak
Flux limit or
SB limit?
It affects X-rays surveys (Ix ≈ne2 ,Pesce et al 1990,Eckert et al 2011) and is predicted to be small in SZ-surveys (ISZ ≈ne, Lin et al 2015, Pipino & Pierpaoli 2010), especially with Planck

10. THE CC STATE OF PLANCK CLUSTERS
DX-BCG is not a direct indicator of the presence of a prominent density peak
Redo the analysis with the concentration parameter (Santos et al 2008)
C=0.02
C=0.30
Abell 2204: CC
Abell 2069: NCC
NEW: M. ROSSETTI, FG, M. DELLA TORRE, G. PANTIRI, D. ECKERT et al (2017) ,
astro-ph:

11. THE CC STATE OF PLANCK CLUSTERS
Santos+ 08
Measured on a almost complete sample (169/189) of Planck high S/N clusters based on PSZ1 cosmo catalogue.
Comparison with 104 ME-MACS clusters (overlap in M and z)
ROSSETTI et al (2017)

12. THE CC STATE OF PLANCK CLUSTERS
Santos+ 08
CC fraction
Planck: ±4%
ME-MACS: 59 ±5%
KS test: KS statistic D=0.35, Null hyp. Prob. p0=
Even more significant difference between Planck and ME-MACS
ROSSETTI et al (2017)

13. Secondary CC peak in simulated distribution
THE ROLE OF THE CC BIAS
Simulations updated from Eckert et al (2011) to reproduce CC-bias in a ME-MACS-like survey starting from a Planck-like sample
Input CC frac: 29%
Output CC frac: 48%
Secondary CC peak in simulated distribution

14. BEYOND THE CC BIAS Other effects can contribute to the difference:
anti-CC bias in Planck due to masked radio-galaxies (1-2%)
bias towards merging clusters in SZ due to shocked regions
bias towards multiple systems in Planck
Possible difference in pressure profiles between CC and NCC at large scales (>R500)
X-ray underluminous clusters

15. X-RAY UNDERLUMINOUS CLUSTERS
Clusters in Planck but NOT in ME-MACS
Presence of X-ray underluminous objects in Planck catalogue (Planck Coll. 2016)
In our sample they are all NCC!

16. RELIC-SHOCK CONNECTION
Relics are diffuse and polarized radio sources on Mpc scales found in the outskirts of GC in merging state

17. ABELL 115
A. BOTTEON, FG, et al (2016) MNRAS 460, L84

18. THE SHOCK IN ABELL 115
A. BOTTEON, FG, et al (2016) MNRAS 460, L84

19. THE SHOCK POSITION
Consistent with an off-axis merger with a mass ratio 1:3, scenario consistent with the optical analysis of Barrena+07

20. ELECTRON RE-ACCELERATION IN A115
DSA requires too large acceleration efficiency for acceleration of thermal electrons, so re-acceleration is invoked (e.g. Bonafede+14, Shimwell+15, van Weeren+17).
A115 also requires very high acceleration efficiency so a population of seed electrons (provided by radio galaxies) alleviates the energetic requirements

21. EL GORDO
A. BOTTEON, FG, et al (2016) MNRAS 463, 1534

22. THE SHOCK IN EL GORDO
A. BOTTEON, FG, et al (2016) MNRAS 463, 1534

23. THE SHOCK IN EL GORDO
One of the few M ≥ 3 shocks known in clusters (Bullet; Markevitch+02, A665; Dasadia+16). For such relatively strong shocks DSA for thermal electrons can not be ruled out

24. EDGE DETECTION
EDGE = rapid change in the brightness of the image
Maximum of the modulus of the gradient
⟶ Gradient Based Methods
Zeros of the Laplacian
⟶ Zero Crossing Methods
One clear problem is Poissonian noise: you need to smooth the image and reach a compromise between reduction of the noise and conservation of the features
G. CAPURRI, FG, et al in prep.

25. GRADIENT BASED METHODS
Image smoothing ⟶ kernel (e.g. Gaussian)
Calculation of the gradient ⟶ finite difference technique
The two processes in one go ⟶ convolution with a single kernel
SOBEL FILTER
G. CAPURRI, FG, et al in prep.

26. CANNY EDGE DETECTOR Developed by Canny (1986) as an optimal filter
Sobel filter + direction of the gradient
Edge definition by choice of two thresholds through an hysteresis process (THINNING)
G. CAPURRI, FG, et al in prep.

27. UNSHARP MASKING Gaussian smoothing Calculation of Laplacian
Numerical approximation → Difference of two Gaussians
G. CAPURRI, FG, et al in prep.

28. BULLET CLUSTER SOBEL FILTER UNSHARP MASKING smoothing σ = 5 pixel
σ₁= 2 pixel σ₂= 15 pixel
G. CAPURRI, FG, et al in prep.

29. BULLET CLUSTER CANNY EDGE DETECTOR CANNY EDGE DETECTOR
Smoothing with σ = 3 pixel
High = Low = 0.1
CANNY EDGE DETECTOR
Smoothing with σ = 3 pixel
High = Low = 0.1
G. CAPURRI, FG, et al in prep.

30. Output of Canny and radio contours by Shimwell+15
BULLET CLUSTER
Output of Canny and radio contours by Shimwell+15
G. CAPURRI, FG, et al in prep.

31. SUMMARY
SZ SAMPLES ARE DIFFERENT IN CC FRACTION THAN X-RAY ONES CLEARLY POINTING TO DIFFERENT SELECTION PROCESSES (ROLE OF CC BIAS IN X-RAY SAMPLES
RELIC-SHOCK CONNECTION STRENGTHNED BY OTHER TWO DISCOVERED CASES, ONE WITH A RELIC IN A VERY UNUSUAL POSITION (A115) AND ONE OF THE FEW STRONG SHOCK CASES IN CLUSTERS (EL GORDO)
EDGE DETECTION TECHNIQUES APPLIED TO MERGING CLUSTERS CAN BE APPLIED TO REVEAL INTERESTING FEATURES

32. THANKS !