Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userAn ultra-deep ISOCAM observation through a cluster-lens
152
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
We present results of ultra-deep ISOCAM observations through a cluster-lens at 7 and 15 micron with the Infrared Space Observatory (ISO) satellite. These observations reveal a large number of luminous Mid-Infrared (MIR) sources. Cross-identification in the optical and Near-Infrared (NIR) wavebands shows that about half of the 7 micron sources are cluster galaxies. The other 7 micron and almost all 15 micron sources are identified as lensed distant galaxies. Thanks to the gravitational amplification they constitute the faintest MIR detected sources, allowing us to extend the number counts in both the 7 and 15 micron bands. In particular, we find that the 15 micron counts have a steep slope alpha_15 = -1.5 +/- 0.3 and are large, with N_15 (>30 microJy}) = 13 +/- 5 per square arcmin. These numbers rule out non-evolutionary models and favour very strong evolution. Down to our counts limit, we found that the resolved 7 and 15 microns background radiation intensity is respectively (2 +/-0.5) 10^(-9) and (5 +/-1) 10^(-9) W m^(-2) sr^(-1).
rate research
Read More
We present imaging results and source counts from an ISOCAM deep and ultra-deep cosmological survey through gravitationally lensing clusters of galaxies at 7 and 15 microns. A total area of about 53 sq.arcmin was covered in maps of three clusters. The lensing increases the sensitivity of the survey. A large number of luminous mid-infrared (MIR) sources were detected behind the lenses, and most could be unambiguously identified with visible counterparts. Thanks to the gravitational amplification, these results include the faintest MIR detections ever recorded, extending source counts to an unprecedented level. The source counts, corrected for cluster contamination and lensing distortion effects, show an excess by a factor of 10 with respect to the prediction of a no-evolution model, as we reported for A2390 alone in Altieri et al. (1999). These results support the A2390 result that the resolved 7 and 15 microns background radiation intensities are 1.7 (+/- 0.5) x 10^-9 and 3.3 (+/- 1.3) x 10^-9 W/m^2/sr, respectively, integrating from 30 microJy to 50 mJy.
ISOCAM was used to perform a deep survey through three gravitationally lensing clusters of galaxies. Nearly seventy sq. arcmin were covered over the clusters A370, A2218 and A2390. We present maps and photometry at 6.7 & 14.3 microns, showing a total of 145 mid-IR sources and the associated source counts. The 15 micron counts reach the faintest level yet recorded. All sources have counterparts in the optical or near-IR. Models of the clusters were used to correct for the effects of lensing, which increases the sensitivity of the survey. Seven of fifteen SCUBA sources were detected at 15 microns. Five have redshift between 0.23 & 2.8, with a median of 0.9. The field sources were counted to a lensing-corrected sensitivity of 30 microJy at 15 microns, and 14 microJy at 7 microns. The counts, corrected for completeness, contamination by cluster sources and lensing, confirm and extend findings of an excess by a factor of ten in the 15 micron population with respect to source models with no evolution. Source redshifts are mostly between 0.4 and 1.5. For the counts at 7 microns, integrating from 14 microJy to 460 microJy, we resolve 0.49+/-0.2 nW.m^(-2).sr^(-1) of the infrared background light (IBL) into discrete sources. At 15 microns we include the counts from other ISOCAM surveys to integrate from 30 microJy to 50 mJy, two to three times deeper than unlensed surveys, to resolve 2.7+/-0.62 nW.m^(-2).sr^(-1) of the IBL. These values are 10% and 55%, respectively, of the upper limit to the IBL, derived from photon-photon pair production of the TeV gamma rays from BL-Lac sources on the IBL photons. However, recent detections of TeV gamma rays from the z=0.129 BL Lac H1426+428 suggest that the 15 micron background reported implies substantial absorption of TeV photons from that source.
Gravitational lensing by massive galaxy clusters allows study of the population of intrinsically faint infrared galaxies that lie below the sensitivity and confusion limits of current infrared and submillimeter telescopes. We present ultra-deep PACS 100 and 160 microns observations toward the cluster lens Abell 2218, to penetrate the Herschel confusion limit. We derive source counts down to a flux density of 1 mJy at 100 microns and 2 mJy at 160 microns, aided by strong gravitational lensing. At these levels, source densities are 20 and 10 beams/source in the two bands, approaching source density confusion at 160 microns. The slope of the counts below the turnover of the Euclidean-normalized differential curve is constrained in both bands and is consistent with most of the recent backwards evolutionary models. By integrating number counts over the flux range accessed by Abell 2218 lensing (0.94-35 mJy at 100 microns and 1.47-35 mJy at 160 microns, we retrieve a cosmic infrared background (CIB) surface brightness of ~8.0 and ~9.9 nW m^-2 sr^-1, in the respective bands. These values correspond to 55% (+/- 24%) and 77% (+/- 31%) of DIRBE direct measurements. Combining Abell 2218 results with wider/shallower fields, these figures increase to 62% (+/- 25%) and 88% (+/- 32%) CIB total fractions, resolved at 100 and 160 microns, disregarding the high uncertainties of DIRBE absolute values.
We present imaging results and source counts from a deep ISOCAM cosmological survey at 15 microns, through gravitationally lensing galaxy clusters. We take advantage of the cluster gravitational amplification to increase the sensitivity of our survey. We detect a large number of luminous mid-IR sources behind the cluster lenses, down to very faint fluxes, which would have been unreachable without the gravitational lensing effect. These source counts, corrected for lensing distortion effects and incompleteness, are in excess of the predictions of no-evolution models that fit local IRAS counts. By integrating the 15 microns source counts from our counts limit, 30 microJy, to 50 mJy we estimate the resolved mid-IR background radiation intensity.
Existing work on understanding deep learning often employs measures that compress all data-dependent information into a few numbers. In this work, we adopt a perspective based on the role of individual examples. We introduce a measure of the computational difficulty of making a prediction for a given input: the (effective) prediction depth. Our extensive investigation reveals surprising yet simple relationships between the prediction depth of a given input and the models uncertainty, confidence, accuracy and speed of learning for that data point. We further categorize difficult examples into three interpretable groups, demonstrate how these groups are processed differently inside deep models and showcase how this understanding allows us to improve prediction accuracy. Insights from our study lead to a coherent view of a number of separately reported phenomena in the literature: early layers generalize while later layers memorize; early layers converge faster and networks learn easy data and simple functions first.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
