text stringlengths 128 2.05k |
|---|
When the network is operating at the critical regime, the dynamics can be seen as balancing between a predictable pattern of activity and uncorrelated random behavior typically present in SOC. One would also expect to find macroscopic magnitudes distributed according to scale-free distributions. Preliminary results ind... |
The spontaneous evolution can be interpreted for instance as activity from other brain areas not considered in the pool of the simulated units, or as stochastic sensory input. Our results indicate that the amount of this stochastic activity that is absorbed by the system is maximized at an optimal state, which in a sen... |
The application of the synaptic rule for information processing is left for future research. We advance, however, that external perturbations when the network is critical would cause a transient activity. During the transient, synapses could be modified according to some other form of learning to encode the proper valu... |
# Source: arxiv 0808.3139 # Title: The star-formation history of K-selected galaxies # Sections: all # Downloaded: 2026-03-02T07:58:31.084775+00:00 |
The star-formation history of [MATH] -selected galaxies (Accepted 2008 August 22. Received 2008 August 21; in original form 2008 May 25) |
Abstract We have studied the [MATH] Jy radio properties of [MATH] -selected galaxy populations detected in the Ultra-Deep Survey (UDS) portion of the United Kingdom Infrared Telescope (UKIRT) Deep Sky Survey (UKIDSS) using 610- and 1,400-MHz images from the Very Large Array (VLA) and the Giant Metre-wave Telescope (GMR... |
[MATH] -selected sub-populations, including sBzK and pBzK star-forming and passive galaxies. We find a strong correlation between the radio and |
[MATH] -band fluxes and a linear relationship between star-formation rate (SFR) and [MATH] -band luminosity. We find no evidence, from either radio spectral indices or a comparison with submm-derived SFRs, that the full sample is strongly contaminated by active galactic nuclei (AGN) at these low flux densities, though ... |
[MATH] , implying that location on the BzK diagram alone is not sufficient to select samples at [MATH] . The sBzK and pBzK galaxies have similar levels of radio flux density, SFR and specific SFR (SSFR – SFR per unit stellar mass) at [MATH] , suggesting there is strong contamination of the pBzK sample by star-forming g... |
keywords: galaxies: active – galaxies: evolution – galaxies: starburst Introduction There has recently been a large advance in our ability to undertake deep, large-area surveys in the near-infrared (IR) due to the introduction of very large IR-sensitive arrays. Such surveys allow us to collect statistically powerful sa... |
[MATH] filter at 2.2 [MATH] m, providing access to the part of the spectral energy distribution (SED) dominated by rest-frame optical light for galaxies at high redshift and by low-mass stars for galaxies at low redshift. Since it is not as strongly affected by dust or biased towards young stars as optical emission, it... |
There are several classifications for [MATH] -selected galaxies – extremely red objects (EROs – Elston, Rieke & Rieke 1989), distant red galaxies (DRGs – van Dokkum et al. 2004) and, more recently, BzK galaxies (Daddi et al. 2004). All these selections are colour based and aim to select – via SED shape – particular kin... |
The strong correlation between far-IR (FIR) and radio emission (Helou et al. 1985) in local star-forming galaxies (SFGs) has allowed the use of radio surveys as a means to quantify star-formation activity in a manner immune to the obscuring effects of dust. The FIR emission is believed to be produced by dust which has ... |
The radio source counts above 1 mJy are dominated by AGN, the powerful Fanaroff-Riley- ii (FR ii – Fanaroff & Riley 1974) sources giving way to the less powerful (but still radio-loud) FR AGN as the flux-density threshold approaches [MATH] mJy (Windhorst et al. 1985; Willott et al. 2002). The evolution of these two typ... |
The exact mix of AGN versus star formation at sub-mJy levels is still the subject of lively debate (see Ibar et al. 2008 and references therein). The first deep radio surveys ascribed the bulk of the sub-mJy activity to SFGs (Hopkins et al. 1998; Gruppioni et al. 1999; Richards 2000) though further investigations with ... |
As we are probing the radio population at flux density levels an order of magnitude fainter than existing studies, it is impossible to know a priori – the relative contributions to the radio flux of accretion and star formation for a given population. This difficulty in distinguishing between star-formation- and AGN-in... |
[MATH] or ratios of radio/optical flux, then we will bias ourselves as extreme starbursts will also have high [MATH] (Ivison et al. 2002) and high radio/optical ratios. Given the findings of Richards et al. (2007), that radio-emitting AGN appear at all radio luminosities and radio/optical ratios, we do not feel that su... |
The layout of the paper is as follows: §2 describes the radio and IR observations. §3 describes the stacking procedure used in the radio. In §4 we present the results of the stacks while in §5 we investigate the possible contamination from AGN using radio spectral indices and submm stacks. In §6 we present the radio-de... |
[MATH] -selected galaxies out to [MATH] and also investigate their specific SFRs as a function of mass and redshift. The cosmology used throughout is [MATH] |
[MATH] and [MATH] km s -1 Mpc -1 Observations 2.1 IR data and photometric redshifts The catalogue used in this paper was selected from the ongoing UDS, which is the deepest of the five surveys that comprise UKIDSS (Lawrence et al. 2007), exploiting the IR Wide-Field Camera (WFCAM – Casali et al. 2007) on UKIRT . The UD... |
[MATH] 100 percent. We matched the UDS [MATH] -selected catalogue to deep optical data in this field from the Subaru XMM-Newton Deep Field survey (Furasawa et al. 2008) in five broad-band filters to 3- [MATH] depths of |
[MATH] (again, 2-arcsec diameter apertures). This deep multi-wavelength coverage of the UDS allowed us to model the galaxy SEDs in detail, as described in Cirasuolo et al. (2008). The SED fitting procedure, to derive photometric redshifts, was performed with a code based largely on the public package Hyperz |
(Bolzonella, Miralles & Pelló 2000) and exploits reliable photometry in 16 broad bands from far-UV to the mid-IR. Both empirical (Coleman, Wu & Weeman 1980; Kinney et al. 1996; Mignoli et al. 2005) and synthetic templates (Bruzual & Charlot 2003) were used to model the galaxy SEDs, including a prescription for dust att... |
For the [MATH] 1,200 galaxies in the UDS with reliable spectroscopic redshifts, the agreement of the photometric redshifts is good over the full redshift range, [MATH] . The distribution of the [MATH] |
has a mean consistent with zero (0.008), a standard deviation, [MATH] , and clear outliers make up less than 2 per cent of the total. A comparison of [MATH] versus [MATH] can be found in fig. 2 of Cirasuolo et al. (2008). This accuracy is comparable to the best available from other surveys such as GOODS and COSMOS (Cap... |
2.2 Radio observations The radio data used here are described by Ivison et al. (2007a) and Ibar et al. (2008). Briefly, we employed the National Radio Astronomy Observatory’s (NRAO ) VLA at 1,400 MHz, with an effective bandwidth of 4 [MATH] [MATH] 3.125 MHz (via correlator mode ‘4’) and central frequencies in the two I... |
[MATH] 15 [MATH] Jy beam -1 . A total of 563 sources were detected at [MATH] [MATH] When imaging, we used only those data within a radius of each pointing centre where the dimensionless BWS term, [MATH] (fractional bandwidth |
[MATH] radius, the latter in units of fwhm ), was [MATH] 1.225. This ensured the reduction in peak flux density due to BWS — a pernicious effect which can lead to the complete loss of real sources from radio catalogues — was always [MATH] 37 per cent [MATH] ). We then calculated error-weighted values of |
[MATH] at every point in the final 3-position mosaic, using these to correct the flux densities extracted at the position of each galaxy. |
Stacking the radio flux of [MATH] -selected populations 3.1 IR-selection criteria The selection criteria for the galaxies to be stacked were [MATH] and total [MATH] for the photometric redshift. This prevents the results becoming contaminated with unreliable sources or redshifts. Sources had to lie in the overlap regio... |
[MATH] data, the SuprimeCam optical images and the VLA 1,400-MHz image (1311.1 arcmin ). The total number of [MATH] -selected sources in this region, meeting the above criteria is 23,185 of which 440 have radio counterparts at [MATH] [MATH] |
A BzK diagram of the sample meeting the [MATH] and photometric redshift selection criteria is shown in Fig. . This splits into several sub-samples based on the BzK colour cut (Daddi et al. 2004), where BzK [MATH] |
1. All galaxies : all objects classified as galaxies (and not lying on the stellar branch of the BzK diagram) meeting the selection criteria. All points on Fig. , except the green stars. |
2. Star-forming, high-redshift : using the sBzK selection of [MATH] we select a sample of 6,626 high-redshift SFGs. These are the blue points on Fig. |
3. Passive, high-redshift : using the pBzK selections of [MATH] and [MATH] we select a sample of 542 high-redshift passive galaxies. These are the red points on Fig. |
4. Non-BzK galaxies : all of the black points in Fig. . Generally galaxies at lower redshifts than the BzK sources. 5. EROs : defined as all objects with [MATH] . Generally these span the space overlapping pBzK, sBzK and non-BzK at the top right of the BzK diagram (see Fig. ). |
Note that a sparse stellar branch is visible on the plot because the BzK diagram is efficient at selecting stars up to 1 mag deeper than the compactness selection used to make the [MATH] catalogue. |
These colour selections are untested at the depths of the UDS and we will investigate their reliability as best we can using the radio data and photometric redshifts. We also note that the sBzK star-forming sample is known to contain a fraction of AGN (Daddi et al. 2004; Reddy et al. 2005) and we will discuss this in m... |
3.2 Radio Stacking To perform the stacking we measured the noise-weighted mean and median of the pixels on the radio image corresponding to the positions of the |
[MATH] -band sources. As the radio image has units of Jy beam -1 this pixel stacking gives the total flux of a source which is no larger than the beam. We only considered regions on the radio image with [MATH] Jy. For the mean stacks we excluded pixels with [MATH] but for the median we included all sources. The errors ... |
We corrected the final mean or median values for the effect of bandwidth smearing (as described in § 2.2 ) by multiplying the value by the mean or median BWS term. The mean and median BWS term was calculated in the same way as the mean and median flux density. The reason the correction was applied to the final stacked ... |
[MATH] on the resulting pixel flux distribution. If we multiply each pixel by the BWS value, but not the noise on that pixel, then we are artificially increasing the S/N of every pixel. This has two consequences: first, we finish with statistics which are not Gaussian (e.g. more [MATH] 3- [MATH] peaks on the stacked ma... |
We performed exactly the same analysis at random positions to check that there were no systematic effects in the radio map which might lead us to a positive flux. Flux was stacked at 21,000 random positions and this stack was repeated 1,000 times. The mean of the 1,000 mean values was 0.015 [MATH] Jy with [MATH] Jy and... |
[MATH] Jy with [MATH] Jy. This is consistent with zero and well below any fluxes quoted from a stack of [MATH] -selected objects. |
We also took a [MATH] -pixel region around every IR position and stacked this to produce a noise-weighted mean image stack (see Fig. ). We excluded any pixels on the radio image with a high noise level ( [MATH] 20 [MATH] Jy) or with a value of |
[MATH] ). These postage-stamp images are not corrected for BWS as the integrated flux is not affected by band-width smearing. The integrated flux is measured directly from the postage stamp images using jmfit in [MATH] and also includes any systematic effects due to positional uncertainty in the [MATH] -band positions,... |
gave consistent ratios between the [MATH] integrated fit and the mean from the pixel stack. Thus positional errors on the [MATH] -selected sources did not vary significantly as a function of [MATH] mag to our limit of [MATH] . The largest effect seems to be from partially resolved sources at low redshifts. The average ... |
and 2.29 for [MATH] . A further correction needs to be made to this integrated flux for the fact that these radio stacks are not ‘ clean ed’. In the VLA image, only sources visible to the eye were clean ed, via clean boxes placed by hand. Sources at the low flux density levels found here resemble the dirty beam with fl... |
We also checked that the noise integrated down as expected following Poissonian statistics by measuring the noise in the stacked postage-stamp images around the detected sources and on the Monte-Carlo images. This confirmed that the noise does integrate down as expected, i.e. as [MATH] |
3.3 Deriving SFRs and luminosities We can use photometric redshifts at the time of stacking the radio data to derive rest-frame 1,400-MHz luminosities, [MATH] and SFRs for the galaxies. The 1,400-MHz luminosity is calculated for each pixel based on the flux at that pixel and the redshift of the galaxy being stacked, us... |
[EQUATION] where [MATH] is the luminosity distance of an individual galaxy in Mpc and [MATH] is the 1,400-MHz flux density of the pixel corresponding to that galaxy in Jy. This assumes a radio spectral index ( [MATH] ) of [MATH] |
The median 1,400-MHz luminosity is then the median of the luminosities for the individual pixels. We note that this is not the same as taking the median stacked flux and median redshift and applying the above equation, since the median redshift of the sub-sample may not be that at which the peak of the radio emission i... |
The SFR is then calculated from the median 1,400-MHz luminosity in two ways. First, following Condon (1992), Haarsma et al. (2000) and Condon et al. (2002), we have |
[EQUATION] where [MATH] is the frequency in GHz. For 1,400 MHz, this becomes [EQUATION] Second, following Bell (2003), we have [EQUATION] |
where [MATH] Both conversions assume a Salpeter-like initial mass function (IMF) with [MATH] and a mass range of [MATH] . The study by Bell (2003) suggests that the radio does not trace star formation as effectively at low values of [MATH] because the non-thermal radio emission is suppressed by a factor of 2–3 in [MATH... |
[MATH] . The Bell (2003) conversion thus boosts the radio SFR at low [MATH] values by a factor [MATH] compared to Condon (1992); however, the overall normalisation of SFR is lower than that used by Condon (1992). |
Results Table shows the results for each sample and for each stacking method. All four sub-sets of [MATH] -selected galaxies (ERO, non-BzK, sBzK and pBzK) are significantly detected in the stacks, their postage-stamp stacked images are shown in Fig. |
The BzK selection (Daddi et al. 2004) is designed to select either star-forming (sBzK) or passively-evolving (pBzK) galaxies in the redshift range [MATH] . We have taken the colour cut used by Lane et al. (2007), as shown in Fig. , which gives us samples of 6,626 sBzK and 542 pBzK galaxies that match the selection crit... |
The redshift distributions for the various samples are shown in Fig. . It is immediately apparent that the distributions for BzK galaxies have tails at redshifts higher and lower than the range in which they are designed to lie. At [MATH] , this amounts to a fraction of 37 (29) per cent for pBzK (sBzK) galaxies. There ... |
objects mostly lie along the boundary with non-BzK objects, suggesting that photometric scatter is the main cause of the low-redshift tail. For pBzK galaxies, however, the [MATH] objects fill the same region of colour space as those at [MATH] . We will investigate the radio properties of BzK galaxies at all redshifts t... |
In Fig. (left) we show the median flux of BzK galaxies as a function of redshift. The average 1,400-MHz flux density of sBzK galaxies is roughly constant (or slowly rising) over the interval |
[MATH] , implying a strong evolution in luminosity. For pBzK galaxies there is a dramatic change in the average radio emission from |
[MATH] to [MATH] . At [MATH] , the pBzK galaxies are relatively bright radio emitters, and are in fact more luminous at this redshift than sBzK galaxies. At [MATH] , their radio emission falls to approximately half that of sBzK galaxies. This is the first piece of evidence that the pBzK sample (as defined by the BzK di... |
[MATH] against redshift for the sBzK and pBzK samples. The flat flux-density-redshift trend for sBzK galaxies in Fig. (left) translates into a steep rise of [MATH] with redshift. In contrast, the pBzK sample behaves quite differently, with median |
[MATH] falling slightly over the range [MATH] before rising dramatically at higher redshifts (in line with sBzK galaxies). The radio luminosity of pBzK galaxies is higher than that of sBzK galaxies at [MATH] ; however, the pBzK galaxies are also more luminous in |
[MATH] over this redshift interval. The elevated luminosity of the pBzK galaxies in this redshift bin is due to the correlation between radio and [MATH] fluxes and luminosities (see Fig. and |
10 ). Indeed, the ratio of radio to [MATH] fluxes in sBzK and pBzK galaxies at [MATH] is the same (see Fig. ). pBzK galaxies are also more luminous in [MATH] than sBzK galaxies in the |
[MATH] bin. This means that the difference between the sBzK and pBzK galaxies in this bin in Fig. (right) is intrinsically larger when normalised to [MATH] luminosity than is shown here. |
In Fig. we show the correlation between [MATH] mag and [MATH] for the full sample (top) and the BzK and ERO galaxies (middle). There is a strong trend, with [MATH] for the full sample, and a steeper relationship for the sBzK galaxies of [MATH] and for the pBzK galaxies of [MATH] . The difference in slope is possibly du... |
[MATH] are shown to have much less radio emission at a given [MATH] mag than either sBzK or [MATH] pBzK galaxies. There is not enough signal to determine a slope to the high-redshift pBzK data. The remaining radio flux from the high-redshift pBzK galaxies could come either from a much lower-level contamination by SFGs,... |
The EROs are a heterogeneous sample. At brighter [MATH] mags ( [MATH] they match the trends of the pBzK galaxies, but at fainter [MATH] mags the correlation between radio and [MATH] seems to break down (Fig. -middle). In the BzK diagram (Fig. ), the EROs typically straddle the sBzK and pBzK colour range. Bright EROs ha... |
Various colour-selected samples have previously been stacked at 1,400 MHz: BzK (Daddi et al. 2005) and EROs (Georgakakis et al. 2006; Simpson et al. 2007; Ivison et al. 2007b). When we account for the differing [MATH] limits of the various samples using Fig. we find good agreement between our ERO and BzK values and tho... |
Reliability of radio emission as a tracer of star formation Integral to this paper is the assumption that we can use the 1,400-MHz radio emission as a tracer of star formation. The main reason why this may not be a valid assumption is the presence of radio-emitting AGN in the sample at flux densities below the detectio... |
1998; Obrić et al. 2006; Mauch & Sadler 2007). In fact, there are thought to be two radio classes of Seyfert galaxy, one where the majority of the radio emission can be attributed to star formation and another containing compact radio cores and extended structures (‘mini-lobes’) which are radio loud and depart more fro... |
So, while the literature suggests we should not be unduly worried about AGN contamination in this stacked sample, we are clearly in uncharted waters as far as radio flux densities and redshift are concerned. We will now attempt to investigate the prevalence of AGN in our stacked sample in two direct ways. First, we wil... |
5.1 Radio spectral indices The spectral indices of the mJy and sub-mJy radio populations can give some clues to the nature of the radio emission. Above 1 mJy, steep-spectrum, radio-loud AGN dominate with [MATH] (Bondi et al. 2007). Below 1 mJy, where the radio counts flatten and the new population emerges, the spectral... |
[MATH] Jy to [MATH] from [MATH] Jy, it then steepens to [MATH] at fainter fluxes. They also find that early-type galaxies have flatter indices ( [MATH] ) than late-type/starburst galaxies [MATH] ). Fomalont et al. (2006) compared 1,400- and 8,400-MHz maps and found that sources with [MATH] Jy have steeper indices than ... |
was thought to be due to an increase in the contribution from flat-spectrum compact radio cores hosted by low-luminosity AGNs, with the subsequent steepening at [MATH] Jy ascribed to an increasing dominance of starburst/late-type galaxies. Unfortunately, this observed steepening at sub-mJy fluxes is not yet a statistic... |
We will now investigate whether stacking the [MATH] -selected populations into radio maps at two frequencies can shed any light on the behaviour of the radio spectral index at [MATH] Jy. |
5.1.1 GMRT observations and stacking For our second frequency, we employ a 610-MHz map taken with the Giant Metre-wave Radio Telescope (GMRT ) near Pune, India. Data were obtained in the UDS field during 2006 February 03–06 and December 05–10, primarily to explore the spectral index of the radio emission from submm-sel... |
Calibration initially followed standard recipes within [MATH] . However, because of concerns that some baselines were picking up signal from local power lines, a raft of new measures were introduced to avoid detrimental effects on the resultant images. These are to be reported in detail in Ibar et al. |
Imaging each of these datasets entailed mosaicing 37 facets, each 512 [MATH] 1.25 -arcsec pixels, to cover the primary beam. A further 6–12 bright sources outside these central regions, identified in heavily tapered maps, were also imaged. Our aim was to obtain the best possible model of the sky. clean boxes were place... |
The final six mosaics, two for each pointing (upper and lower sidebands), were convolved to a common beam size, then knitted together using flatn . An appropriate correction was made for the shape of the primary beam and data beyond the quarter-power point of the primary beam were rejected. Data for each pointing and s... |
The stacking was performed in the same way as for the 1,400-MHz map, except that the BWS correction is negligible for the GMRT’s narrow channels. Regions with [MATH] Jy were included in the stack. We calculated the correction for integrated flux and dirty-to- clean as we have for the 1,400-MHz data. The final correctio... |
Due to the noise being much higher in the GMRT data, only stacks of total populations could be made (not split as a function of redshift, |
[MATH] mag, etc.). We also stacked sets of random positions into the GMRT map to check for systematics and we found that the means are consistent with zero. The medians are generally a little below the means and there is a possible [MATH] Jy offset, consistent with the median background removal applied to the map which... |
5.1.2 Spectral index results The results of the stacking for the [MATH] -selected populations are listed in Table and the postage-stamp images are shown in Fig. . It must be noted here that our spectral index measurements are the ratio of two ‘average’ quantities, flux at 610 MHz and flux at 1,400 MHz. As such, we cann... |
[MATH] , following [MATH] and for each source we assigned a value of [MATH] between [MATH] and [MATH] . We provided the [MATH] values according to two models, one where [MATH] is random and another where [MATH] depends on [MATH] as [MATH] [MATH] [MATH] . Below 50 [MATH] Jy which we introduce an asymptote to [MATH] as [... |
[MATH] 30–75 [MATH] Jy is 2.0–2.5 (Fomalont et al. 2006; Biggs & Ivison 2006). If this holds at fainter fluxes, our method should not significantly bias the measured spectral indices; indeed, it probably gives values closer to the true spectral index than surveys which use detected sources. |
The spectral indices are consistent, within the errors, across all the samples. The very faint flux regimes we are probing here are below any other work which has measured spectral indices. The average for the whole [MATH] -selected sample at [MATH] ) is [MATH] This is consistent with values for [MATH] in star-forming ... |
then sBZK galaxies if radio-quiet AGN were the dominant contributors to the stacked radio flux. There is a hint that such a flattening is present but, with the current noise levels on the pBZK stack, this difference is not significant. |
The corrections we have had to make for integrated flux versus peak flux mean that there are further systematic errors ( [MATH] atop those quoted (which are from the stack statistics only). The GMRT beam is particularly unpleasant in the UDS field due to poor [MATH] coverage, meaning the correction factor to |
clean -integrated flux density is more uncertain than that derived at 1,400 MHz, where it was clear that source size and positional errors were producing a broadening compared to the VLA beam. Although we can compare samples in our data in a consistent way and are confident that they are as similar or different as they... |
5.2 Submm stacking of BzK galaxies As another test of the reliability of the radio as a tracer of star formation, we compared it to the SFR from stacking the sBzK and pBzK samples into the submm map of the SXDF from the SHADES survey (Mortier et al. 2005; Coppin et al. 2006). Noise-weighted means and medians were calcu... |
[MATH] galaxies. For the mean stacks, we excluded pixels with [MATH] . The submm beam is 14 arcsec and so the stacked pixel fluxes should represent the total flux for the [MATH] galaxies. To check that there were no systematic effects, we examined the distribution of pixel signal-to-noise ratios (SNRs) in the map, as s... |
The resulting 850- [MATH] m stacked flux was extrapolated to [MATH] using a modified black-body SED ( [MATH] ) with a dust emissivity index [MATH] and dust temperatures of 30 and 35 . Two reasonable temperatures were chosen to provide an indication of the effect temperature has on the estimate of [MATH] [MATH] was then... |
The total pBzK sample was not detected and the small number of sources overlapping the submm map means that the upper limit is not sensitive enough to constrain the nature of the radio emission. The upper limit to the submm-derived SFR is consistent with that determined through radio stacking. On splitting the pBzK gal... |
[MATH] and [MATH] , we see a 2.6- [MATH] detection in the submm band in the low-redshift bin. Although not formally significant, this suggests that the [MATH] pBzK galaxies are a strongly star-forming sample, and their submm emission is consistent with radio emission powered by star formation. At [MATH] , there is no s... |
Another way to look at the stacked submm and radio fluxes is in terms of the FIR/radio correlation. The FIR/radio ratio is defined as |
[EQUATION] where [MATH] is given by Helou et al. (1985) as [EQUATION] where [MATH] and [MATH] are the 60- and 100- [MATH] m flux densities in Jy. [MATH] is the radio flux in units of [MATH] and several different frequencies are used in the literature. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.