We have a developed a new method for fitting spectral energy distributions (SEDs) to identify and constrain the physical properties of high-redshift (4 < z < 8) galaxies. Our approach uses an implementation of Bayesian based Markov Chain Monte Carlo
(PiMC^2) that allows us to compare observations to arbitrarily complex models and to compute 95% credible intervals that provide robust constraints for the model parameters. The work is presented in 2 sections. In the first, we test PiMC^2 using simulated SEDs to not only confirm the recovery of the known inputs but to assess the limitations of the method and identify potential hazards of SED fitting when applied specifically to high redshift (z>4) galaxies. Our tests reveal five critical results: 1) the ability to confidently constrain metallicity, population ages, and Av all require photometric accuracy better than what is currently achievable (i.e. less than a few percent); 2) the ability to confidently constrain stellar masses (within a factor of two) can be achieved without the need for high-precision photometry; 3) the addition of IRAC photometry does not guarantee that tighter constraints of the stellar masses and ages can be defined; 4) different assumptions about the star formation history can lead to significant biases in mass and age estimates; and 5) we are able to constrain stellar age and Av of objects that are both young and relatively dust free. In the second part of the paper we apply PiMC^2 to 17 4<z<8 objects, including the GRAPES Ly alpha sample (4<z<6), supplemented by HST/WFC3 near-IR observations, and several broad band selected z>6 galaxies. Using PiMC^2, we are able to constrain the stellar mass of these objects and in some cases their stellar age and find no evidence that any of these sources formed at a redshift much larger than z_f=8, a time when the Universe was ~ 0.6 Gyr old.
Of the many ways of detecting high redshift galaxies, the selection of objects due to their redshifted Ly-alpha emission has become one of the most successful. But what types of galaxies are selected in this way? Until recently, Ly-alpha emitters wer
e understood to be small star-forming galaxies, possible building-blocks of larger galaxies. But with increased number of observations of Ly-alpha emitters at lower redshifts, a new picture emerges. Ly-alpha emitters display strong evolution in their properties from higher to lower redshift. It has previously been shown that the fraction of ultra-luminous infrared galaxies (ULIRGs) among the Ly-alpha emitters increases dramatically between redshift three and two. Here, the fraction of AGN among the LAEs is shown to follow a similar evolutionary path. We argue that Ly-alpha emitters are not a homogeneous class of objects, and that the objects selected with this method reflect the general star forming and active galaxy populations at that redshift. Ly-alpha emitters should hence be excellent tracers of galaxy evolution in future simulations and modeling.
Lyman-break galaxies are now regularly found in the high redshift Universe by searching for the break in the galaxy spectrum caused by the Lyman-limit redshifted into the optical or even near-IR. At lower redshift, this break is covered by the GALEX
UV channels and small samples of z ~ 1 LBGs have been presented in the literature. Here we give results from fitting the spectral energy distributions of a small sub-set of low redshift LBGs and demonstrate the advantage of including photometric points derived from HST ACS slitless grism observations. The results show these galaxies to have very young, star forming populations, while still being massive and dusty. LBGs at low and high redshift show remarkable similarities in their properties, indicating that the LBG selection method picks similar galaxies throughout the Universe.
We study the multi-wavelength properties of a set of 171 Ly-alpha emitting candidates at redshift z = 2.25 found in the COSMOS field, with the aim of understanding the underlying stellar populations in the galaxies. We especially seek to understand w
hat the dust contents, ages and stellar masses of the galaxies are, and how they relate to similar properties of Ly-alpha emitters at other redshifts. The candidates here are shown to have different properties from those of Ly-alpha emitters found at higher redshift, by fitting the spectral energy distributions (SEDs) using a Monte-Carlo Markov-Chain technique and including nebular emission in the spectra. The stellar masses, and possibly the dust contents, are higher, with stellar masses in the range log M_* = 8.5 - 11.0 M_sun and A_V = 0.0 - 2.5 mag. Young population ages are well constrained, but the ages of older populations are typically unconstrained. In 15% of the galaxies only a single, young population of stars is observed. We show that the Ly-alpha fluxes of the best fit galaxies are correlated with their dust properties, with higher dust extinction in Ly-alpha faint galaxies. Testing for whether results derived from a light-weighted stack of objects correlate to those found when fitting individual objects we see that stellar masses are robust to stacking, but ages and especially dust extinctions are derived incorrectly from stacks. We conclude that the stellar properties of Ly-alpha emitters at z = 2.25 are different from those at higher redshift and that they are diverse. Ly-alpha selection appears to be tracing systematically different galaxies at different redshifts.
The work of astronomers is getting more complex and advanced as the progress of computer development occurs. With improved computing capabilities and increased data flow, more sophisticated software is required in order to interpret, and fully exploi
t, astronomic data. However, it is not possible for every astronomer to also be a software specialist. As history has shown, the work of scientists always becomes increasingly specialised, and we here argue in favour of another, at least partial, split between programmers and interpreters. In this presentation we outline our vision for a new approach and symbiosis between software specialists and scientists, and present its advantages along with a simple test case.
The traditional view that Ly-alpha emission and dust should be mutually exclusive has been questioned more and more often; most notably, the observations of Ly-alpha emission from ULIRGs seem to counter this view. In this paper we seek to address the
reverse question. How large a fraction of Ly-alpha selected galaxies are ULIRGs? Using two samples of 24/25 Ly-alpha emitting galaxies at z = 0.3/2.3, we perform this test, including results at z = 3.1, and find that, whereas the ULIRG fraction at z = 3.1 is very small, it systematically increases towards lower redshifts. There is a hint that this evolution may be quite sudden and that it happens around a redshift of z ~ 2.5. After measuring the infrared luminosities of the Ly-alpha emitters, we find that they are in the normal to ULIRG range in the lower redshift sample, while the higher redshift galaxies all have luminosities in the ULIRG category. The Ly-alpha escape fractions for these infrared bright galaxies are in the range 1-100 % in the z = 0.3 galaxies, but are very low in the z = 2.3 galaxies, 0.4 % on average. The unobscured star formation rates are very high, ranging from 500 to more than 5000 M_sun/yr, and the dust attenuation derived are in the range 0.0 < A_V < 3.5.
We show that with the simple assumption of no correlation between the Ly-alpha equivalent width and the UV luminosity of a galaxy, the observed distribution of high redshift galaxies in an equivalent width - absolute UV magnitude plane can be reprodu
ced. We further show that there is no dependence between Ly-alpha equivalent width and Ly-alpha luminosity in a sample of Ly-alpha emitters. The test was expanded to Lyman-break galaxies and again no dependence was found. Simultaneously, we show that a recently proposed lack of large equivalent width, UV bright galaxies (Ando et al. 2006) can be explained by a simple observational effect, based on too small survey volumes.
This publication contains the conference summary of the Understanding Lyman-alpha Emitters conference held at the Max Planck Institute for Astronomy in Heidelberg October 6 - 10, 2008. The scope of the conference was to bring together most of the sci
entists working in the field of Lyman-alpha emitters, whether at low or high redshift, or on observational or theoretical aspects, and to summarise how far the field of study of galaxies with Lyman-alpha emission has come. An outlook towards the future of the field was also desired. As part of the conference, two days were dedicated to in total six discussion sessions. The topics were i) new methods and selection methods, ii) morphology, iii) what can the local Universe observations tell us about the high redshift Universe?, iv) clustering, v) SED fitting and vi) Ly-alpha blobs. The chairs of those sessions were asked to summarise the discussions, as presented in these proceedings.
Context: Narrow-band surveys for Ly-alpha emitters (LAEs) is a powerful tool in detecting high, and very high, redshift galaxies. Even though samples are growing at redshifts z = 3 - 6, the nature of these galaxies is still poorly known. Aims: To stu
dy the properties of z = 2.25 LAEs and compare those with the properties of z > 3 LAEs. Methods: We present narrow-band imaging made with the MPG/ESO 2.2m telescope with the WFI detector. We have made a selection for emission-line objects and find 170 candidate typical LAEs and 17 candidates which we regard as high UV-transmission LAEs. We have derived the magnitudes of these objects in 8 bands from u* to Ks, and studied if they have X-ray and/or radio counterparts. Results: We show that there has been significant evolution in the properties of LAEs between redshift z ~ 3 and z = 2.25. The spread in spectral energy distributions (SEDs) at the lower redshift is larger and we detect a significant AGN contribution in the sample. The distribution of the equivalent widths is narrower than at z ~ 3, with only a few candidates with rest-frame equivalent width above the predicted limit of 240 A. The star formation rates derived from the Ly-alpha emission compared to that derived from the UV emission are lower by on average a factor of ~ 1.8, indicating a large absorption of dust. Conclusion: LAEs at redshift z = 2.25 may be more evolved than LAEs at higher redshift. The red SEDs imply more massive, older and/or more dusty galaxies at lower redshift than observed at higher redshifts. The decrease in equivalent widths and star formation rates indicate more quiescent galaxies, with in general less star formation than in higher redshift galaxies. At z = 2.25, AGN appear to be more abundant and also to contribute more to the LAE population. [Abridged]
Context: Many current and future surveys aim to detect the highest redshift (z >~ 7) sources through their Lyman-alpha (Ly-alpha) emission, using the narrow-band imaging method. However, to date the surveys have only yielded non-detections and upper
limits as no survey has reached the necessary combination of depth and area to detect these very young star forming galaxies. Aims: We aim to calculate model luminosity functions and mock surveys of Ly-alpha emitters at z >~ 7 based on a variety of approaches. Methods: We calculate model luminosity functions at different redshifts based on three different approaches: a semi-analytical model based on CDM, a simple phenomenological model, and an extrapolation of observed Schechter functions at lower redshifts. The results of the first two models are compared with observations made at redshifts z ~ 5.7 and z ~ 6.5, and they are then extrapolated to higher redshift. Results: We present model luminosity functions for redshifts between z = 7 - 12.5 and give specific number predictions for future planned or possible narrow-band surveys for Ly-alpha emitters. We also investigate what constraints future observations will be able to place on the Ly-alpha luminosity function at very high redshift. Conclusion: It should be possible to observe z = 7 - 10 Ly-alpha emitters with present or near-future instruments if enough observing time is allocated. In particular, large area surveys such as ELVIS (Emission Line galaxies with VISTA Survey) will be useful in collecting a large sample. However, to get a large enough sample to constrain well the z >= 10 Ly-alpha luminosity function, instruments further in the future, such as an ELT, will be necessary.
