A new strategy for estimating photometric redshifts of quasars

Based on the SDSS and SDSS-WISE quasar datasets, we put forward two schemes to estimate the photometric redshifts of quasars. Our schemes are based on the idea that the samples are firstly classified into subsamples by a classifier and then a photometric redshift estimation of different subsamples is performed by a regressor. Random Forest is adopted as the core algorithm of the classifiers, while Random Forest and k NN are applied as the key algorithms of regressors. The samples are divided into two subsamples and four subsamples, depending on the redshift distribution. The performances based on different samples, different algorithms and different schemes are compared. The experimental results indicate that the accuracy of photometric redshift estimation for the two schemes generally improves to some extent compared to the original scheme in terms of the percents in (|?z|)/(1+zi) 0.1 and (|?z|)/(1+zi) 0.2 and mean absolute error. Only given the running speed, k NN shows its superiority to Random Forest. The performance of Random Forest is a little better than or comparable to that of k NN with the two datasets. The accuracy based on the SDSS-WISE sample outperforms that based on the SDSS sample no matter by k NN or by Random Rorest. More information from more bands is considered and helpful to improve the accuracy of photometric redshift estimation. Evidently, it can be found that our strategy to estimate photometric redshift is applicable and may be applied to other datasets or other kinds of objects. Only talking about the percent in (|?z|)/(1+zi) 0.3, there is still large room for further improvement in the photometric redshift estimation.     

Recovering the real-space correlation function from photometric redshift surveys

Measurements of clustering in large-scale imaging surveys that make use of photometric redshifts depend on the uncertainties in the redshift determination. We have used light-cone simulations to show how the deprojection method successfully recovers the real-space correlation function when applied to mock photometric redshift surveys. We study how the errors in the redshift determination affect the quality of the recovered two-point correlation function. Considering the expected errors associated with the planned photometric redshift surveys, we conclude that this method provides information on the clustering of matter useful for the estimation of cosmological parameters that depend on the large-scale distribution of galaxies.     

Estimating the redshift distribution of photometric galaxy samples – II. Applications and tests of a new method

In Lima et al. we presented a new method for estimating the redshift distribution,   N ( z )  , of a photometric galaxy sample, using photometric observables and weighted sampling from a spectroscopic subsample of the data. In this paper, we extend this method and explore various applications of it, using both simulations and real data from the Sloan Digital Sky Survey (SDSS). In addition to estimating the redshift distribution for an entire sample, the weighting method enables accurate estimates of the redshift probability distribution,   p ( z )  , for each galaxy in a photometric sample. Use of   p ( z )  in cosmological analyses can substantially reduce biases associated with traditional photometric redshifts, in which a single redshift estimate is associated with each galaxy. The weighting procedure also naturally indicates which galaxies in the photometric sample are expected to have accurate redshift estimates, namely those that lie in regions of photometric-observable space that are well sampled by the spectroscopic subsample. In addition to providing a method that has some advantages over standard photo- z estimates, the weights method can also be used in conjunction with photo- z estimates e.g. by providing improved estimation of   N ( z )  via deconvolution of   N ( z _phot)  and improved estimates of photo- z scatter and bias. We present a publicly available   p ( z )  catalogue for ∼78 million SDSS DR7 galaxies.     

Error analysis of the photometric redshift technique

We present a calculation of the systematic component of the error budget in the photometric redshift technique. We make use of it to describe a simple technique that allows the assignment of confidence limits to redshift measurements obtained through photometric methods. We show that our technique, through the calculation of a redshift probability function, gives complete information on the probable redshift of an object and its associated confidence intervals. This information can and must be used in the calculation of any observable quantity that makes use of the redshift.     

MegaZ-LRG: a photometric redshift catalogue of one million SDSS luminous red galaxies

We describe the construction of MegaZ-LRG, a photometric redshift catalogue of over one million luminous red galaxies (LRGs) in the redshift range  0.4 < z < 0.7  with limiting magnitude   i < 20  . The catalogue is selected from the imaging data of the Sloan Digital Sky Survey (SDSS) Data Release 4. The 2dF-SDSS LRG and Quasar (2SLAQ) spectroscopic redshift catalogue of 13 000 intermediate-redshift LRGs provides a photometric redshift training set, allowing use of ann z, a neural network-based photometric-redshift estimator. The rms photometric redshift accuracy obtained for an evaluation set selected from the 2SLAQ sample is  σ _z = 0.049  averaged over all galaxies, and  σ _z = 0.040  for a brighter subsample  ( i < 19.0)  . The catalogue is expected to contain ∼5 per cent stellar contamination. The ann z code is used to compute a refined star/galaxy probability based on a range of photometric parameters; this allows the contamination fraction to be reduced to 2 per cent with negligible loss of genuine galaxies. The MegaZ-LRG catalogue is publicly available on the World Wide Web from http://www.2slaq.info .     

CDFS天区中喑蓝星系的测光红移研究

从COMBO-17数字巡天数据里,选择了CDFS(Chandra Deep Field South)天区中1231个测光红移在0.1～0.3之间的暗蓝星系作为样本,研究了这些星系分别在只有光学波段和光学加近红外波段数据情况下做测光红移得到的红移分布,以及这些星系在静止参考系下的能谱分布(Spectral Energy Distributions,SEDs)特征.结果表明有183个星系在利用光学加近红外波段数据做测光红移时得到的红移大于1.2,它们的误差为0.046,提高测光的信噪比也有利于区分这类被光学波段误认为低红移的星系.这些暗蓝星系中高红移星系的观测近红外流量相对于光学流量有上升的趋势,而低红移星系的观测近红外流量相对于光学流量有下降的趋势.     

Photometric redshifts: spectroscopy AT Δλ / λ ≈ 0.1

I discuss the advantages that photometric redshift techniques offer over traditional (spectroscopic) redshift determination methods. It is shown that the former represents the only means we have of studying the faint distant galaxy population as a whole, and that, in its range of applicability, it delivers excellent results that can add to our knowledge of galaxy formation and evolution processes. Along this line, I also present some of the results extracted from a photometric redshift catalogue of galaxies in the Hubble Deep Field (HDF), including measurements of the star formation density in the high-redshift Universe, morphological evolution of galaxies, and detection of some of the most distant galaxies ever observed. This revised version was published online in July 2006 with corrections to the Cover Date.     

Reconstructing galaxy fundamental distributions and scaling relations from photometric redshift surveys. Applications to the SDSS early-type sample

Noisy distance estimates associated with photometric rather than spectroscopic redshifts lead to a biased estimate of the luminosity distribution, and produce a correlated misestimate of the sizes. We consider a sample of early-type galaxies from the Sloan Digital Sky Survey Data Release 6 for which both spectroscopic and photometric information is available, and apply the generalization of the V _max method to correct for these biases. We show that our technique recovers the true redshift, magnitude and size distributions, as well as the true size–luminosity relation. We find that using only 10 per cent of the spectroscopic information randomly spaced in our catalogue is sufficient for the reconstructions to be accurate within  ∼3 per cent  , when the photometric redshift error is  δ z ≃ 0.038  . We then address the problem of extending our method to deep redshift catalogues, where only photometric information is available. In addition to the specific applications outlined here, our technique impacts a broader range of studies, when at least one distance-dependent quantity is involved. It is particularly relevant for the next generation of surveys, some of which will only have photometric information.     

Light-cone distortion of the clustering and abundance of massive galaxies at high redshifts

Observational surveys of galaxies are not trivially related to single-epoch snapshots from computer simulations. Observationally, an increase in the distance along the line of sight corresponds to an earlier cosmic time at which the properties of the surveyed galaxy population may change. The effect of observing a survey volume along the light cone must be considered in the regime where the mass function of galaxies varies exponentially with redshift. This occurs when the haloes under consideration are rare, that is either when they are very massive or observed at high redshift. While the effect of the light cone is negligible for narrow-band surveys of Lyα emitters, it can be significant for dropout surveys of Lyman-break galaxies (LBGs) where the selection functions of the photometric bands are broad. Since there are exponentially more haloes at the low-redshift end of the survey, the low-redshift tail of the selection function contains a disproportionate fraction of the galaxies observed in the survey. This leads to a redshift probability distribution for the dropout LBGs with a mean less than that of the photometric selection function (PHSF) by an amount of order the standard deviation of the PHSF. The inferred mass function of galaxies is then shallower than the true mass function at a single redshift with the abundance at the high-mass end being twice or more as large as expected. Moreover, the statistical moments of the count of galaxies calculated ignoring the light-cone effect deviate from the actual values.     

Classical cosmological tests for galaxies of the Hubble Ultra Deep Field

Images of the Hubble Ultra Deep Field are analyzed to obtain a catalog of galaxies for which the angular sizes, surface brightness, photometric redshifts, and absolute magnitudes are found. The catalog contains a total of about 4000 galaxies identified at a high signal-to-noise ratio, which allows the cosmological relations angular size—redshift and surface brightness-redshift to be analyzed. The parameters of the evolution of linear sizes and surface brightness of distant galaxies in the redshift interval 0.5–6.5 are estimated in terms of a grid of cosmological models with different density parameters (Ω _V ; Ω _m ). The distribution of photometric redshifts of galaxies is analyzed and possible superlarge inhomogeneities in the radial distribution of galaxies are found with scale lengths as large as 2000 Mpc.     

Photometric redshifts with surface brightness priors

We use galaxy surface brightness as prior information to improve photometric redshift (photo- z ) estimation. We apply our template-based photo- z method to imaging data from the ground-based VVDS survey and the space-based GOODS field from HST , and use spectroscopic redshifts to test our photometric redshifts for different galaxy types and redshifts. We find that the surface brightness prior eliminates a large fraction of outliers by lifting the degeneracy between the Lyman and 4000-Å breaks. Bias and scatter are improved by about a factor of 2 with the prior in each redshift bin in the range  0.4 < z < 1.3  , for both the ground and space data. Ongoing and planned surveys from the ground and space will benefit, provided that care is taken in measurements of galaxy sizes and in the application of the prior. We discuss the image quality and signal-to-noise ratio requirements that enable the surface brightness prior to be successfully applied.     

Photometric redshifts of galaxies from SDSS and 2MASS

In order to find the physical parameters which determine the accuracy of pho- tometric redshifts, we compare the spectroscopic and photometric redshifts (photo-z's) for a large sample of ～ 80 000 SDSS-2MASS galaxies. Photo-z's in this paper are es- timated by using the artificial neural network photometric redshift method (ANNz). For a subset of～40000 randomly selected galaxies, we find that the photometric redshift recovers the spectroscopic redshifi distribution very well with rms of 0.016. Our main results are as follows: (1) Using magnitudes directly as input parameters produces more accurate photo-z's than using colors; (2) The inclusion of 2MASS (3, H, Ks) bands does not improve photo-z's significantly, which indicates that near infrared data might not be important for the low-redshift sample; (3) Adding the concentration index (essentially the steepness of the galaxy brightness profile) as an extra input can improve the photo-z's estimation up to～10 percent; (4) Dividing the sample into early- and late-type galaxies by using the concentration index, normal and abnormal galaxies by using the emission line flux ratios, and red and blue galaxies by using color index (g - r), we can improve the accuracy of photo-z's significantly; (5) Our analysis shows that the outliers (where there is a big difference between the spectroscopic and photometric redshifts) are mainly correlated with galaxy types, e.g., most outliers are late-type (blue) galaxies.     

Spectroscopic needs for imaging dark energy experiments

Ongoing and near-future imaging-based dark energy experiments are critically dependent upon photometric redshifts (a.k.a. photo-z’s): i.e., estimates of the redshifts of objects based only on flux information obtained through broad filters. Higher-quality, lower-scatter photo-z’s will result in smaller random errors on cosmological parameters; while systematic errors in photometric redshift estimates, if not constrained, may dominate all other uncertainties from these experiments. The desired optimization and calibration is dependent upon spectroscopic measurements for secure redshift information; this is the key application of galaxy spectroscopy for imaging-based dark energy experiments.Hence, to achieve their full potential, imaging-based experiments will require large sets of objects with spectroscopically-determined redshifts, for two purposes:

• •Training: Objects with known redshift are needed to map out the relationship between object color and z (or, equivalently, to determine empirically-calibrated templates describing the rest-frame spectra of the full range of galaxies, which may be used to predict the color-z relation). The ultimate goal of training is to minimize each moment of the distribution of differences between photometric redshift estimates and the true redshifts of objects, making the relationship between them as tight as possible. The larger and more complete our “training set” of spectroscopic redshifts is, the smaller the RMS photo-z errors should be, increasing the constraining power of imaging experiments.

• Requirements: Spectroscopic redshift measurements for ∼30,000 objects over >∼15 widely-separated regions, each at least ∼20 arcmin in diameter, and reaching the faintest objects used in a given experiment, will likely be necessary if photometric redshifts are to be trained and calibrated with conventional techniques. Larger, more complete samples (i.e., with longer exposure times) can improve photo-z algorithms and reduce scatter further, enhancing the science return from planned experiments greatly (increasing the Dark Energy Task Force figure of merit by up to ∼50%).
• Options: This spectroscopy will most efficiently be done by covering as much of the optical and near-infrared spectrum as possible at modestly high spectral resolution (λ/Δλ > ∼3000), while maximizing the telescope collecting area, field of view on the sky, and multiplexing of simultaneous spectra. The most efficient instrument for this would likely be either the proposed GMACS/MANIFEST spectrograph for the Giant Magellan Telescope or the OPTIMOS spectrograph for the European Extremely Large Telescope, depending on actual properties when built. The PFS spectrograph at Subaru would be next best and available considerably earlier, c. 2018; the proposed ngCFHT and SSST telescopes would have similar capabilities but start later. Other key options, in order of increasing total time required, are the WFOS spectrograph at TMT, MOONS at the VLT, and DESI at the Mayall 4 m telescope (or the similar 4MOST and WEAVE projects); of these, only DESI, MOONS, and PFS are expected to be available before 2020. Table 2-3 of this white paper summarizes the observation time required at each facility for strawman training samples. To attain secure redshift measurements for a high fraction of targeted objects and cover the full redshift span of future experiments, additional near-infrared spectroscopy will also be required; this is best done from space, particularly with WFIRST-2.4 and JWST.
• Calibration: The first several moments of redshift distributions (the mean, RMS redshift dispersion, etc.), must be known to high accuracy for cosmological constraints not to be systematics-dominated (equivalently, the moments of the distribution of differences between photometric and true redshifts could be determined instead). The ultimate goal of calibration is to characterize these moments for every subsample used in analyses - i.e., to minimize the uncertainty in their mean redshift, RMS dispersion, etc. – rather than to make the moments themselves small. Calibration may be done with the same spectroscopic dataset used for training if that dataset is extremely high in redshift completeness (i.e., no populations of galaxies to be used in analyses are systematically missed). Accurate photo-z calibration is necessary for all imaging experiments.
• Requirements: If extremely low levels of systematic incompleteness (<∼0.1%) are attained in training samples, the same datasets described above should be sufficient for calibration. However, existing deep spectroscopic surveys have failed to yield secure redshifts for 30–60% of targets, so that would require very large improvements over past experience. This incompleteness would be a limiting factor for training, but catastrophic for calibration. If <∼0.1% incompleteness is not attainable, the best known option for calibration of photometric redshifts is to utilize cross-correlation statistics in some form. The most direct method for this uses cross-correlations between positions on the sky of bright objects of known spectroscopic redshift with the sample of objects that we wish to calibrate the redshift distribution for, measured as a function of spectroscopic z. For such a calibration, redshifts of ∼100,000 objects over at least several hundred square degrees, spanning the full redshift range of the samples used for dark energy, would be necessary.
• Options: The proposed BAO experiment eBOSS would provide sufficient spectroscopy for basic calibrations, particularly for ongoing and near-future imaging experiments. The planned DESI experiment would provide excellent calibration with redundant cross-checks, but will start after the conclusion of some imaging projects. An extension of DESI to the Southern hemisphere would provide the best possible calibration from cross-correlation methods for DES and LSST.

We thus anticipate that our two primary needs for spectroscopy – training and calibration of photometric redshifts – will require two separate solutions. For ongoing and future projects to reach their full potential, new spectroscopic samples of faint objects will be needed for training; those new samples may be suitable for calibration, but the latter possibility is uncertain. In contrast, wide-area samples of bright objects are poorly suited for training, but can provide high-precision calibrations via cross-correlation techniques. Additional training/calibration redshifts and/or host galaxy spectroscopy would enhance the use of supernovae and galaxy clusters for cosmology. We also summarize additional work on photometric redshift techniques that will be needed to prepare for data from ongoing and future dark energy experiments.     

GeneticKNN:a weighted KNN approach supported by genetic algorithm for photometric redshift estimation of quasars

We combine K-nearest neighbors(KNN)with a genetic algorithm(GA)for photometric redshift estimation of quasars,short for GeneticKNN,which is a weighted KNN approach supported by a GA.This approach has two improvements compared to KNN:one is the feature weighted by GA;the other is that the predicted redshift is not the redshift average of K neighbors but the weighted average of median and mean of redshifts for K neighbors,i.e.p×zmedian+(1-p)×zmean.Based on the SDSS and SDSS-WISE quasar samples,we explore the performance of GeneticKNN for photometric redshift estimation,comparing with the other six traditional machine learning methods,i.e.the least absolute shrinkage and selection operator(LASSO),support vector regression(SVR),multi-layer perceptrons(MLP),XGBoost,KNN and random forest.KNN and random forest show their superiority.Considering the easy implementation of KNN,we make improvement on KNN as GeneticKNN and apply GeneticKNN on photometric redshift estimation of quasars.Finally the performance of GeneticKNN is better than that of LASSO,SVR,MLP,XGBoost,KNN and random forest for all cases.Moreover the accuracy is better with the additional WISE magnitudes for the same method.     

Two Novel Approaches for Photometric Redshift Estimation based on SDSS and 2MASS

We investigate two training-set methods: support vector machines (SVMs) and Kernel Regression (KR) for photometric redshift estimation with the data from the databases of Sloan Digital Sky Survey Data Release 5 and Two Micron All Sky Survey. We probe the performances of SVMs and KR for different input patterns. Our experiments show that with more parameters considered, the accuracy does not always increase, and only when appropriate parameters are chosen, the accuracy can improve. For different approaches, the best input pattern is different. With different parameters as input, the optimal bandwidth is dissimilar for KR. The rms errors of photometric redshifts based on SVM and KR methods are less than 0.03 and 0.02, respectively. Strengths and weaknesses of the two approaches are summarized. Compared to other methods of estimating photometric redshifts, they show their superiorities, especially KR, in terms of accuracy.     

Incorporating photometric redshift probability density information into real-space clustering measurements

The use of photometric redshifts in cosmology is increasing. Often, however these photo- z are treated like spectroscopic observations, in that the peak of the photometric redshift, rather than the full probability density function (PDF), is used. This overlooks useful information inherent in the full PDF. We introduce a new real-space estimator for one of the most used cosmological statistics, the two-point correlation function, that weights by the PDF of individual photometric objects in a manner that is optimal when Poisson statistics dominate. As our estimator does not bin based on the PDF peak, it substantially enhances the clustering signal by usefully incorporating information from all photometric objects that overlap the redshift bin of interest. As a real-world application, we measure quasi-stellar object (QSO) clustering in the Sloan Digital Sky Survey (SDSS). We find that our simplest binned estimator improves the clustering signal by a factor equivalent to increasing the survey size by a factor of 2–3. We also introduce a new implementation that fully weights between pairs of objects in constructing the cross-correlation and find that this pair-weighted estimator improves clustering signal in a manner equivalent to increasing the survey size by a factor of 4–5. Our technique uses spectroscopic data to anchor the distance scale and it will be particularly useful where spectroscopic data (e.g. from BOSS) overlap deeper photometry (e.g. from Pan-STARRS, DES or the LSST). We additionally provide simple, informative expressions to determine when our estimator will be competitive with the autocorrelation of spectroscopic objects. Although we use QSOs as an example population, our estimator can and should be applied to any clustering estimate that uses photometric objects.     

Infrared photometric study of type II quasars

For this paper, we collected almost all of the type II quasars so far discovered. Among them, 485 sources have photometric data at JHK bands, mainly from Two-Micron All-Sky Survey observations, 65 sources have IRAS photometric data in at least one of the three IRAS bands at 25, 60 and 100 μm and 15 sources have IRAS photometric data in all three IRAS bands. We find that in nearly half of all type II quasars, both the near- and far-infrared radiations are dominated by starlight or thermal reprocessing of starlight by dust in the underlying galaxy. The infrared radiation of the other group (slightly over half) is dominated by a non-thermal radiation in the near-infrared, and mostly in the far-infrared also (although there is a mixture particularly for the longer wavelengths). It is proposed that for the later group, hidden broad lines may exist in the infrared. On the basis of our and previous results, we also discuss the possibility that there are two distinct classes of type II quasars: 'true' type II quasars without a broad-line region and heavily obscured type I quasars, in full analogy with the case for type II Seyfert galaxies. No relationships can be found for either the near-infrared or the far-infrared colours and the redshift. Correlations between absolute magnitude in the near- and far-infrared with redshift are found, but could be due to a flux limit (Malmquist bias).     

Confirmation of the effectiveness of submm source redshift estimation based on rest-frame radio–FIR photometry

We present a comparison between the published optical, infrared (IR) and CO spectroscopic redshifts of 15 (sub)mm galaxies and their photometric redshifts as derived from long-wavelength (radio–mm–far-IR) photometric data. The redshift accuracy measured for 12 submillimetre (submm) galaxies with at least one robustly determined colour in the radio–mm–far-IR regime is  δ z ≈ 0.30 (rms)  . Despite the wide range of spectral energy distributions in the local galaxies that are used in an unbiased manner as templates, this analysis demonstrates that photometric redshifts can be efficiently derived for submm galaxies with a precision of  δ z < 0.5  using only the rest-frame far-IR to radio wavelength data.     

Photometric classification of normal galaxies and their redshift determination

Two-color diagrams including magnitudes from several photometric systems are investigated with respect to the possibility of determining from the integrated photometry the galaxy type and its redshift. The color indices of galaxies are calculated using the energy distribution curves of Colemanet al. (1980). It is shown that redshift determination at low values ofz requires an ultraviolet magnitude with λ₀ shorter than 300 nm. At redshiftsz>0.6, the infrared magnitude photometry becomes important. Different sources of errors and uncertainties are discussed.     

Optical monitoring of quasars – I. Variability

We present an analysis of quasar variability from data collected during a photometric monitoring of 50 objects carried out at CNPq/Laboratório Nacional de Astrofísicá, Brazil, between 1993 March and 1996 July. A distinctive feature of this survey is its photometric accuracy, ∼0.02  V  mag, achieved through differential photometry with CCD detectors, which allows the detection of faint levels of variability. We find that the relative variability, δ σ L , observed in the V band is anticorrelated with both luminosity and redshift, although we have no means of discovering the dominant relation, given the strong coupling between luminosity and redshift for the objects in our sample. We introduce a model for the dependence of quasar variability on frequency that is consistent with multiwavelength observations of the nuclear variability of the Seyfert galaxy NGC 4151. We show that correcting the observed variability for this effect slightly increases the significance of the trends of variability with luminosity and redshift. Assuming that variability depends only on the luminosity, we show that the corrected variability is anticorrelated with luminosity and is in good agreement with predictions of a simple Poissonian model. The energy derived for the hypothetical pulses, ∼10⁵⁰ erg, agrees well with those obtained in other studies. We also find that the radio-loud objects in our sample tend to be more variable than the radio-quiet ones, for all luminosities and redshifts.