Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userThe PAU Survey: narrowband photometric redshifts using Gaussian processes
144
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
We study the performance of the hybrid template-machine-learning photometric redshift (photo-$z$) algorithm Delight, which uses Gaussian processes, on a subset of the early data release of the Physics of the Accelerating Universe Survey (PAUS). We calibrate the fluxes of the $40$ PAUS narrow bands with $6$ broadband fluxes ($uBVriz$) in the COSMOS field using three different methods, including a new method which utilises the correlation between the apparent size and overall flux of the galaxy. We use a rich set of empirically derived galaxy spectral templates as guides to train the Gaussian process, and we show that our results are competitive with other standard photometric redshift algorithms. Delight achieves a photo-$z$ $68$th percentile error of $sigma_{68}=0.0081(1+z)$ without any quality cut for galaxies with $i_mathrm{auto}<22.5$ as compared to $0.0089(1+z)$ and $0.0202(1+z)$ for the BPz and ANNz2 codes, respectively. Delight is also shown to produce more accurate probability distribution functions for individual redshift estimates than BPz and ANNz2. Common photo-$z$ outliers of Delight and BCNz2 (previously applied to PAUS) are found to be primarily caused by outliers in the narrowband fluxes, with a small number of cases potentially indicating spectroscopic redshift failures in the reference sample. In the process, we introduce performance metrics derived from the results of BCNz2 and Delight, allowing us to achieve a photo-$z$ quality of $sigma_{68}<0.0035(1+z)$ at a magnitude of $i_mathrm{auto}<22.5$ while keeping $50$ per cent objects of the galaxy sample.
rate research
Read More
In this paper we introduce the textsc{Deepz} deep learning photometric redshift (photo-$z$) code. As a test case, we apply the code to the PAU survey (PAUS) data in the COSMOS field. textsc{Deepz} reduces the $sigma_{68}$ scatter statistic by 50% at $i_{rm AB}=22.5$ compared to existing algorithms. This improvement is achieved through various methods, including transfer learning from simulations where the training set consists of simulations as well as observations, which reduces the need for training data. The redshift probability distribution is estimated with a mixture density network (MDN), which produces accurate redshift distributions. Our code includes an autoencoder to reduce noise and extract features from the galaxy SEDs. It also benefits from combining multiple networks, which lowers the photo-$z$ scatter by 10 percent. Furthermore, training with randomly constructed coadded fluxes adds information about individual exposures, reducing the impact of photometric outliers. In addition to opening up the route for higher redshift precision with narrow bands, these machine learning techniques can also be valuable for broad-band surveys.
In this paper we present and characterize a nearest-neighbors color-matching photometric redshift estimator that features a direct relationship between the precision and accuracy of the input magnitudes and the output photometric redshifts. This aspect makes our estimator an ideal tool for evaluating the impact of changes to LSST survey parameters that affect the measurement errors of the photometry, which is the main motivation of our work (i.e., it is not intended to provide the best photometric redshifts for LSST data). We show how the photometric redshifts will improve with time over the 10-year LSST survey and confirm that the nominal distribution of visits per filter provides the most accurate photo-$z$ results. The LSST survey strategy naturally produces observations over a range of airmass, which offers the opportunity of using an SED- and $z$-dependent atmospheric affect on the observed photometry as a color-independent redshift indicator. We show that measuring this airmass effect and including it as a prior has the potential to improve the photometric redshifts and can ameliorate extreme outliers, but that it will only be adequately measured for the brightest galaxies, which limits its overall impact on LSST photometric redshifts. We furthermore demonstrate how this airmass effect can induce a bias in the photo-$z$ results, and caution against survey strategies that prioritize high-airmass observations for the purpose of improving this prior. Ultimately, we intend for this work to serve as a guide for the expectations and preparations of the LSST science community with regards to the minimum quality of photo-$z$ as the survey progresses.
We present methods for emulating the matter power spectrum which effectively combine information from cosmological $N$-body simulations at different resolutions. An emulator allows estimation of simulation output by interpolating across the parameter space of a handful of simulations. We present the first implementation of multi-fidelity emulation in cosmology, where many low-resolution simulations are combined with a few high-resolution simulations to achieve an increased emulation accuracy. The power spectrums dependence on cosmology is learned from the low-resolution simulations, which are in turn calibrated using high-resolution simulations. We show that our multi-fidelity emulator can achieve percent-level accuracy on average with only $3$ high-fidelity simulations and outperforms a single-fidelity emulator that uses $11$ simulations. With a fixed number of high-fidelity training simulations, we show that our multi-fidelity emulator is $simeq 100$ times better than a single-fidelity emulator at $k leq 2 ,htextrm{Mpc}{^{-1}}$, and $simeq 20$ times better at $3 leq k < 6.4 ,htextrm{Mpc}{^{-1}}$. Multi-fidelity emulation is fast to train, using only a simple modification to standard Gaussian processes. Our proposed emulator shows a new way to predict non-linear scales by fusing simulations from different fidelities.
We present a robust method to estimate the redshift of galaxies using Pan-STARRS1 photometric data. Our method is an adaptation of the one proposed by Beck et al. (2016) for the SDSS Data Release 12. It uses a training set of 2313724 galaxies for which the spectroscopic redshift is obtained from SDSS, and magnitudes and colours are obtained from the Pan-STARRS1 Data Release 2 survey. The photometric redshift of a galaxy is then estimated by means of a local linear regression in a 5-dimensional magnitude and colour space. Our method achieves an average bias of $overline{Delta z_{rm norm}}=-2.01 times 10^{-4}$, a standard deviation of $sigma(Delta z_{rm norm})=0.0298$, and an outlier rate of $P_o=4.32%$ when cross-validating on the training set. Even though the relation between each of the Pan-STARRS1 colours and the spectroscopic redshifts is noisier than for SDSS colours, the results obtained by our method are very close to those yielded by SDSS data. The proposed method has the additional advantage of allowing the estimation of photometric redshifts on a larger portion of the sky ($sim 3/4$ vs $sim 1/3$). The training set and the code implementing this method are publicly available at www.testaddress.com.
Precision photometric redshifts will be essential for extracting cosmological parameters from the next generation of wide-area imaging surveys. In this paper we introduce a photometric redshift algorithm, ArborZ, based on the machine-learning technique of Boosted Decision Trees. We study the algorithm using galaxies from the Sloan Digital Sky Survey and from mock catalogs intended to simulate both the SDSS and the upcoming Dark Energy Survey. We show that it improves upon the performance of existing algorithms. Moreover, the method naturally leads to the reconstruction of a full probability density function (PDF) for the photometric redshift of each galaxy, not merely a single best estimate and error, and also provides a photo-z quality figure-of-merit for each galaxy that can be used to reject outliers. We show that the stacked PDFs yield a more accurate reconstruction of the redshift distribution N(z). We discuss limitations of the current algorithm and ideas for future work.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
