arXiv subjects
Publications and source records attributed to Ofer Lahav.
When do we stop an ongoing science project to make room for something new? Decision-making is a complex process, ranging from budgetary considerations and tension between ongoing projects, to progress assessments and allowance for novel science developments.
Knowing the redshift of galaxies is one of the first requirements of many cosmological experiments, and as it's impossible to perform spectroscopy for every galaxy being observed, photometric redshift (photo-z) estimations are still of particular interest. Here, we investigate different deep learning methods for obtaining photo-z estimates directly from images, comparing these with traditional machine learning algorithms which make use of magnitudes retrieved through photometry. As well as testing a convolutional neural network (CNN) and inception-module CNN, we introduce a novel mixed-input model which allows for both images and magnitude data to be used in the same model as a way of further improving the estimated redshifts. We also perform benchmarking as a way of demonstrating the performance and scalability of the different algorithms. The data used in the study comes entirely from the Sloan Digital Sky Survey (SDSS) from which 1 million galaxies were used, each having 5-filter (ugriz) images with complete photometry and a spectroscopic redshift which was taken as the ground truth. The mixed-input inception CNN achieved a mean squared error (MSE)=0.009, which was a significant improvement (30%) over the traditional Random Forest (RF), and the model performed even better at lower redshifts achieving a MSE=0.0007 (a 50% improvement over the RF) in the range of z<0.3. This method could be hugely beneficial to upcoming surveys such as the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) which will require vast numbers of photo-z estimates produced as quickly and accurately as possible.
We present full-sky maps of the Integrated Sachs-Wolfe effect (ISW) for the MICE Grand Challenge lightcone simulation up to redshift 1.4. The maps are constructed in the linear regime using spherical Bessel transforms. We compare and contrast this procedure against analytical approximations found in the literature. By computing the ISW in the linear regime, we remove the substantial computing and storage resources required to calculate the non-linear Rees-Sciama effect. Since the linear ISW at low redshift $z\leq1$, at large angular scales, and after matter domination is $\sim 10^{2}\times$ larger in $ΔT/T$, this has a negligible impact on the maps produced and only becomes relevant on scales which are dominated by cosmic microwave background (CMB) anisotropies. The MICE simulation products have been extensively used for studies involving current and future galaxy surveys. The availability of these maps will allow MICE to be used for future galaxy and CMB cross-correlation studies, ISW reconstruction studies, and ISW void-stacking studies probed by galaxy surveys such as Dark Energy Survey, Dark Energy Spectroscopic Instrument, Euclid, and Rubin Legacy Survey of Space and Time. The pipeline developed in this study is provided as a public Python package pyGenISW. This could be used in the future studies for constructing the ISW from existing and future simulation suites probing vast sets of cosmological parameters and models.
Measurements of large-scale structure are interpreted using theoretical predictions for the matter distribution, including potential impacts of baryonic physics. We constrain the feedback strength of baryons jointly with cosmology using weak lensing and galaxy clustering observables (3$\times$2pt) of Dark Energy Survey (DES) Year 1 data in combination with external information from baryon acoustic oscillations (BAO) and Planck cosmic microwave background polarization. Our baryon modeling is informed by a set of hydrodynamical simulations that span a variety of baryon scenarios; we span this space via a Principal Component (PC) analysis of the summary statistics extracted from these simulations. We show that at the level of DES Y1 constraining power, one PC is sufficient to describe the variation of baryonic effects in the observables, and the first PC amplitude ($Q_1$) generally reflects the strength of baryon feedback. With the upper limit of $Q_1$ prior being bound by the Illustris feedback scenarios, we reach $\sim 20\%$ improvement in the constraint of $S_8=σ_8(Ω_{\rm m}/0.3)^{0.5}=0.788^{+0.018}_{-0.021}$ compared to the original DES 3$\times$2pt analysis. This gain is driven by the inclusion of small-scale cosmic shear information down to 2.5 arcmin, which was excluded in previous DES analyses that did not model baryonic physics. We obtain $S_8=0.781^{+0.014}_{-0.015}$ for the combined DES Y1+Planck EE+BAO analysis with a non-informative $Q_1$ prior. In terms of the baryon constraints, we measure $Q_1=1.14^{+2.20}_{-2.80}$ for DES Y1 only and $Q_1=1.42^{+1.63}_{-1.48}$ for DESY1+Planck EE+BAO, allowing us to exclude one of the most extreme AGN feedback hydrodynamical scenario at more than $2 σ$.
Obtaining accurate photometric redshift estimations is an important aspect of cosmology, remaining a prerequisite of many analyses. In creating novel methods to produce redshift estimations, there has been a shift towards using machine learning techniques. However, there has not been as much of a focus on how well different machine learning methods scale or perform with the ever-increasing amounts of data being produced. Here, we introduce a benchmark designed to analyse the performance and scalability of different supervised machine learning methods for photometric redshift estimation. Making use of the Sloan Digital Sky Survey (SDSS - DR12) dataset, we analysed a variety of the most used machine learning algorithms. By scaling the number of galaxies used to train and test the algorithms up to one million, we obtained several metrics demonstrating the algorithms' performance and scalability for this task. Furthermore, by introducing a new optimisation method, time-considered optimisation, we were able to demonstrate how a small concession of error can allow for a great improvement in efficiency. From the algorithms tested we found that the Random Forest performed best in terms of error with a mean squared error, MSE = 0.0042; however, as other algorithms such as Boosted Decision Trees and k-Nearest Neighbours performed incredibly similarly, we used our benchmarks to demonstrate how different algorithms could be superior in different scenarios. We believe benchmarks such as this will become even more vital with upcoming surveys, such as LSST, which will capture billions of galaxies requiring photometric redshifts.
We use Density Estimation Likelihood-Free Inference, $Λ$ Cold Dark Matter simulations of $\sim 2M$ galaxy pairs, and data from Gaia and the Hubble Space Telescope to infer the sum of the masses of the Milky Way and Andromeda (M31) galaxies, the two main components of the Local Group. This method overcomes most of the approximations of the traditional timing argument, makes the writing of a theoretical likelihood unnecessary, and allows the non-linear modelling of observational errors that take into account correlations in the data and non-Gaussian distributions. We obtain an $M_{200}$ mass estimate $M_{\rm MW+M31} = 4.6^{+2.3}_{-1.8} \times 10^{12} M_{\odot}$ ($68 \%$ C.L.), in agreement with previous estimates both for the sum of the two masses and for the individual masses. This result is not only one of the most reliable estimates of the sum of the two masses to date, but is also an illustration of likelihood-free inference in a problem with only one parameter and only three data points.
The Cosmological Constant Lambda, a concept introduced by Einstein in 1917, has been with us ever since in different variants and incarnations, including the broader concept of Dark Energy. Current observations are consistent with a value of Lambda corresponding to about present-epoch 70% of the critical density of the Universe. This is causing the speeding up (acceleration) of the expansion of the Universe over the past 6 billion years, a discovery recognised by the 2011 Nobel Prize in Physics. Coupled with the flatness of the Universe and the amount of 30% matter (5% baryonic and 25% Cold Dark Matter), this forms the so-called Lambda-CDM standard model, which has survived many observational tests over about 30 years. However, there are currently indications of inconsistencies (`tensions' ) within Lambda-CDM on different values of the Hubble Constant and the clumpiness factor. Also, time variation of Dark Energy and slight deviations from General Relativity are not ruled out yet. Several grand projects are underway to test Lambda-CDM further and to estimate the cosmological parameters to sub-percent level. If Lambda-CDM will remain the standard model, then the ball is back in the theoreticians' court, to explain the physical meaning of Lambda. Is Lambda an alteration to the geometry of the Universe, or the energy of the vacuum? Or maybe it is something different, that manifests a yet unknown higher-level theory?
We measure the projected number density profiles of galaxies and the splashback feature in clusters selected by the Sunyaev--Zeldovich (SZ) effect from the Advanced Atacama Cosmology Telescope (AdvACT) survey using galaxies observed by the Dark Energy Survey (DES). The splashback radius for the complete galaxy sample is consistent with theoretical measurements from CDM-only simulations, and is located at $2.4^{+0.3}_{-0.4}$ Mpc $h^{-1}$. We split the sample based on galaxy color and find significant differences in the profile shapes. Red galaxies and those in the green valley show a splashback-like minimum in their slope profile consistent with theoretical predictions, while the bluest galaxies show a weak feature that appears at a smaller radius. We develop a mapping of galaxies to subhalos in $N$-body simulations by splitting subhalos based on infall time onto the cluster halos. We find that the location of the steepest slope and differences in the shapes of the profiles can be mapped to differences in the average time of infall of galaxies of different colors. The minima of the slope in the galaxy profiles trace a discontinuity in the phase space of dark matter halos. By relating spatial profiles to infall time for galaxies of different colours, we can use splashback as a clock to understand galaxy quenching. We find that red galaxies have on average been in their clusters for over $3.2 ~\rm Gyrs$, green galaxies about $2.2 ~\rm Gyrs$, while blue galaxies have been accreted most recently and have not reached apocenter. Using the information from the complete radial profiles, we fit a simple quenching model and find that the onset of galaxy quenching in clusters occurs after a delay of about a gigayear, and that galaxies quench rapidly thereafter with an exponential timescale of $0.6$ Gyr.
We explore mass estimation of the Local Group via the use of the simple, dynamical `timing argument' in the context of a variety of theories of dark energy and modified gravity: a cosmological constant, a perfect fluid with constant equation of state $w$, quintessence (minimally coupled scalar field), MOND, and symmetrons (coupled scalar field). We explore generic coupled scalar field theories, with the symmetron model as an explicit example. We find that theories which attempt to eliminate dark matter by fitting rotation curves produce mass estimates in the timing argument which are not compatible with the luminous mass of the galaxies alone. Assuming that the galaxies are approaching their first encounter, MOND gives of around $2.7\times 10^{10} M_\odot$, roughly 10\% of the luminous mass of the LG, although a higher mass can be obtained in the case of a previous fly-by event between the MW and M31. The symmetron model suggests a mass too high to be explained without additional dark matter ($\mathcal{O}(10^{12}) M_\odot$), suggesting that there is a missing mass problem in this model. We also demonstrate that tensions in measurements of $H_0$ can produce an uncertainty in the Local Group mass estimate comparable to observational uncertainties on the separation and relative velocity of the galaxies, with values for the mass ranging from $4.5 - 5.4 \times 10^{12} M_{\odot}$ varying $h$ between 0.67 and 0.76.
We present the GEOMAX algorithm and its Python implementation for a two-step compression of bispectrum measurements. The first step groups bispectra by the geometric properties of their arguments; the second step then maximises the Fisher information with respect to a chosen set of model parameters in each group. The algorithm only requires the derivatives of the data vector with respect to the parameters and a small number of mock data, producing an effective, non-linear compression. By applying GEOMAX to bispectrum monopole measurements from BOSS DR12 CMASS redshift-space galaxy clustering data, we reduce the $68\%$ credible intervals for the inferred parameters $\left(b_1,b_2,f,σ_8\right)$ by $\left(50.4\%,56.1\%,33.2\%,38.3\%\right)$ with respect to standard MCMC on the full data vector. We run the analysis and comparison between compression methods over one hundred galaxy mocks to test the statistical significance of the improvements. On average GEOMAX performs $\sim15\%$ better than geometrical or maximal linear compression alone and is consistent with being lossless. Given its flexibility, the GEOMAX approach has the potential to optimally exploit three-point statistics of various cosmological probes like weak lensing or line-intensity maps from current and future cosmological data-sets such as DESI, Euclid, PFS and SKA.
In this work we investigate the systematic uncertainties that arise from the calculation of the peculiar velocity when estimating the Hubble constant ($H_0$) from gravitational wave standard sirens. We study the GW170817 event and the estimation of the peculiar velocity of its host galaxy, NGC 4993, when using Gaussian smoothing over nearby galaxies. NGC 4993 being a relatively nearby galaxy, at $\sim 40 \ {\rm Mpc}$ away, is subject to a significant effect of peculiar velocities. We demonstrate a direct dependence of the estimated peculiar velocity value on the choice of smoothing scale. We show that when not accounting for this systematic, a bias of $\sim 200 \ {\rm km \ s ^{-1}}$ in the peculiar velocity incurs a bias of $\sim 4 \ {\rm km \ s ^{-1} \ Mpc^{-1}}$ on the Hubble constant. We formulate a Bayesian model that accounts for the dependence of the peculiar velocity on the smoothing scale and by marginalising over this parameter we remove the need for a choice of smoothing scale. The proposed model yields $H_0 = 68.6 ^{+14.0}_{-8.5}~{\rm km\ s^{-1}\ Mpc^{-1}}$. We demonstrate that under this model a more robust unbiased estimate of the Hubble constant from nearby GW sources is obtained.
Cosmology now has a standard model - a remarkably simple description of the universe, its contents and its history. A symposium held last September in Cambridge, UK, gave this model a 'health check' and discussed fascinating questions that lie beyond it.
We perform a joint analysis of the counts and weak lensing signal of redMaPPer clusters selected from the Dark Energy Survey (DES) Year 1 dataset. Our analysis uses the same shear and source photometric redshifts estimates as were used in the DES combined probes analysis. Our analysis results in surprisingly low values for $S_8 =σ_8(Ω_{\rm m}/0.3)^{0.5}= 0.65\pm 0.04$, driven by a low matter density parameter, $Ω_{\rm m}=0.179^{+0.031}_{-0.038}$, with $σ_8-Ω_{\rm m}$ posteriors in $2.4σ$ tension with the DES Y1 3x2pt results, and in $5.6σ$ with the Planck CMB analysis. These results include the impact of post-unblinding changes to the analysis, which did not improve the level of consistency with other data sets compared to the results obtained at the unblinding. The fact that multiple cosmological probes (supernovae, baryon acoustic oscillations, cosmic shear, galaxy clustering and CMB anisotropies), and other galaxy cluster analyses all favor significantly higher matter densities suggests the presence of systematic errors in the data or an incomplete modeling of the relevant physics. Cross checks with X-ray and microwave data, as well as independent constraints on the observable--mass relation from SZ selected clusters, suggest that the discrepancy resides in our modeling of the weak lensing signal rather than the cluster abundance. Repeating our analysis using a higher richness threshold ($λ\ge 30$) significantly reduces the tension with other probes, and points to one or more richness-dependent effects not captured by our model.
We present the first reconstruction of dark matter maps from weak lensing observational data using deep learning. We train a convolution neural network (CNN) with a Unet based architecture on over $3.6\times10^5$ simulated data realizations with non-Gaussian shape noise and with cosmological parameters varying over a broad prior distribution. We interpret our newly created DES SV map as an approximation of the posterior mean $P(κ| γ)$ of the convergence given observed shear. Our DeepMass method is substantially more accurate than existing mass-mapping methods. With a validation set of 8000 simulated DES SV data realizations, compared to Wiener filtering with a fixed power spectrum, the DeepMass method improved the mean-square-error (MSE) by 11 per cent. With N-body simulated MICE mock data, we show that Wiener filtering with the optimal known power spectrum still gives a worse MSE than our generalized method with no input cosmological parameters; we show that the improvement is driven by the non-linear structures in the convergence. With higher galaxy density in future weak lensing data unveiling more non-linear scales, it is likely that deep learning will be a leading approach for mass mapping with Euclid and LSST.
Jacob Bekenstein's pioneering contributions to the foundations of Gravity cover a broad range of scales, from Black Holes to the entire Universe. He is well known as the father of Black Hole Thermodynamics and the inventor of the TeVeS modified gravity theory. I reflect here on my interaction with him, from the time he supervised me for my MSc project at Ben-Gurion University in the early 1980s to his sabbatical visit at UCL in 2014, about a year before he passed away (sadly just a month before the first LIGO discovery of Gravitational Waves from a Binary Black Hole). His quote "I feel more comfortable in the world if I understand why things are happening" reflects the way he viewed the world as a Physicist as well as his personal belief. He also allegedly said "Sometimes theories are smarter than their inventors".
This is a review article for The Review of Particle Physics 2020 (aka the Particle Data Book). It forms a compact review of knowledge of the cosmological parameters at the end of 2019. Topics included are Parametrizing the Universe; Extensions to the standard model; Probes; Bringing observations together; Outlook for the future.
Cosmological studies of large-scale structure have relied on two-point statistics, not fully exploiting the rich structure of the cosmic web. In this paper we show how to capture some of this cosmic web information by using the minimum spanning tree (MST), for the first time using it to estimate cosmological parameters in simulations. Discrete tracers of dark matter such as galaxies, $N$-body particles or haloes are used as nodes to construct a unique graph, the MST, that traces skeletal structure. We study the dependence of the MST on cosmological parameters using haloes from a suite of COLA simulations with a box size of $250\ h^{-1}{\rm Mpc}$, varying the amplitude of scalar fluctuations $\left(A_{\rm s}\right)$, matter density $\left(Ω_{\rm m}\right)$, and neutrino mass $\left(\sum m_ν\right)$. The power spectrum $P$ and bispectrum $B$ are measured for wavenumbers between $0.125$ and $0.5$ $h{\rm Mpc}^{-1}$, while a corresponding lower cut of $\sim12.6$ $h^{-1}{\rm Mpc}$ is applied to the MST. The constraints from the individual methods are fairly similar but when combined we see improved $1σ$ constraints of $\sim 17\%$ ($\sim 12\%$) on $Ω_{\rm m}$ and $\sim 12\%$ ($\sim 10\%$) on $A_{\rm s}$ with respect to $P$ ($P+B$) thus showing the MST is providing additional information. The MST can be applied to current and future spectroscopic surveys (BOSS, DESI, Euclid, PSF, WFIRST, and 4MOST) in 3D and photometric surveys (DES and LSST) in tomographic shells to constrain parameters and/or test systematics.
We propose a principled Bayesian method for quantifying tension between correlated datasets with wide uninformative parameter priors. This is achieved by extending the Suspiciousness statistic, which is insensitive to priors. Our method uses global summary statistics, and as such it can be used as a diagnostic for internal consistency. We show how our approach can be combined with methods that use parameter space and data space to identify the existing internal discrepancies. As an example, we use it to test the internal consistency of the KiDS-450 data in 4 photometric redshift bins, and to recover controlled internal discrepancies in simulated KiDS data. We propose this as a diagnostic of internal consistency for present and future cosmological surveys, and as a tension metric for data sets that have non-negligible correlation, such as LSST and Euclid.