Dec 2018
Dec. 5, Wednesday
12:00 pm (DESI Lunch, please note you need LBL bus pass or other access to get in)
Ben Wibking, OSU
LBL 50-4205
Predicting small-scale galaxy clustering and lensing with simulations
The statistical precision of galaxy clustering and lensing measurements
is greatest at small (~Mpc) scales, but the complexities of structure
formation and galaxy bias make this regime the most difficult for which
to provide accurate predictions for use in cosmological analyses. I have
used N-body simulations to predict structure formation, combined with a
halo occupation distribution model to populate halos with galaxies,
which provides accurate predictions in this regime for both galaxy
clustering and galaxy-galaxy lensing across a wide range of galaxy and
cosmological parameter space by using Gaussian process interpolation.
In order to approach data, I have examined selection effects in the
CMASS galaxy sample and suggest possible modifications to the standard
halo model in order to incorporate such effects. I conclude with future
directions and prospects for cosmological inference on small scales with
DES and DESI.
Dec 7, Friday
12:00 pm (INPA)
Carlos Garcia Garcia, Madrid
LBL 50-5132
Theoretical priors for quintessence
Dark energy is a key unsolved problem. An enormous number of theories try to
explain the accelerated expansion of the universe, ranging from the simplicity
of a cosmological constant to the inclusion of new gravitational fields that
affect space-time dynamics. We need clever methods to test the landscape of
theories to make the most of next-generation experiments. I will present a
novel framework to study dark energy and apply it to general quintessence
models, reducing their functional freedom. Expanding the dark energy density as
a truncated polynomial series, we are able to reproduce the observables with
less than 1% error, with just 2 parameters. This economic yet precise
description will allow dark energy to be constrained with next generation instruments
in a general and efficient way.
Dec 14, Friday
12:00 pm (INPA)
Kimmy Wu, Chicago
LBL 50-5132
Delensing, Neural Networks, the H_0 problem — a perspective from the CMB
Dec. 20, Thursday
4 pm (RPM)
Sunny Vagnozzi, Stockholm
LBL 50-5132
“Recent Developments in Neutrino Cosmology”
A robust detection of neutrino masses is avowedly among the key goals of several upcoming Cosmic Microwave Background (CMB) and Large-Scale Structure (LSS) surveys. In this talk, I will describe recent progress in neutrino cosmology on three fronts. Firstly, I will illustrate the wealth of information on the sum of the neutrino masses obtainable from current cosmological probes, focusing on LSS data. Current upper limits begin favoring the normal neutrino mass ordering, emphasizing the need to develop statistical tools for quantifying this preference. Next, I will discuss galaxy bias as a limitation towards fully capitalizing on neutrino information hidden in LSS data, proposing a method for calibrating the scale-dependent galaxy bias using CMB lensing-galaxy cross-correlations. Moreover, in massive neutrino cosmologies the bias as usually defined is scale-dependent even on large scales: neglecting this effect will lead to incorrectly inferred parameters. Finally, I will take on a different angle and discuss degeneracies between neutrinos and other cosmological parameters. I will show how in certain physically motivated dynamical dark energy models the neutrino mass upper limits tighten instead of broadening, discussing implications for future laboratory determinations of the mass ordering. I will also discuss how neutrino unknowns affect constraints on inflationary models.
Jan 2019
Jan 25, Friday
12:00 pm (INPA)
Anna Ijjas, Harvard
LBL 50-5132
Jan 29, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Paco Villaescusa Navarro, CCA
Campbell 131
February 2019
Feb 1, Friday
12 pm (INPA)
Hillary Child, Chicago
LBL 50-5132
Feb 5, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Lloyd Knox, Davis
Campbell 131
Feb 12, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Eric Baxter, Penn
Campbell 131
Feb 19, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Feb 22, Friday
12 pm (INPA)
Jia Liu, Princeton
LBL 50-5132
Feb 26, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Stephen Feeney, CCA
Campbell 131
March 2019
Mar 5, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Camille Avestruz, Chicago
Campbell 131
Mar 12, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Andrej Obuljen, Perimeter
Campbell 131
Mar 19, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Mar 26, Tuesday
Spring break, no seminar
April 2019
Apr 2, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Apr 9, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Liang Dai, IAS
Campbell 131
Apr 16, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Apr 23, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Apr 30, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Miguel Aragon-Calvo, UNAM
Campbell 131
Past Months
July 2018
July 19, Thursday
4 pm (RPM)
Laura Newburgh, Yale
LBL 50-5132
“New Probes of Old Structure: Cosmology with 21cm Intensity Mapping and the Cosmic Microwave Background”
Current cosmological measurements have left us with deep questions about our Universe: What caused the expansion of the Universe at the earliest times? How did structure form? What is Dark Energy and does it evolve with time? New experiments like CHIME, HIRAX, and ACTPol are poised to address these questions through 3-dimensional maps of structure and measurements of the polarized Cosmic Microwave Background. In this talk, I will describe how we will use 21cm intensity measurements from CHIME and HIRAX to place sensitive constraints on Dark Energy between redshifts 0.8 — 2.5, a poorly probed era corresponding to when Dark Energy began to impact the expansion history of the Universe. I will also discuss how we will use data from new instruments on the ACT telescope to constrain cosmological parameters like the total neutrino mass and probe structure at late times.
Aug 2018
Aug 9, Thursday
2:00 pm (LBL special seminar)
John Walsh, Unlearn.AI, Inc.
LBL 50-5026
Machine Learning
I will give a broad overview of machine learning, defining the main areas of supervised, unsupervised, and reinforcement learning, including a brief history and a view on the current state of the field. I will discuss what aspects of machine learning are interesting to me as a physicist and data scientist, highlighting some open problems in supervised learning before delving into unsupervised learning. Physics-inspired machine learning models have the potential to make a tremendous impact on unsupervised learning, and I will present some of my recent work on a class of models known as Boltzmann machines.
Aug 28, Tuesday
1:10 pm (Cosmology/ BCCP)
Matt Lewandowski, IPhT (Saclay)
Campbell 131
Analytic IR-resummation for the BAO peak
We develop an analytic method for implementing the IR-resummation of arXiv:1404.5954, which allows one to correctly and consistently describe the imprint of baryon acoustic oscillations (BAO) on statistical observables in large-scale structure. We show that the final IR-resummed correlation function can be computed analytically without relying on numerical integration, thus allowing for an efficient and accurate use of these predictions on real data in cosmological parameter fitting. In this work we focus on the one-loop correlation function, where the challenge is to reproduce the BAO peak. We show that, compared with the standard numerical integration method of IR-resummation, the new method is accurate to better than 0.2% of the numerical result, and is quite easily improvable. We also give an approximate resummation scheme which is based on using the linear displacements of a fixed fiducial cosmology, which when combined with the method described above, is about six times faster than the standard numerical integration. Finally, we show that this analytic method is generalizable to higher loop computations.
Sep 2018
Sep 4, Tuesday
1:10 pm (Cosmology/ BCCP)
Joe DeRose, Stanford
Campbell 131
Modeling the Non-Linear Universe using Cosmological Simulations
Current and upcoming cosmological surveys are turning their eyes towards the late time universe in an attempt to answer fundamental questions about the structure of our world and the physical laws that govern it. Given the non-linear nature of the observables studied by these surveys, perturbative modeling approaches have limited faculty. Thus, non-perturbative simulation approaches are necessary in order to connect theory with observation and wring the most information out of the deluge of data that will be collected in the coming years. In this talk I will discuss a number of efforts I am involved with to interpret data using simulations within the Dark Energy Survey, focusing on their use in our Y1 analyses, and how we are extending this for upcoming analyses. I will also discuss more forward looking efforts to build emulators for galaxy survey observables that promise to unlock the information contained in highly non-linear observables, such as the halo mass function, redshift-space distortions and small scale projected clustering and galaxy-galaxy lensing.
Sep 11, Tuesday
1:10 pm (Cosmology/ BCCP)
Yi-Kuan Chiang, JHU
Campbell 131
Broadband Intensity Tomography: Spectral Tagging the Cosmic UV Background
Most of the photons ever recorded by astronomers are in the form of images—or broadband intensity mapping, while the photons’ redshift and frequency information has largely lost. I will introduce a data-driven technique to recover these otherwise collapsed dimensionalities by exhausting information in the spatial fluctuations. As the first application, we measure the spectrum of the ultraviolet background (UVB) at 0 < z < 2 in GALEX Imaging surveys. We spatially cross-correlate photons in the FUV and NUV bands with spectroscopic objects in SDSS as a function of redshift, and use the observed shapes of the K-corrections to constrain the rest-frame spectrum of the UVB. We fit simultaneously a parameterized UVB volume emissivity and a clustering bias factor both as function of frequency and cosmic time. With minimum assumptions, our measured non-ionizing continuum emissivity is broadly consistent with that in the Haardt & Madau model. Cosmic Lya emission is tentatively detected at the 2sigma level at z=1, with the luminosity density consistent with being powered by cosmic star-formation with an effective Lya escape fraction of 10%. Our approach probes all sources of radiation without surface brightness thresholding, which includes potential IGM emission. The technique brings some of the future line-intensity mapping science to within the reach of existing/upcoming broadband data at all wavelengths.
Sep 13, Thursday
4:10 pm (Astronomy Colloquium)
George Becker, Riverside
1 LeConte Hall
Connecting Galaxies and the Intergalactic Medium Near Reionization
The reionization of hydrogen was a landmark event in cosmic history. Within one billion years of the Big Bang the first galaxies emitted enough ultraviolet photons to ionize the gas in deep space, permanently transforming the Universe. Determining exactly when and how reionization occurred is therefore central to our efforts to understand these early sources, as well as the physics that governs the interaction between galaxies and their environments. I will describe what we know about reionization from the study of quasar absorption lines and other probes of the high-redshift Universe. I will especially focus on what we’re learning about the intergalactic medium (IGM) shorty after reionization is believed to end. By combining observations of high-redshift quasars with wide-field galaxy surveys we are beginning to better appreciate the complexity of the IGM at this epoch, and recognize how it may help us to construct a more complete model of reionization.
Sep 18, Tuesday
1:10 pm (Cosmology/ BCCP)
Johannes Lange , Yale
Campbell 131
New Insights into Cosmology and the Galaxy-Halo Connection from non-linear Scales
Current and future galaxy surveys have the potential to transform our understanding of galaxy formation. Information on small, non-linear scales holds the greatest statistical power and potential insight into the galaxy-halo connection. In this talk, I will concentrate on constraints derived from galaxy clustering, galaxy-galaxy lensing and satellite kinematics, showing that there is consistent tension between those different probes. I will present new measurements of the tension between clustering and lensing in the Baryonic Acoustic Oscillations Survey (BOSS). The most promising explanations for this tension, baryonic feedback, assembly bias and cosmological parameters different from the Planck CMB constraints, are discussed. Furthermore, I will show that previous tension between satellite kinematics and other probes can be attributed to systematics in the modeling. We developed an updated, more robust analysis to extract constraints on the galaxy-halo connection from satellite kinematics. The accuracy of this approach is tested using a large number of realistic mock catalogs and shown to yield unbiased, highly competitive constraints. We then apply this updated analysis to the Sloan Digital Sky Survey (SDSS) and compare our inferences from satellite kinematics to those from previous studies.
Sep 20, Thursday
4:00 pm (LBL-RPM)
Martin White, UCB/LBL
LBL 50-5132
The cosmological legacy of Planck
The anisotropies in the cosmic microwave background radiation have
become our most important cosmological fossil. The study of these
"echoes of gravity" has revolutionized cosmology, stringently tested our models and allowed precise measurement of a host of important
cosmological parameters. I will discuss how far we've come since the
early detections of CMB anisotropies and in particular the cosmological legacy of the Planck mission.
Sep 24, Monday
12:10 pm (TAC)
Vadim Semenov, Chicago
Campbell 131
"How Galaxies Form Stars"
Observed star-forming galaxies convert their gas into stars
inefficiently. The typical time on which available gas in galaxies is
depleted is ~5-10 Gyr which is orders of magnitude longer than any
physical timescale relevant for star formation. The origin of long
depletion times is a long-standing puzzle. Many galaxy simulations can
reproduce observed long depletion times but the physical mechanism
controlling their values is not understood. In addition, some of the
simulations show rather counter-intuitive behavior: global depletion
times appear to be almost insensitive to the assumptions about local
star formation in individual star-forming regions, a phenomenon
described as "self-regulation." Yet another part of the puzzle is the
observed tight and near-linear correlation between star formation rates
and the amount of molecular gas on kiloparsec and larger scales. A
linear correlation implies that depletion time of molecular gas is
almost independent of molecular gas density on >kiloparsec scales, while
popular models of star formation in galaxies predict a strong
dependence. I will present results from a suite of isolated disk galaxy
simulations in which we systematically explored the behavior of
depletion times. Using insights from these simulations we formulated a
physical model that explains both the origin of long gas depletion times
in observed galaxies and the results of galaxy formation simulations.
This model also sheds light on the origin of the observed linear
correlation between star formation rates and molecular gas.
Sep 25, Tuesday
1:10 pm (Cosmology/ BCCP)
Christiane Lorenz, Oxford
Campbell 131
'Neutrino cosmology and large scale structure'
In this talk, I will present studies of the model-dependence of cosmological neutrino mass constraints. In particular, I will focus on two phenomenological parameterizations of time-varying dark energy (early dark energy and barotropic dark energy) that can exhibit degeneracies with the cosmic neutrino background over extended periods of cosmic time, and show how the combination of multiple probes across cosmic time can help to distinguish between the two components. In addition, I will discuss how neutrino mass constraints can change in extended neutrino mass models, and how current tensions between low- and high-redshift cosmological data might be affected in these models. Finally, I will discuss whether lensing magnification and other relativistic effects that affect the galaxy distribution contain additional information about dark energy and neutrino parameters, and how much parameter constraints can be biased from neglecting them.
October 2018
Oct 2, Tuesday
1:10 pm (Cosmology/ BCCP)
Kris Pardo, Princeton
Campbell 131
Astrophysical Tests of Gravitation and Dark Matter
Dark energy and dark matter make up most of the energy content of the Universe; yet, we still know very little about either of them. During my talk, I will highlight two ways of using astrophysical data to constrain theories of dark energy and dark matter. First, I will discuss how we can use the recent multi-messenger gravitational wave event GW170817 to test higher-dimensional theories of gravity, which are often invoked to explain dark energy. I will show how we can use just this one event to constrain the number of large extra dimensions to 4, within a few percent. Then, I will discuss how we can use warps in galactic disks to constrain self-interacting dark matter. Our preliminary results show that we can obtain constraints on the self-interaction cross section that are competitive with the Bullet Cluster constraints. As we continue to collect ever larger amounts of data on astrophysical systems, it is important to think through unique applications of this data to fundamental physics.
Oct 4, Thursday
4 pm (LBL RPM)
Kyle Helson, Goddard
LBL 50-5132
"The Status and Outlook of Cosmic Microwave Background Polarization Measurement"
The detailed characterization of the intensity and polarization Cosmic Microwave Background (CMB) radiation provides a powerful tool to constrain the properties of early Universe. The polarization induced by a stochastic gravitational wave background in this epoch induces a distinctive and measurable signature of primordial inflationary processes. Presently the field has a healthy contingent of ground, balloon, and space based missions. In this talk I will discuss a brief history of CMB measurements, the current status of the field and outlook for the future. I will also discuss on-going work at NASA Goddard Space Flight Center on the development of polarization-sensitive detectors for a future satellite CMB polarization mission and their application in ground based instruments.
Oct 5, Friday
12:00 pm (INPA)
Arka Banerjee, Stanford
LBL 50-5132
Signatures of massive neutrinos on Large Scale Structure
Neutrino oscillation experiments have shown that there are at least two massive neutrino eigenstates, in a mass range that can produce observable signatures in current and future cosmological surveys. I will talk about the challenges and progress in correctly including the effects of massive neutrinos in N-body simulations of structure formation. Finally, I will talk about how these simulations can be used to study novel effects in massive neutrino cosmology - in particular, scale-dependent bias of nonlinear objects such as halos and voids on large scales.
Oct 9, Tuesday
1:10 pm (Cosmology/ BCCP)
Hung-Jin Huang, CMU
Campbell 131
Astrophysical systematics in weak lensing
Current and upcoming large galaxy surveys will provide a wealth of cosmological information about our Universe. Weak lensing is one of the primary probes to uncover the nature of dark energy through its sensitivity to the growth of structure across cosmic time. However, a detailed understanding about systematic effects is required to make robust cosmological inferences from data. Besides shear calibration and photometric redshift systematics, intrinsic alignment and the effects of baryons on the observables constitute two major astrophysical systematics for weak lensing. In this talk, I will focus on discussing these two sources of astrophysical systematics. I will first introduce two methods to mitigate the effects of baryons for cosmic shear, and validate their performances for a future LSST-like survey under a variety of baryonic scenarios constructed from hydrodynamical simulations. In the second part of my talk, I will discuss the importance of intrinsic alignment, and present our measurements of intrinsic alignments of central and satellite galaxies in redMaPPer clusters.
Oct 16, Tuesday
1:10 pm (Cosmology/ BCCP)
Gigi Guzzo, Milan
Campbell 131
“Measuring the Universe with Galaxy Clustering and Motions”
I will review some recent advances in studies of large-scale structure, focusing on the latest
clustering results from our VIPERS project at the ESO VLT, together with parallel development
of analysis and modeling techniques. With 90,000 galaxy redshifts and multi-band photometric
information, VIPERS has delivered at redshift 0.5 < z < 1.2 both cosmological measurements
(as e.g. various estimates of the growth rate of structure through Redshift Space Distortions)
and new insights on the evolution of galaxies within their large-scale environment at these
epochs. Having the latter aspects under control allows us to investigate how well specific
galaxies trace the underlying mass distribution and to reduce systematic errors in the modeling,
e.g. when using RSD to test gravity. I will then conclude discussing a recent complementary
result, in which forward modeling of redshift-space clustering through numerical simulations,
pushing well into the non-linear regime, provides a sensitive and seemingly robust probe of
General Relativity.
Oct 23, Tuesday
1:10 pm (Cosmology/ BCCP)
Fred Davies, UCSB
Campbell 131
Measuring the Reionization History with Quasar Damping Wings
The Lyman-alpha damping wing from neutral hydrogen in the intergalactic medium is predicted to be a key signature of the reionization epoch in the spectra of high-redshift quasars. There are substantial challenges in measuring and interpreting this signal, however: the intrinsic (i.e. unabsorbed) spectrum of the quasar near its Lyman-alpha line is unknown, and the impact of neutral gas on the spectrum depends on the large-scale structure of reionization, the small-scale distribution of baryons, and the age of the quasar. We have developed a Principal Component Analysis-based machine-learning approach to predict the intrinsic quasar spectrum, and have combined semi-numerical simulations of reionization with 1D radiative transfer through hydrodynamical simulations to predict the full range of proximity zone and damping wing morphologies. Using a Bayesian statistical formalism we can then translate an observed quasar spectrum into joint constraints on the neutral fraction and the quasar lifetime. I will demonstrate the application of these methods to the highest redshift quasars known, and discuss the potential for existing quasar spectra to constrain the reionization history between redshift 6 and 7 and the growth of the first supermassive black holes.
Oct 24, Wednesday
6-7:30 pm (Distinguished Lecture in Astronomy)
Bob Kirshner, Harvard
Sibley Auditorium, Bechtel Engineering Center
From the Accelerating Universe to Accelerating Science
Twenty years ago, astronomers were astonished to learn from observations of exploding stars that cosmic expansion is speeding up. We attribute this to a mysterious “dark energy” that pervades the universe and makes up 70% of it. Scientists are working in many ways to learn more about the nature of dark energy, but our reservoir of ignorance is deep. This talk will summarize the present state of knowledge and look ahead to new ways to use infrared observations of supernovae to improve our grip on dark energy. Accelerating scientific discovery is a mission of the Gordon and Betty Moore Foundation and Professor Kirshner will illustrate some of the ways we do that at Berkeley and beyond.
Oct 25, Thursday
4pm (LBL RPM)
Laura Newburgh, Yale
LBL 50-5132
Forefronts of Cosmology: Beyond Standard Model Physics with Upcoming CMB Instruments
Upcoming new instruments to measure the polarized CMB promise to provide discriminatory limits on inflation, the number of light relic particles, and the sum of the neutrino masses, ushering in a new era of using the CMB as a probe of particle physics. Achieving these science goals requires highly sensitive instruments that are composed of enormous arrays of low noise detectors. In addition, systematic errors and foreground removal must be improved to lower the systematics floor below the statistical errors, necessitating dramatic improvements in calibration precision. In this talk, I will describe Simons Observatory (coming online in ~2021) and CMB-S4 (coming online in ~2027), their science goals, and how the twin requirements for sensitivity and systematics require a new approach to software for readout, data acquisition, and control systems. I will also discuss future directions for cosmology, including work in 21cm instruments to improve our understanding of the nature of Dark Energy.
Oct 25, Thursday
4:10 pm (Astronomy Colloquium)
Bob Kirshner, Harvard
1 LeConte Hall
Using HST: looking inside SN 1987A and making better measures of cosmic expansion
The Hubble Space Telescope has been the perfect tool for studying SN 1987A, providing a way to dissect the expansion geometry and infer the 3D distribution of the stellar debris that was ejected 31 years ago. Together with radio and x-ray observations, the HST data provide stringent limits on emission from the neutron star that was formed in the collapse. HST is also a powerful tool for making rest-frame infrared observations of distant supernovae to measure cosmic expansion history. The obvious advantage is that infrared observations suffer less from extinction in the supernova host, but nature has been very kind and exploding white dwarfs are better standard candles in the near infrared. I will show new results from our analysis of the world’s low-redshift sample and give a report on our RAISIN program to use SNIA in the IR to constrain dark energy properties.
Oct 30, Tuesday
1:10 pm (Cosmology/ BCCP)
Alex Malz, NYU
Campbell 131
Maximizing LSST science with probabilistic data products
LSST will produce massive catalogs including detected objects down to unprecedented floors in signal-to-noise ratio, opening the door to a new space of potential discoveries, from illuminating the dark energy accelerating the expansion of the universe to revealing the physical processes underlying transients and variable stars. The anticipated deluge of uncertainty-dominated data, however, demands an unprecedented degree of statistical rigor. Posterior probabilities that quantify complex uncertainties are appropriate successors to the conventional point estimates of physical parameters that suffice for more informative data. In contrast with traditional science analysis pipelines for point estimates and Gaussian errors, inferential infrastructure compatible with probabilistic data products remains underdeveloped. I present mathematically self-consistent techniques for validating, storing, and using such probabilities in the contexts of the ongoing PLAsTiCC light curve classification competition and photometric redshifts with applications in cosmology. A statistically principled propagation of information will enable us to use every part of the animal and do the best science possible with LSST.
November 2018
Nov 1, Thursday
4pm (LBL RPM)
Vivian Poulin, JHU
LBL 50-5132
“Shedding Light on Dark Matter with the CMB: Implications for EDGES 21cm Signal”
In this talk, I would like to review how the Cosmic Microwave Background (in particular its temperature and polarization anisotropies) can be used to perform both direct and indirect detection of Dark Matter. In a first part, I will show how the CMB can be used to put stringent constraints on DM models leading to energy injection, such as DM annihilations and decay or Primordial Black Holes matter accretion. I will compare CMB bounds to those coming from galactic cosmic and gamma ray searches and illustrate how the 21cm signal, a major target of next generation surveys, can be used to significantly improve over current limits. In a second part, I will discuss a recent puzzling absorption feature in the global 21cm signal observed by the EDGES experiment. Such feature could be explained by efficient scattering between DM and baryons in the late universe. I will show how the CMB strongly challenges such explanation of that signal.
Nov 2, Friday
12:00 pm (INPA)
Marco Raveri, Chicago
LBL 50-5132
Concordance cosmology and beyond
I will discuss the relevance of statistical tools aimed at testing the concordance of cosmological observations and how they can be used as the most model independent test for the standard cosmological model. These statistical tests are a powerful way of assessing whether a specific model explains the observed measurements or whether there are indications of residual, unaccounted, systematic effects or hints toward new physical phenomena. I will show the results of these tests when applied to state of the art cosmological measurements and discuss the overall level of agreement of the standard cosmological model with observations. I will then move to discuss extensions of the standard model and how these might help in explaining some of the discrepant aspects of observations. In particular I will focus on dark energy and modified gravity models discussing their cosmological phenomenology and highlighting the observational aspects that they improve and that can be used to test them.
Nov 5, Monday
12:10 pm (TAC)
Ue-Li Pen, CITA
Campbell 131
Neutrino Torquing
We show how galaxy spin alignment directly measures the initial tidal field, thus doubling the information content of galaxy surveys. The initial density field can be recovered to $k\sim 1$ using spin alignment and non-linear galaxy displacement reconstruction, separately. By measuring two scalar fields, we measure the gravitational field of cold dark matter and neutrinos. A next generation 21cm survey can achieve the required sensitivity to measure the contribution of primordial neutrinos to galaxy spin.
Nov 6, Tuesday
1:10 pm (Cosmology/ BCCP)
Denise Schmitz, Caltech
Campbell 131
A Perturbative Treatment of Intrinsic Alignments
Intrinsic alignments (IA), correlations between the intrinsic shapes and orientations of galaxies on the sky, are both a significant systematic in weak lensing and a probe of the effect of large-scale structure on galactic structure and angular momentum. In the era of precision cosmology, it is thus especially important to model IA with high accuracy. Previous models for IA fall broadly into two categories: linear alignment models, which are linear in the matter density field, and tidal torquing models, which are quadratic in the matter density. More generally, these contributions can be considered part of an effective expansion in all potentially relevant cosmological fields at a given order. I have developed a full, self-consistent formalism up to third order in standard perturbation theory (SPT) for such a perturbative treatment of the IA field, analogous to the use of bias coefficients to quantify clustering. I will present this model and discuss the implications for weak lensing systematics as well as for studies of galaxy formation and evolution. This work includes a complete treatment of time-evolution effects, which is not only required for a self-consistent SPT expansion, but also allows for the use of IA to probe properties of galaxy formation.
Nov 9, Friday
12:00 pm (INPA)
Doug Finkbeiner, Harvard
LBL 50-5132
Making neural net classifiers more robust and explainable: Lessons from Adversarial AI
As deep neural nets achieve ever greater successes, efforts to break them and learn about their failure modes are also ramping up. Security experts and malicious actors are interested in weaknesses per se, and we scientists are more interested in what we can learn about robustness to inputs somewhat different from training data. I will give examples of attacks and defenses, and talk about a measure of credibility at inference time.
Nov 13, Tuesday
1:10 pm (Cosmology/ BCCP)
Yuuki Omori, Stanford
Campbell 131
Survey synergy: combining CMB and galaxy surveys
Large scale structure (LSS) alters the appearance of both background galaxies and the cosmic microwave background (CMB) through the effect of gravitational lensing. Auto and cross-correlations of these observables provide us independent information, and the various probes can be combined to constrain cosmology. In this talk, I will present the recent results from combining SPT and DES year 1 data sets. I will also discuss what we have learnt, and the improvements we are going to make in the year 3 analysis.
Nov 16, Friday
12:00 pm (INPA)
Andrej Dvornik, Leiden
LBL 50-5132
KiDS and biases
The current ongoing large imaging surveys are an excellent tool for studying the origin and evolution of the Universe and the galaxy - dark matter connection, using the weak gravitational lensing as the main probe. Using the predicting power of the halo model formalism, the weak gravitational lensing (together with other large scale probes) can be used to constrain the origin of the scale dependence of the galaxy bias - the relation between the galaxies and the dark matter distribution, as well as studying the dependence of the formation time of galaxies on their halo masses - so called assembly bias. In this talk I will present the KiDS survey and way it can be used to shine a light on the different aspects of galaxy-halo connection.
Nov 26, Monday
4:15 pm (physics colloquium)
Shirley Ho, CCA
1 LeConte
Machine Learning for the Universe: Steps towards Opening the Blackbox
To fully understand the structure formation of the Universe is one of the holy grails of modern astrophysics. Astrophysicists survey large volumes of the Universe and employ a large ensemble of computer simulations to compare with the observed data in order to extract the full information of our own Universe. However, to evolve trillions of galaxies over billions of years even with the simplest physics is a daunting task. In this talk, we discuss our recent work on building a deep neural network to predict the non-linear structure formation of the Universe from simple linear perturbation theory. Our extensive analysis, demonstrates that the deep learning model outperforms the second order perturbation theory, the commonly used fast approximate simulation method, in point-wise comparison, 2-point correlation, and 3-point correlation. We also show that the deep learning model is able to accurately extrapolate far beyond its training data, and predict structutre formation for significantly different cosmological parameters. Our study proves, for the first time, that deep learning is a practical and accurate alternative to approximate simultaions of the gravitational structure formation of the Universe. We will also discuss our efforts in understanding why the deep learning model is able to capture the non-linear structure formation of the Universe.
Nov 27, Tuesday
1:10 pm (Cosmology/ BCCP)
Natalia Porqueres, Garching
Campbell 131
Inferring high-redshift large-scale structure dynamics from the
Lyman-alpha forest
I will present a Bayesian framework to infer the
three-dimensional large-scale density field and its dynamics at
high-redshift from the Lyman-alpha forest. This framework allows the
exploration of the cosmic web at distances where galaxy surveys are too
sparse to trace the large-scale structure and provides access to the
initial conditions and dynamical properties such as the velocity field. I
tested and validated the method with the CLAMATO mock data, showing that
it can account for the baryonic bias and recover the dark matter
distribution at high-z.
Nov 30, Friday
12:00 pm (INPA)
Simon Foreman, CITA
LBL 50-5132
Gravitational lensing of line intensity maps
Gravitational lensing of the cosmic microwave background (CMB) has emerged as a powerful cosmological probe, made possible by the development and characterization of nearly-optimal estimators for extracting the lensing signal from temperature and polarization maps. One can ask whether similar tools can be applied to upcoming "intensity maps" of emission lines at various wavelengths (e.g. 21cm). In this talk, I will present recent work in this direction, focusing in particular on the impact of nonlinear gravitational clustering on standard CMB lensing estimators when applied to intensity maps. I will show how these nonlinearities can provide a significant contaminant to lensing reconstruction, but will also describe how this contamination can largely be mitigated by modifying the lensing estimator. Finally, I will present estimates for the detectability of lensing in ongoing and future intensity mapping surveys, and highlight related work on reconstructing large-angle information in galaxy surveys and CMB maps.
|
