August2023
August 29, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Group intro
131 Campbell
September 2023
Sept 5, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Bruce Allen Max Planck Institute for Gravitational Physics
131 Campbell
Detecting nano-Hertz gravitational waves
Four pulsar timing array (PTA) collaborations have recently
published the results of their latest searches for a stochastic
background of very low frequency gravitational waves. Their
conclusions span the range from "weak evidence" to "compelling
evidence". The detectors operate at a Galactic-scale, by exploiting
radio pulsars (very stable rapidly-spinning neutron stars) as high
precision clocks. I'll explain how these detectors operate -- the
gravitational waves leave detectable imprints on the arrival times of
the radio pulses -- and review the current state of the field and its
prospects. I'll also describe some recent work (arXiv:2205.05637,
arXiv:2208.07230) on properties of the "Hellings and Downs
correlation". This pattern of pulsar timing correlations is the
"smoking gun" signature that gravitational waves have modified the
pulse arrival times.
Sept 8, Friday
12 noon
Naim Göksel Karaçaylı, Ohio State University
LBL 50-5132 and ZOOM
1D Lyman-alpha forest power spectrum from the DESI early data
The one-dimensional power spectrum P1D of the Lya forest provides important information including constraints on warm dark matter models, the sum of the neutrino masses, and the thermal state of the intergalactic medium. I will present the first measurement of P1D with the quadratic maximum likelihood estimator (QMLE) applied to 54,600 quasars from DESI early data. Our analysis demonstrates the percent-level performance of the spectroscopic pipeline noise estimation and the spectrograph resolution matrix. Even though the pipeline is exceptional, noise calibration and spectrograph resolution remain as major sources of systematics.
Sept 11, Monday
4:10 pm (Physics Colloquium)
Uros Seljak, UCB
Physics North lecture Hall 1
AI for Physics, Physics for AI
Artificial Intelligence (Machine Learning) is revolutionizing many aspects of our life, but its success stories in physics and astronomy are rare and limited to a few subfields only. I will argue that this is because physics applications require development of physics specific AI methods, rather than using off the shelf methods from AI community. A few examples of physics specific nature of the data are large dimensionality of the data, stochastic nature of the data, and symmetries. I will argue that learning the data structures first using generative learning approaches such as Normalizing Flows enables not only better learning, but also provides additional information on robustness, such as anomaly detection. These methods applied to cosmology data show the promise of up to an order of magnitude improvement relative to traditional methods. Physics ideas have also influenced the development of AI, and many of these have been based on stochastic processes and sampling. I will discuss recently developed MicroCanonical Hamiltonian and Langevin Monte Carlo, which are a new class of sampling methods that outperform previous state of the art such as Hamiltonian Monte Carlo, in some cases by orders of magnitude. These new sampling methods will in turn enable solutions of physics problems that were not possible before, in a wide range of fields from cosmology to lattice QCD.
Sept 12, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
YiFei Luo, Santa Cruz
131 Campbell
The Merian Survey: Characterizing dark matter and feedback in star-forming dwarf galaxies with medium-band filters
The Merian survey is an ongoing medium-band imaging survey on the CTIO 4-meter Blanco telescope. We will use 64 nights with the Dark Energy Camera (DECam) to cover 800 deg^2 of the sky within the HSC SSP wide field, aiming to characterize dark matter halo and stellar feedback in star-forming dwarf galaxies. With two custom made medium-band filters targeting at Halpha and OIII emission lines at z~0.1, we expect to detect 80,000 star-forming dwarf galaxies with 90% completeness. The Merian survey will allow us to measure the full dark matter profile of dwarf galaxies via weak gravitational lensing for the first time. In this talk, I will present the filter design, survey description and the current status of the Merian survey. I will also discuss several other science cases of Merian, such as searching Lya emitters at z>3 with Merian filters.
Sept 19, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Jessie Muir, Perimeter
131 Campbell
Probing cosmological structure growth with the Dark Energy Survey
Within the framework of ΛCDM as a standard cosmological model, we can predict the evolution of matter density fluctuations over the history of the Universe. By measuring the large-scale distribution of matter and comparing its growth history to ΛCDM expectations, galaxy surveys can thus seek clues for how to answer some of the most significant open questions in physics, such as the nature of dark energy and dark matter. I will discuss how we approach this with the Dark Energy Survey, a 5000 square-degree imaging survey, both by constraining parameters of ΛCDM and by extending that analysis to physics beyond the standard cosmological model. I'll highlight results from DES Year 3 tests of growth history, as well as some challenges which can guide future studies and preparation for next-generation cosmological experiments.
Sept 26, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Tomomi Sunayama, Arizona
131 Campbell
Towards an accurate optical cluster cosmology with SDSS redMaPPer clusters and HSC Y3 lensing measurements
Galaxy clusters are the most massive gravitationally self-bound objects in the Universe. These clusters form at the rare high peaks of the primordial density fluctuations, and they subsequently trace the growth of structure in the Universe as they grow in mass and abundance. As such, clusters constitute a natural cosmological probe for constraining the properties of the primordial fluctuations as well as cosmological parameters including the nature of dark energy. Clusters are, however, also known to be susceptible to many systematics such as selection biases and systematics. Furthermore, the recent cosmology result from DES using photometrically-identified clusters, which favored lower Omega_m and higher sigma8 compared against other constraints from CMB and large-scale structure, question optical clusters as a cosmological probe.
We developed a novel analysis method that fully forward models the abundances, weak lensing, and clustering of galaxy clusters including an accurate modeling of systematics such as projection effects. Projection effects are the misidentification of interloper galaxies as a member of clusters and are considered the biggest systematics. We found that projection effects not only alter mass-observable relation but also boost the amplitude of clustering and lensing signals due to the anisotropic distribution of optical clusters and modeling projection effects is a key to an accurate cosmological analysis using optical clusters. In this talk, I will present the result of a joint analysis using Sloan Digital Sky Survey (SDSS) redMaPPer clusters and Hyper-Suprime Cam (HSC)-Y3 shape catalog.
Sept 26, Tuesday
4 pm LBL RPM
Jia Liu, IPMU
LBL 50-5132 and ZOOM
Neutrinos, Baryons, Cosmic Web, and COVID19
This won’t be your usual seminar, but rather a report back since I left Berkeley two years ago. Despite barely having made it to campus or the lab due to mat leave + COVID, I managed to work with many Berkeley people on various projects, almost all of which happened serendipitously. I will touch upon the topics mentioned in the title. I will also discuss prospects for Stage IV cosmology, in terms of joint simulations and analysis with CMB and LSS surveys.
October 2023
Oct. 3, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Leander Thiele, Princeton
131 Campbell
Advancing Cosmology with Robust ML
Upcoming cosmological data sets will constitute a qualitative leap in statistical power and, consequently, sensitivity to complicated systematics.
The potential scientific return could not be more exciting: from neutrino mass to dark energy to our universe's primordial matter distribution,
cosmological information could bear on fundamental physics across a wide range of energy scales.
Unlocking this potential, however, will hinge on our ability to accurately extract the richest information from the measurements.
The non-linear character of the late-time density field calls for novel analysis methods, often involving analytically intractable probability distributions.
Furthermore, great care will be required in modeling the formation and evolution of galaxies as well as the phenomenology of violent small-scale energy release.
I will demonstrate how the past decade of rapid advancements in machine learning sets the stage for performant and robust methods that will be instrumental in meeting these challenges.
Oct. 4, Tuesday
12:10 pm (DESI lunch)
Leander Thiele, Princeton
131 Campbell
Neutrino mass constraint from an Implicit Likelihood Analysis of BOSS voids
Gravity causes the Universe to evolve from Gaussian initial perturbations to the non-linear structures that constitute the cosmic web. Consequently, traditional two-point statistics are likely unable to capture the entire information content of late-time observables, such as redshift-space positions of galaxies. I will present our efforts to constrain the neutrino mass sum using galaxies measured by SDSS/BOSS. Being a low-density, diffuse component, neutrinos are expected to leave imprints on cosmic voids. Thus, we augment the usual galaxy auto-power spectrum with void statistics. Inference is based on simulations and performed through a neural implicit likelihood procedure. I will pay special attention to improvements that are potentially required for an application to DESI.
Oct. 10, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Suchetha Cooray, National Astronomical Observatory of Japan
131 Campbell
A forward modeling approach to understanding the light from galaxies
I am developing an empirical galaxy formation model that can model consistent galaxy growth histories within the context of cosmological structure formation. The observational constraints of the previous works include; stellar mass functions, quenched fractions, and average star formation rates (Behroozi et al., 2019). However, the inherent uncertainties that come with modeling choices in deriving galaxy physical properties from spectral energy distributions (SEDs) mean that we at least have a ~0.35 dex uncertainty. Therefore, I am developing the next-generation empirical galaxy formation model that directly matches all the observed SEDs across cosmic time. The self-consistent evolution of the galaxy properties should reduce the uncertainty on galaxy properties down to ~0.1 dex level. I will discuss recent progress and some scientific questions we can start to answer with the approach. I will also discuss how generative models could be used to make inferences at the image level (field level) for galaxies possible. Key outcomes include a fully physical, self-consistent picture of galaxy stellar masses, star formation histories, dust, and metallicity from z = 0 to 15; significantly reduced uncertainties on the galaxy-halos connection; and highly realistic mock catalogs and images for arbitrary current and future surveys that match the latest observations.
Oct. 11, Wednesday
12:10 pm, DESI lunch
Xinyi Chen, Yale
LBL 50-5132 and zoom
Probing primordial non-Gaussianity by reconstructing the initial conditions with machine learning
Inflation remains one of the enigmas in fundamental physics. While it is difficult to distinguish different inflation models, information contained in primordial non-Gaussianity (PNG) offers a route to break the degeneracy. In galaxy surveys, the local type PNG is usually probed by measuring the scale-dependent bias in the power spectrum on large scales, where cosmic variance is also large. I will introduce a new approach to measure the local type PNG by using the reconstructed density field, a density field reversed to the initial conditions from late time. This is then followed by computing a near optimal bispectrum estimator and fitting a new template. This new approach offers an alternative way to the existing method with different systematics and also follows organically the procedure of baryon acoustic oscillation (BAO) analysis in large galaxy surveys. By reconstructing the initial conditions, we remove the nonlinearity induced by gravity, which is a source of confusion when measuring PNG. I will present a reconstruction method using convolutional neural networks that significantly improves the performance of traditional reconstruction algorithms in the matter density field, which is crucial for more effectively probing PNG. This pipeline can enable new observational constraints on PNG from the ongoing Dark Energy Spectroscopic Instrument (DESI) and Euclid surveys, as well as from upcoming surveys, such as that of the Nancy Grace Roman Space Telescope.
Oct. 17, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Nikhil Padmanabhan, Yale
131 Campbell
Baryon Acoustic Oscillations with Galaxy Surveys : Present State and Some Future Prospects
I will weave three separate threads. The first will be to describe recent and ongoing results from the Dark Energy Spectroscopic Survey. I will present the BAO results from the early DESI data, and some of the preparatory work for the Year 1 data. I will then present a new approach to reconstruction based on optimal transport, highlighting some recent results and ideas for future applications. I will end by discussing merging perturbative reconstruction ideas with convolutional neural networks, and a possible new application to primordial non-gaussianity.
Oct. 24, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Alejandro Aviles, UNAM
131 Campbell
Large scale structure formation in theories with scale dependent linear growth
In this talk, I will present recent developments in perturbation theory regarding the study of the large-scale clustering of galaxies in theories with scale-dependent linear growth, such as those involving massive neutrinos and modified gravity. The challenge in these models is that the introduced scale is incorporated into the perturbative kernels. We utilize the Lagrangian approach to compute these kernels and then map them to the Eulerian frame to obtain the power spectrum. Unfortunately, our method for computing statistics is quite slow. Therefore, we simplify the modified kernels by retaining only the terms that contain the scale-dependent growth factor. This approach enables us to estimate parameters from BOSS data using standard MCMC methods. In the final part of the presentation, I will delve into how the full-shape approach extracts the neutrino mass information from the power spectrum.
Oct. 31, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Molly Wolfson, UCSB
131 Campbell
Constraining reionization with the z > 5 Lyman-α forest
One of the most important open problems in cosmology is understanding the reionization of the intergalactic medium (IGM) by the first luminous sources. During reionization, ionization fronts propagate through the IGM, heating the reionized gas. This heat injection can be observed over a redshift interval of ∆z ~ 1 due to the long cooling times in the low-density IGM. Simultaneously, the mean free path of ionizing photons (λmfp) describing the ultraviolet background (UVB) rapidly evolves as bubbles of reionized gas, where the UVB is stronger, merge. Thus, constraining the thermal state of the IGM and the evolution of λmfp can, in turn, be used to constrain reionization. Transmission in the Lyman-a (Lya) forest, the ubiquitous Lya absorption lines produced by residual neutral hydrogen in the IGM along quasar sightlines, offers a powerful tool to investigate these phenomena. First, I will discuss my work demonstrating that the Lya forest flux auto-correlation function can be used to constrain λmfp and the thermal state of the IGM. I will also introduce my measurements of the flux auto-correlation function at z > 5 from the XQR-30 extended data set, which are the first measurements of Lya forest clustering extending to z=6. Finally, I will discuss the constraints on λmfp and the progress towards achieving reionization constraints based on these measurements.
November 2023
Nov 7, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
Yeshukei Jagvaral, CMU
131 Campbell
Galaxies on graph neural networks: towards robust synthetic galaxy
catalogs with deep generative models
Future cosmological weak lensing surveys are expected to constrain
cosmological parameters with unprecedented precision. Systematics and
analysis pipelines of these surveys will require large volume high
resolution simulations with accurate galaxy properties. However, these
types of simulations are intractable given its computational cost. In
this work we demonstrate an alternative approach based on Graph Neural
Networks to emulate galaxy properties and generate synthetic galaxy
catalogs. We developed a novel Deep Generative model that is
statistically consistent with those of a reference high-resolution
hydrodynamical simulation. Additionally, as an extension we are
developing a novel Score-Based Diffusion Model specifically for the
SO(3) manifold to model the full 3D orientation of galaxies along with
scalar properties.
Nov 14, Tuesday
4:00 pm (BCCP/Cosmology Seminar and LBL RPM)
Aleksandra Kusiak, Columbia
LBL 50-5132 and ZOOM
Probing the Ionized Gas Thermodynamics in Distant Galaxies with
the Sunyaev-Zel’dovich Effect
The Sunyaev-Zel’dovich Effect—the Doppler boost of low-energy Cosmic
Microwave Background photons scattering off free electrons—is an
excellent probe of ionized gas residing in distant galaxies. Its two
main constituents are the kinematic SZ effect (kSZ), where electrons
have a non-zero line-of-sight (LOS) velocity and which probes the
electron line-of-sight momentum, and the thermal SZ effect (tSZ),
where electrons have high energies due to their temperature, and which
probes the electron integrated pressure. These two effects provide
complementary information to constrain the thermodynamic profile of
gas residing in distant galaxies, which can be further used to
understand feedback processes, a necessary ingredient to describe the
evolution of the large-scale structure in our Universe. Both tSZ and
kSZ can be measured in cross-correlation with large-scale structure.
In this talk, I will discuss my past and ongoing measurements of the
SZ-galaxy cross-correlation with unWISE galaxies. unWISE is a galaxy
catalog containing over 500 million galaxies on the full sky and
consists of three subsamples of mean redshifts z=0.5, 1.1, 1.5, whose
halo occupation distribution I have already constrained. If time
permits, I will also present my ongoing work on mitigating foregrounds
in the SZ cross-correlations, particularly the Cosmic Infrared
Background (CIB).
Nov 21, Tuesday
no 1 pm seminar
Nov 21, Tuesday
4:00 pm (LBL RPM)
Roohi Dalal, Princeton
LBL 50-5132 and ZOOM
New Cosmology Constraints from the Hyper Suprime-Cam Year 3 Data Release
The Hyper Suprime-Cam (HSC) survey is the deepest present-day weak lensing experiment, going to ~26 mag with exquisite image quality. This combination of depth and image quality achieved by HSC allows us to probe cosmology using weak gravitational lensing up to high redshifts, and these analyses serve as important preparatory studies for the next generation of weak lensing surveys, particularly the Vera C. Rubin Observatory. I will discuss our constraints on cosmological parameters using weak lensing cosmic shear power spectra measured from the Year 3 shear catalog of HSC, covering 416 square degrees of the northern sky. I will describe our cosmological analysis, including the steps we take to prevent confirmation bias as well as our modeling of various systematic effects. The quantity best constrained by our analysis is the parameter S8, which describes the clumpiness of the matter distribution in the universe. From our analysis, and other weak lensing analyses with different surveys, there is a 2-3 sigma detection of a tension in the S8 measurements from weak lensing and those from the cosmic microwave background. While this could suggest that our cosmological model is incomplete, I will discuss ongoing and future work that can further shed light on this tension, including an improved modeling of baryonic feedback at small scales.
Nov 28, Tuesday
1:10 pm (BCCP/Cosmology Seminar)
no talk (two next week)
131 Campbell
December 2023
Dec. 7, Thursday
4 pm (LBL RPM)
Gerrit Farren, Cambridge
LBL 50-5132 and ZOOM
The Atacama Cosmology Telescope: Probing the large scale structure with ACT DR6 CMB lensing and cross-correlation with unWISE
I will present work on probing the large scale structure of the universe using CMB lensing from the upcoming Data Release 6 of the Atacama Cosmology Telescope (ACT) and cross-correlations with galaxies from the unWISE galaxy catalog. My talk will focus on how our highly competitive constraints from CMB lensing cross-correlations can provide insight into the widely discussed “S8/sigma8 tension”. For this purpose I will briefly introduce the high fidelity CMB lensing reconstruction obtained by the ACT Collaboration and results from the analysis of the lensing auto-correlation. I will discuss results from the cross-correlation between ACT CMB lensing and unWISE galaxies, highlighting improvements to the analysis pipeline compared to previous work on the cross-correlation between Planck CMB lensing and unWISE by some of my collaborators (Krolewski et al. 2021). I will also touch on our recent work to extend such cross-correlation analyses beyond two-point correlations using the first detection of the galaxy-galaxy-CMB lensing bispectrum.
Dec. 8, Friday
12 pm (INPA seminar)
Richard Feder-Staehle, JPL/Caltech
LBL 50-5132 and ZOOM
Uncovering the Near-Infrared Universe through Galaxy Surveys and Intensity Mapping
In this talk I will discuss two complementary methods for measuring large-scale structure, namely through galaxy surveys and broad-band intensity mapping. In the first half will present updated galaxy simulations and redshift forecasts for SPHEREx, NASA's next MIDEX mission which will conduct the first all-sky spectrophotometric survey in the near-infrared (0.75 - 5 um, R~30-140). In the second half I will present preliminary results from a fluctuation analysis of imager data taken at 1.1 and 1.8 um during the final flight of the Cosmic Infrared Background ExpeRiment (CIBER), a sounding rocket payload designed to characterize and decompose the extragalactic background light.
January 2024
Jan 16, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Jan 23, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Jan 30, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
February 2024
Feb 6, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Feb 13, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Feb 20, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Feb 27, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
March 2024
Mar 5, Tuesday
1:10 pm (BCCP/Cosmology seminar)
,
Campbell 131
Mar 12, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Sam Goldstein, Columbia
Campbell 131
|
