August 2025
Aug 18, Monday
11 am (CS/DD Seminar)
Ben Horowitz
, University of Tokyo
LBL 50A-5132
Novel Forward Models for Field-Level Inference in Cosmology and Astrophysics
Disentangling complex astrophysical processes—such as baryonic feedback—from the underlying cosmological signal is critical as observations probe ever-smaller scales. Yet, no existing simulation framework has self-consistently reproduced more than a handful of key cosmological observables.
In this talk, I will present recent advances in differentiable hydrodynamical simulations that enable rapid exploration of high-dimensional parameter spaces. These forward models allow in situ training of embedded neural networks, which can either constrain flexible subgrid physics models or accelerate the underlying simulation itself. I will also discuss complementary stochastic interpolant based approaches, that provide fast and accurate alternatives for forward modeling when trained on existing simulations.
September 2025
Sep 2, Tuesday
1:10 pm (BCCP/Cosmology seminar)
First meeting
, one minute slides
Campbell 131
Introductory one minute slides
Sep 9, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Emmanuel Schaan, Stanford
Campbell 131
Line intensity mapping: enabling cross-correlations to increase confidence in cosmological detections
Line intensity mapping (LIM) is a tantalizing approach for mapping the matter density fluctuations at high-redshifts where galaxy surveys would become prohibitively expensive. By avoiding the need to resolve individual objects, LIM promises fast mapping speeds, but the faint signals and large foregrounds make it difficult to detect or trust LIM auto-spectra. Cross-correlations thus appear crucial to make the most of future LIM experiments. At the highest redshifts where galaxy surveys are too sparse, projected 2D tracers like CMB lensing and the CIB may be the only candidates for LIM cross-correlations. However, foreground filtering in LIM would naively destroy the Fourier mode overlap with these projected fields. In this talk, I will present new approaches to circumvent this issue by relying either on the non-Gaussianity of the matter density field, or the overall redshift-evolution of LIM signals.
Sep 11, Thursday
12 noon (BIDMaP seminar)
Rana Adhikari, Caltech
373 Cory Hall
Machines Learning to Make Better Experiments
Today’s most sensitive physics experiments require optimization of experimental design, nonlinear feedback controls, and materials design, among other challenges. In this talk, Professor Rana Adhikari will describe recent work in these areas in the context of gravitational wave detectors, with the goal of improving their precision for measuring the vibrations of black holes and exploring gravitation and cosmology. He will also discuss how the simultaneous use of highly entangled quantum sensing and thin films without two-level tunneling systems could open up the dark side of the universe to the deepest exploration.
Sep 16, Tuesday
1:10 pm (BCCP/Cosmology seminar)
John Franklin Crenshaw, Stanford/SLAC
Campbell 131
First Light and Future Frontiers: Commissioning the Rubin Observatory and Prospects for Cosmology with Lyman-break Galaxies
The Vera C. Rubin Observatory is poised to transform our understanding of the redshift > 2 universe, and its journey to first light represents a major technical and scientific milestone. In this talk I will discuss commissioning the Rubin Observatory, with a focus on the challenges associated with employing active optics to deliver pristine image quality across Rubin’s 9.6 sq deg field of view. I will then discuss prospects for high-redshift cosmology using hundreds of millions of Lyman-break galaxies (LBGs) detected by Rubin’s Legacy Survey of Space and Time (LSST), including forecasts for constraints on the evolution of dark energy and early results on detection of LBGs in Rubin’s DP1 dataset.
Sep 18, Thursday
3:30 pm (Astronomy Colloquium)
Emmanuel Schaan, Stanford/SLAC
Campbell 131
Backlighting the Universe with the Cosmic Microwave Background
Upcoming large-scale structure (LSS) surveys promise to turn the Universe into a particle physics laboratory and determine the properties of dark matter, dark energy and the neutrinos. However, doing so requires disentangling cosmology from the astrophysics of galaxy formation, which obscures the connection between visible and dark matter, and modifies their spatial distributions. A powerful approach to simultaneously answer these astrophysics and cosmology questions is to use the cosmic microwave background (CMB) as a backlight for the LSS, to reveal its invisible properties like watermarks on a bill. I will present innovative methods to produce percent-precision maps of the baryonic matter (Sunyaev-Zel'dovich effects) and dark matter (gravitational lensing) on cosmic scales, from combinations of CMB experiments like the Atacama Cosmology Telescope and Simons Observatory with LSS experiments like the Dark Energy Spectroscopic Instrument and the Rubin Observatory. These could transform our understanding of galaxy formation, and shed light on dark matter, dark energy and the neutrinos.
Sep 22, Monday
12:30 pm (4D seminar)
Aurora Ireland, Stanford
classroom 402, Physics South
The B-Side of Gravitational Waves: Imprints of Primordial Tensor Perturbations in CMB B-Modes
Primordial sources of gravitational waves (GWs) have traditionally been probed through their contribution to the stochastic GW background, detectable via pulsar timing arrays and ground-based laser interferometers. However, these same tensor perturbations can also leave an imprint on the cosmic microwave background (CMB) in the form of B-mode polarization. While a detection of primordial B-modes has long been regarded as a smoking gun for inflation, recent work has shown that non-inflationary sources in the early universe can also produce B-mode polarization on observable scales with amplitudes potentially within reach of upcoming CMB experiments.
In this talk, we examine two such sources: 1) first-order cosmological phase transitions occurring in the late, but pre-recombination, universe, and 2) scalar-induced tensor perturbations, generated at second order in cosmological perturbation theory from enhanced small-scale curvature perturbations. For each, we compute the angular spectrum of B-mode polarization and compare to that from inflation assuming tensor-to-scalar ratios targeted by next-generation CMB experiments. We find that in both cases, there exists viable parameter space yielding signals rivaling those from inflation. The scenarios can in principle be distinguished by making observations at multiple angular scales, since the spectral shapes are distinct from the inflationary case, featuring more power on smaller scales.
These results motivate broader observational strategies for uncovering primordial sources of GWs beyond inflation.
Sep 23, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Dhayaa Anbajagane, Chicago
Campbell 131
Bringing the Cosmological collider into the Nonlinear Universe
Primordial non-Gaussianities (PNGs) are imprints in the density field caused by particle interactions during the inflationary epoch. A subclass of these signatures ("cosmological colliders") encode information about the interactions and properties of any additional particle fields present during inflation. I introduce a novel method for producing cosmological simulations with arbitrary collider signals, and present signatures from over thirty collider models in the deeply non-linear regime of the density field. I show how weak lensing surveys can offer competitive constraints that are complementary to other probes of PNGs in the late Universe. This simulation framework expands the probes of nonlinear structure (beyond lensing) that are usable in constraining non-Gaussian signatures, which in turn expands the space of testable models and enhances the precision with which we constrain them.
Sep 25, Thursday
3:40 pm (Astronomy Colloquium)
Ana Bonaca, Carnegie
Campbell 131 and ZOOM
The nature of dark matter with stellar streams in the LSST era
More than 130 stellar streams, long thin filaments of stars produced by tidal dissolution of star clusters and dwarf galaxies, have been discovered in the Milky Way halo. In addition to outlining the rich accretion history of the Milky Way, these streams are sensitive tracers of dark matter in the Galaxy. 3D positions and velocities of stream member stars identified in large kinematic surveys like Gaia and DESI have allowed us to constrain the mass and shape of the Milky Way's dark matter halo, with tantalizing evidence of a tilt with respect to the Galactic disk. Further, streams provide tentative evidence of dark matter substructure: a handful of best-observed streams have features consistent with impacts of low-mass dark-matter subhalos, although alternative explanations have not been ruled out fully. Upcoming facilities like the Vera Rubin Observatory's LSST (science verification in process) and the Via Project (first light early 2027) will provide more detailed observations of the structure and kinematics for a hundred stellar streams in the Milky Way halo. I will discuss how these data will isolate dark matter signals in stellar streams, constrain the linear matter power spectrum on scales below the threshold for galaxy formation (~10^6Msun), to ultimately explore dark matter particle candidates in the mass regime of <~25keV.
Sep 25, Thursday
4 pm (sharp, RPM, LBL)
Jim Strait, LBNL
LBL 50A-5132 and ZOOM
CMB-S4: birth, life and death of a major science project
CMB-S4 was envisioned to be a comprehensive ground-based CMB experiment to make transformative discoveries and new insights in fundamental physics, cosmology, astrophysics, and astronomy. It was conceived as a way to unite the entire CMB community to enable a ground-based experiment with capabilities beyond that of any of the smaller individual experiment to realize the enormous potential of CMB measurements for understanding the origin and evolution of the Universe, from the highest energies at the dawn of time through the growth of structure to the present day. CMB-S4 was launched with great promise and enthusiasm at the 2013 Snowmass Summer Study, grew into a large collaboration, built a capable project organization, developed an advanced technical design and scientific strategies, was endorsed by multiple high-level advisory committees, but ultimately was cancelled by the funding agencies in 2025. In this talk I will summarize the history and accomplishments of CMB-S4, and provide an analysis of why it did not succeed.
Sep 29, Monday
4:15 pm (Physics Colloquium)
Aaron Roodman, Stanford
#1 Physics North
First Look with the Vera C. Rubin Observatory and the World’s Largest Camera
The Vera C. Rubin Observatory, which features the world’s largest digital Camera, will observe billions of galaxies, billions of stars in our own galaxy the Milky Way, as well as millions of objects closer to home in the solar system. Every night over a ten year survey, the Rubin Observatory will observe much of the night sky, so that every portion of the sky will be imaged nearly a thousand times. Rubin has just released its first images, with thousands of asteroids and millions of stars and galaxies. In this talk I will describe the capabilities of this new observatory, some of the early results from the initial images and its future discovery potential.
Sep 30, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Raagini Patki, Cornell
Campbell 131
Probing Fundamental Physics in New Ways with the kSZ Effect
The kinematic Sunyaev-Zel'dovich (kSZ) effect has become a powerful probe of baryons and cosmic velocity fields, with current high-resolution CMB data and upcoming maps from the Simons Observatory (SO). By pairing CMB maps with large-scale structure (LSS) surveys such as DESI, forthcoming kSZ measurements will potentially help shed light on enduring questions in physics. In this talk, I will present two new statistical estimators employing the kSZ effect, which can help advance our understanding of these questions.
I will first discuss a novel temperature-temperature-density bispectrum, which extracts the kSZ signal using any projected-field of LSS tracers. This signal has better scale-separation than the existing kSZ-squared method, and peaks for squeezed triangles. We forecast high SNRs for SO crossed with a linear galaxy field from WISE/Rubin and competitive constraints on neutrino masses. In the second half, I will present a new way to test gravity on the largest scales possible, by combining kSZ velocity-reconstruction with another CMB secondary - CMB lensing. We construct an estimator of the 'E_G' statistic from galaxy cross-correlations with convergence and kSZ-reconstructed velocities. I will show forecasts for SO and ACT with DESI galaxies, and discuss their potential to distinguish between general relativity and modified gravity at linear scales.
October 2025
Oct 7, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Natali de Santi, UCB
Campbell 131
Learning Cosmology with Graph Neural Networks
It is well known that there is no consensus on the optimal summary statistic for constraining cosmological parameters. Traditional approaches, such as the power spectrum or bispectrum, often fail to capture the rich non-Gaussian and non-linear information present in large-scale structure. In this talk, I will present an alternative approach to predicting the matter density parameter, Ωm, using graph neural networks (GNNs). By translating galaxy and/or halo catalogs into graphs, GNNs provide a natural framework to incorporate geometric constraints, symmetries, prior physical knowledge, and some systematics. I will show that these models are not only accurate but also robust: they can generalize across different hydrodynamical simulations, cosmological and astrophysical parameter variations, different halo and subhalo finders, and even semi-analytical models. This robustness makes GNNs a promising tool to bridge the gap between simulations and observations.
Oct 13, Monday
4:15 pm (Physics Colloquium)
Risa Wechsler, Stanford
#1 Physics North
Unveiling the nature of dark matter with small-scale cosmic structure
Cosmological and astrophysical measurements provide clear evidence for the existence of dark matter, and in recent years they have started to map dark matter over large volumes of the cosmos – but we still do not know the mass or interaction properties of the dark matter particle(s?). I will describe how measurements of small-scale cosmic structure – including dwarf galaxies, gravitational lenses, and the properties of stellar streams – are providing insight into the microphysical properties of dark matter. Large photometric and spectroscopic surveys coming online in the near future will accelerate this search in a way that is complementary to other dark matter probes, and have the potential to detect truly dark halos below the limit of galaxy formation and confirm or rule out a key prediction of our standard cosmological model.
Oct 14, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Anshuman Acharya, Berkeley
Campbell 131
An outside-in approach to the Epoch of Reionization: one simulation to rule them all?
While the Lambda-CDM model has been incredibly successful, it faces persistent tensions between high- and low-redshift observations, particularly in the Hubble constant (H0) and the matter clustering parameter (S8). A key avenue to resolving these discrepancies lies in the interplay between astrophysical processes and the underlying cosmology. In this talk, I will motivate the Epoch of Reionization (EoR) as the ideal laboratory to probe this interplay. However, currently, interpreting observables from this period is a disjointed effort. Models built for galactic observations from telescopes like JWST often lack the large scales needed to interpret observations of the Intergalactic Medium (IGM), such as the 21-cm signal. Conversely, models built for IGM scales typically lack the physical prescriptions and resolution needed for galactic observables. To address this, I have been involved in building the Polar simulation suite, a singular platform designed for such multi-scale, multi-probe studies. I will demonstrate how Polar can be used to explore the impact of varying cosmological and astrophysical parameters, laying the groundwork for a unified, simulation-based inference platform applicable to a wide range of observables.
Finally, I will highlight the necessity of this approach, showing that while individual observations may not resolve the cosmological tensions, combining multiple observables will be crucial. With upcoming facilities such as SKA-Low, the Simons Observatory, and COMAP, we can simultaneously constrain the cosmological and astrophysical parameters that model the EoR. I will also discuss the current limitations of this work, which directly inform the future research I hope to undertake here.
Oct 21, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Erik Zaborowski, OSU/LBL
Campbell 131
tba
Oct 23, Thursday
12 noon (BIDMaP seminar, RSVP requested)
Shivam Pandey, JHU
373 Cory Hall
tba
Oct 28, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Irene Abril Cabezas, Cambridge
Campbell 131
tba
November 2025
Nov 4, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Calvin Preston, Cambridge
Campbell 131
tba
Nov 10, Monday
12:10 pm (TAC seminar)
Neha Anil Kumar, JHU
Campbell 131
tba
Nov 11, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Aaron Ouellette, Illinois
Campbell 131
tba
Nov 18, Tuesday
1:10 pm (BCCP/Cosmology seminar)
tba
Campbell 131
tba
Nov 25, Tuesday
No seminar
December 2025
Dec 2, Tuesday
1:10 pm (BCCP/Cosmology seminar)
Thomas Bakx, Utrecht
Campbell 131
tba
|
