Benjamin Horowitz
Who I work with...
Learning UnlimitedAbout Me
I’m a Kavli Fellow at the Institute for the Mathematics and Physics of the Universe (IPMU), formerly at Lawrence Berkeley National Lab and Princeton University. My research has mainly focused on understanding the distribution of “stuff” in the universe (the “large scale structure”) and using that information to make inferences about underlying physics. Much of my recent work has focused on developing statistical methods to reconstruct distant cosmic structures through a variety of probes.
In addition to my work in physics, I’m deeply involved with “Splash” educational outreach/enrichment programs where university students teach short classes on non-standard topics to area high school and middle school students. Classes can be on anything ranging from the physics of black holes to german fairy tales. Students get to learn about things that they normally wouldn’t have access to and empower them to find new passions. Through the national umberella organization, Learning Unlimited, I’ve mentored students at UC Berkeley, UC Merced, Oxford University, Bard College, and Nothwestern University to run their own successful program.
Current Research Focus
Reconstructing the Large Scale Structure of the Universe
The distribution of matter in the universe holds a wealth of information about the fundemental nature of the cosmos. As time moves forward, structure evolve under the force of gravity given their initial conditions from the early universe. By understanding the current distributon of matter, we can rewind back time to study the initial conditions as well as constrain the gravitational evolution itself.
Some of my recent work has been focused on understanding the emergent structures in the universe from a variety of cosmological probes using maximum likelihood methods.
Improved Lyman Alpha Tomography using Optimized Reconstruction with Constraints on Absorption (ORCA)
Efficient Optimal Reconstruction of Linear Fields and Band-powers from Cosmological Data
Connecting Gas/Galaxies in the Universe to the Underlying Dark Matter with Machine Learning
While the majority of matter in the universe is composed of mysterious dark matter, the objects we can actually observe consist of baryons. While the two are closely correlated, the complicated physical processes of galaxy formation and gas shock heating make drawing a direct connection between the underlying dark matter and the observed structures difficult. In practice, we often have to run hydrodynamical simulations which are incredibly expensive computationally. My recent work has focused on accelerating this process, and making the underlying models differentiable, via use of generative neural network models.
HyPhy: Deep Generative Conditional Posterior Mapping of Hydrodynamical Physics
Fast, high-fidelity Lyman α forests with convolutional neural networks
Understanding the Structure of the Universe through the Cosmic Microwave Background
The Cosmic Microwave Background provides a backlight on universe and we can analyze structures not only through the radiation they emit but through the interaction of the matter with light from the cosmic microwave background. This has the advantege of allowing us to analyze aspects of galaxy clusters that cannot be inferred from the visible light, such as the temperature of the intracluster medium (through the thermal Sunyaev Zeldovich effect) or the overall mass of the cluster (through their gravitational lensing).
Reconstructing Small Scale Lenses from the Cosmic Microwave Background
Cosmological constraints from thermal Sunyaev Zeldovich power spectrum revisited
Exploring Cosmological Signatures of Beyond the Standard Model Physics
Perhaps the most exciting aspect of modern cosmology is the ability to use the large datasets available to constrain exotic physics models. Of particular interest recently has been the theory that dark matter is composed primarily (or exclusively) from a large primordial black hole population. In this work, I used observations from the Planck Satellite’s observations of the Cosmic Microwave Background to constrain those models.
Revisiting Primordial Black Holes Constraints from Ionization History