Cameron Hummels


Cameron Hummels



About me

As a National Science Foundation Postdoctoral Fellow at Caltech I spend most of my time using computer simulations to model how galaxies form and evolve over the age of the universe, but I also perform research in other areas of astronomy. I have used the Arecibo radio telescope to observe fuel for star formation in other galaxies, and I have investigated how explosions occur on the surface of the Moon. Computationally, I contribute code to the Enzo hydrodynamics simulation code, the yt visualization and analysis suite, and the Trident synthetic spectral generation tool, and I am a member of the FIRE simulation group, the AGORA simulation group, and the PI of the Tempest simulations.

I am very involved in astronomy outreach, currently directing the Caltech Astronomy outreach efforts around Pasadena and LA. As a graduate student at Columbia University, I worked as Director of Outreach, where I helped build up their program to be one of the premier astronomy public education efforts in the country.

My Curriculum Vitae
ADS Search

My Research Topics

Computational Galaxy Evolution

I investigate how galaxies evolve over cosmological timescales and the forces responsible for this evolution. My research goals focus on understanding the nature of star formation and stellar feedback in galaxies, the primary mechanisms by which gas, energy, and metals are expelled into the outer parts of galaxies and beyond. These processes are crucial to the dynamical and chemical evolution of galactic systems, yet they are still not currently understood in computational and observational contexts.

Circumgalactic Medium

One new avenue for comparing observations and theoretical predictions of galaxies is the circumgalactic medium, the vast reservoir of tenuous gas surrounding each galaxy out to several hundred kiloparsecs. Observations are now being made of the state of this gas, and they are providing a very useful tool for understanding the flow of material and energy into and out of galaxies. I was awarded a Hubble Space Telescope Theory Grant to investigate the nature of the CGM using computer simulations to better understand its origin, evolution, and to explain some of its peculiar observational signatures. Please visit this page for more information on my past results as well as downloadable data.

Tempest Simulations

The Tempest simulations are a series of state-of-the-art cosmological "zoom-in" galaxies employing enhanced halo resolution in the outskirts of the galaxy to better model the gas in the circumgalactic medium. In traditional simulations, spatial resolution is typically tied to the gas density, leaving low-density environments like the CGM with a lack of resolution. This coarse resolution leads to a number of unphysical effects, and by enhancing the resolution in the halo, the Tempest Simulations addresses many long-standing discrepancies between simulations and observations of the CGM. For more about these simulations and the technique of Enhanced Halo Resolution (EHR), please read our recent paper and check out our movies.


Trident is a tool for generating synthetic spectra from hydrodynamical simulations enabling direct comparison between theory and observational data of the CGM and IGM. Britton Smith, Devin Silvia, and I are the lead developers for this new utility crucial for understanding simulations of low-density, extragalactic gas.

AGORA Simulation Comparison

I am involved in the AGORA Galaxy Simulation Comparison Project an effort to directly compare results from galaxy simulations produced by the leading astrophysical hydrodynamics codes. The goal of the project is to identify what characteristics and behaviors of simulated galaxies are real and which are simply products of the hydrodynamical methods employed in various codes. I am co-leader of the working group investigating the characteristics of the simulated CGM in cosmological environments, which will produce synthetic observations to directly compare against each other and real observations. This study should provide insight into why galaxies behave the way they do and what physical processes (e.g. what star formation and feedback prescriptions) are most responsible for producing realistic CGM characteristics.


I develop and use the adaptive mesh refinement hydrodynamics code, Enzo, to perform cosmological simulations following the formation and evolution of individual disk galaxies to present redshift. I am investigating new subgrid models in these simulations to better prescribe the detailed physics on small scales, specifically star formation and efficient stellar feedback. With better models for these two processes, we may be able to avoid dynamical pitfalls like the angular momentum problem and produce galaxies consistent with observations.


I am one of the core developers for the yt project, a software suite for visualization and analysis of computational hydrodynamical datasets. I'm contribute to the project in many ways, but my focus is on halo mergers and tracking through cosmological volumes, volume rendering and visualization, and developing methods for building realistic synthetic observations from simulation outputs for direct comparison against observations.


I am a collaborator and observer on the Galex Arecibo SDSS Survey (GASS), a multiwavelength project survey that targeted 1000 massive nearby galaxies. GASS was the first statistically significant sample of massive transition galaxies with homogeneously measured stellar masses, star formation rates and gas properties. It provides a means of understanding how galaxies react to their environments and their cold gas content, and why the bimodality in the galaxy color-magnitude diagram exists.

Transient Lunar Phenomena

Lastly, I had a brief foray into planetary science a few years ago, investigating how eruptions of gas out of the interior of the Moon would impact its surface. These models predicted sub-surface lunar ice with similar observational characteristics to ice discovered later that year by NASA mission scientists.

My Publications

See our movies. Read our method paper.

The Tempest simulations are a series of cosmological "zoom" hydrodynamics simulations following the evolution of a Milky Way mass galaxy from just after the Big Bang to present day. Unlike most traditional simulations, they employ Enhanced Halo Resolution (EHR), a technique to assure that the outskirts of the target galaxy resolve small spatial scales. As we demonstrate, this is important for the physical modeling of the gas, resulting in qualititave changes in the behavior of the circumgalactic medium (CGM), the low-density gas filling the galactic halo. The global trend is that it makes it more mist-like.

Traditional cosmological hydrodynamics simulations fail to spatially resolve the CGM, usually 16x or coarser spatial resolutions than what is used in the dense gas of the galactic disk. Employing varying levels of EHR, the Tempest simulations demontrate the impact of galactic halo resolution on the simulated CGM. Above, you can see views of two simulated galaxies, one utilizing traditional AMR and the other EHR, in projected neutral hydrogen density. While the AMR simulation coarsely resolves the CGM to 8 comoving kpc spatial resolution near the virial radius, the EHR run resolves it to 500 comoving pc throughout much of the halo volume. The differences in behavior are striking.

Among its many effects, EHR (1) changes the thermal balance of the CGM, increasing its cool gas content and decreasing its warm/hot gas content; (2) preserves cool gas structures for longer periods; and (3) enables these cool clouds to exist at progressively smaller size scales. Observationally, this results in a boost in "low ions" like neutral hydrogen (H I) and a drop in "high ions" like O VI throughout the CGM. These effects of EHR do not converge in the Tempest simulations, but extrapolating these trends suggests that the CGM in reality is a mist consisting of ubiquitous, small, long-lived, cool clouds suspended in a hot medium at the virial temperature of the halo.

In our paper, we explore the physical mechanisms to explain why EHR produces the above effects, proposing that it works both by (1) better sampling the distribution of CGM phases enabling runaway cooling in the denser, cooler tail of the phase distribution; and (2) preventing cool gas clouds from artificially mixing with the ambient hot halo and evaporating. Evidence is found for both EHR mechanisms occurring in the Tempest simulations. Watch our movies to see these mechanisms in action.

The Tempest simulations were performed using the Enzo hydrodynamics code and analyzed with the Trident synthetic observation and yt analysis tools. The Tempest simulations were run on the Blue Waters Supercomputer at the National Center for Supercomputing Applications on the campus of University of Illinois, Urbana-Champaign. Our team of researchers consists of members from institutions across the USA. Support for the Tempest Simulations come from NSF AAPF grant 1501443 and NASA Hubble Space Telescope grant AR-13917.

Simulation Movies

(Downloadable links for each movie are available below.)

Tempest Simulations

Download movie file. Please cite as: "Tempest Simulations: Hummels et al. 2018"

The Tempest simulations are a series of cosmological "zoom" hydrodynamics simulations following the evolution of a Milky Way mass galaxy from just after the Big Bang to present day. Unlike most traditional simulations, they employ Enhanced Halo Resolution (EHR), a technique to assure that the outskirts of the target galaxy resolve small spatial scales. As you will witness, this is important for the physical modeling of the gas throughout the galactic volume.

This movie shows two galaxies taken from the Tempest Simulations, identical aside from their differences in spatial resolution. The galaxy on the left uses a traditional resolution scheme (AMR) only able to resolve at a coarse 4 comoving kpc near the virial radius, whereas the galaxy on the right employs the new Enhanced Halo Resolution (EHR) scheme requiring spatial resolution elements to be no larger than 500 comoving parsecs (16x better) throughout the halo.

The movie displays neutral hydrogen (H I) column density maps for each galaxy. These views are projected face-on to the galactic disk. Both simulations accrete the same cosmological structures: cool filamentary inflows and merging galaxies. But the EHR scheme results in substantially more H I, increasingly found in smaller, denser clouds able to survive for longer.

Details: The Tempest simulations (Hummels et al. 2018) were produced using the Enzo simulation code (Bryan et al. 2014). Image frames were generated using the yt analysis suite (Turk et al. 2010).

Tempest Filament

Download movie file. Please cite as: "Tempest Simulations: Hummels et al. 2018"

The Tempest simulations are a series of cosmological "zoom" hydrodynamics simulations following the evolution of a Milky Way mass galaxy from just after the Big Bang to present day. Unlike most traditional simulations, they employ Enhanced Halo Resolution (EHR), a technique to assure that the outskirts of the target galaxy resolve small spatial scales. As you will witness, this is important for the physical modeling of the gas throughout the galactic volume.

This video follows the evolution of a cosmological filament feeding cool, dense gas into our target galaxy seen in the lower right. The movie consists of four panels: the two on the left correspond to a simulation using the traditional resolution scheme (AMR), whereas the two on the right employ the new Enhanced Halo Resolution (EHR) scheme capable of resolving 500 comoving parsecs (16x better) throughout the halo. The top row shows projected views of the filament in neutral hydrogen column density, whereas the bottom row examines the density and temperature distribution of the gas contained inside the white dotted circle, a sphere 30 kpc across.

Enhanced Halo Resolution (EHR) leads to an increase in cool, dense gas (probed by neutral hydrogen) in the galactic halo. This gas sometimes manifests itself in the form of runaway cooling and precipitation of small dense gas flows and clouds not unlike terrestrial precipitation. This movie allows you to witness precipitation in action, taking place in this cool filament in the EHR simulations (right panels), but not in the AMR simulations (left panels).

In the uneventful, first half of the movie, it can be difficult to trace the continuous flow of the filament as it gets disrupted by other activity in the outskirts of the galaxy. By z~0.8 though, the filament finally stabilizes and the video is slowed down. Look for dense clouds and flows precipitating out of the filament from z=0.7 to z=0.5. Furthermore, look at the difference in the phase diagrams in the bottom panels to see that the precipitating gas in the EHR simulation (right) is ten times denser and cooler than the gas in the AMR simulation (left). In particular, the phase diagram in the lower right panel reveals the smoking gun of runaway cooling, that the gas follows lines of constant pressure (isobaric) from top left to bottom right. For more information about the Tempest Simulations and the effects of EHR on galaxy evolution, please read our paper: Hummels et al. 2018.

Details: The Tempest Simulations (Hummels et al. 2018) were produced using the Enzo simulation code (Bryan et al. 2014). Image frames were generated using the yt analysis suite (Turk et al. 2010).

FIRE Simulations

Download movie file. Please cite as: "FIRE CGM: Hummels et al., in prep."

A movie of one of the FIRE simulations following the evolution of a hypothetical galaxy from just after the Big Bang to present. This simulated galaxy has a similar mass and composition as to what we believe the Milky Way had throughout its evolution.

On the left panel, you can see the projected density evolution, which effectively shows how mass and material in space build up to form this galaxy over time. On the right, you see the temperature of this gas. Notice how the galaxy grows by accreting smaller galaxies. This demonstrates how galaxies form hierarchically.

As the relatively cool gas from these smaller galaxies is absorbed, it provides fuel for star formation (and later supernovae!) in the interior of the galaxy, which results hot blasts of material to erupt from the galaxy during periods of high accretion. Later in time, there is not as much material flying around the Universe, so by the time we reach the end of the movie (present day), the galaxy becomes pretty quiescent and stable, just like we see in the Milky Way today.

Details: This is a 100 comoving Megaparsec box of a cosmological hydrodynamics simulation zoomed on a single L* galaxy. The FIRE simulations (Hopkins et al. 2018) incorporate metal cooling, star formation, and a sophisticated stellar feedback prescription including radiative, mechanical, and thermal effects from type I and II supernovae and hot massive stars. The simulation was conducted with the Gizmo hydrodynamics code (Hopkins et al. 2015) and visualized using the yt analysis suite (Turk et al. 2010).

Cosmological Evolution

Download movie file. Please cite as: "Movie from C. Hummels"

A simulation following the evolution of a large representative volume of the Universe from just after the Big Bang to the present day. We're looking at the projected gas density from the side of the cube, and our view follows the reference frame of the expanding Universe over this time.

As you can see, initially the Universe starts out dense and uniform, but over time slight overdensities in the gas are amplified by the attractive effects of gravity to form the walls, filaments, and clusters seen here. Also note the axes scales change over time as the Universe expands by a factor of 100 between z=100 and z=0, and observe the significant drop in the general column density over that period.

Details: This is a 100 comoving Megaparsec box of a cosmological hydrodynamics simulation conducted with the Enzo hydrodynamics code (Bryan et al. 2014) and visualized using the yt analysis suite (Turk et al. 2010).

Disk Galaxy

Download movie file. Please cite as: "Movie from Hummels & Bryan 2012"

This is a fly-around of an output from an Enzo hydrodynamics simulation which followed the evolution of a Milky-Way-like galaxy from just after the Big Bang to present (z=99 to z=0). This output occurred at z=1.

Initially, you see the full cosmological volume of the simulation of 30 Megaparsecs (~100 million light years) on a side. You're looking at isodensity contours of the gas, so brighter and greener regions imply higher gas densities there. You can see the filaments of material which have formed over time by gravitational collapse of matter in the Universe.

After a few rotations, the camera zooms in on the target galaxy, a 10^12 solar mass disk galaxy, similar to our own Milky Way. You can see it has a disk and some spiral structure in the density contours. The movie switches to showing just column density to better display the galaxy's structure.

Lastly, the movie changes modes to displaying the stellar component of the galaxy. Red stars represent old stellar populations, whereas blue stars represent younger stars. Each particle is about 10,000 stars in the simulation. From this, one can see that younger stars tend to be clustered in the disk region as is true in our own Milky Way.

Details: Simulation movie comes from Hummels & Bryan (2012) rendering a Milky-Way-like galaxy at z=0 produced using the Enzo simulation code (Bryan et al. 2014) and visualized using yt (Turk et al. 2010).

Synthetic Spectra

Download movie file. Please cite as: "Trident Code: Hummels et al. 2017"

An illustration of how absorption spectra are generated by light passing through gas. The left panel is a two-dimensional slice taken from a hydrodynamics simulation of a spiral galaxy where we are viewing the gas number density and a light ray passing through that gas distribution. Overplot on the gas density are the velocity vectors of the gas.

The right side shows the density of silicon-bearing gas encountered by the light ray as well as the line of sight velocity of that gas. In the bottom right panel, we see the spectrum generated by the light ray as it passes along the sightline zoomed in on the Si II 1260 Angstrom feature. The green line represents the "pure spectrum" and the black channel maps have added noise (S/N = 10) as well as a background quasar and Milky Way foreground added.

Details: Simulation output is for a sub-L* galaxy at z=0.2 produced using the Enzo simulation code (Bryan et al. 2014). Spectra were generated using the Trident synthetic spectral generation code (Hummels et al. 2017) and visualized using yt (Turk et al. 2010).

Synthetic Spectra

Download movie file. Please cite as: "Trident Code: Hummels et al. 2017"

This movie demonstrates how absorption features are generated (e.g., Lyman alpha forest) for a spectrum taken from a distant bright background object. It uses a hydrodynamical simulation following the evolution of a large and representative volume of the universe.

In the top panel, we view our chunk of the universe in projected gas density, where you can see the distribution of matter on very large scales in our universe, mostly visible as filamentary structures composing the cosmic web. The movie begins at z=0.1, or about 1.5 before the present age of the universe. There is a red dot to represent the location in our simulated space of a hypothetical quasar, which will act as our bright background light source. On the bottom panel, we see an emission spectrum for that quasar, displaying the light it emits spread out over its various colors, or wavelengths.

The quasar emits light in all directions, but in this movie, we will just follow one ray of light (yellow arrow) as it travels from the quasar through the vast distances to our telescopes at the present day. As this ray of light travels through the volume in the top panel, we see its resulting spectrum at that moment in time on the bottom panel. When the ray of light passes through a dense patch of gas composing one of the filaments, part of its light is absorbed at certain characteristic wavelengths, powering the particles in that gas to jump to higher internal energy levels. In the case of this movie, the light is absorbed at wavelengths capable of ionizing hydrogen gas at 1216 angstroms and 1026 angstroms, referred to as Lyman alpha and Lyman beta respectively.

If the quasar were nearby, then the ray of light would pass through a filament, get an absorption feature at 1216 and 1026 angstroms, and then hit our telescopes and that would be it. But because the quasar is so far, the travel time for the light ray to get to us is extremely long, even though light travels at a very high speed. It takes so much time that we have to account for the expansion of the universe, which stretches and "redshifts" our ray of light to ever larger wavelengths as the universe evolves. The effect of this is that our original spectrum for the quasar in our bottom panel begins to shift and stretch over to the right as the universe evolves over time.

Each subsequent time that the ray of light intersects a dense hydrogen-rich filament, an absorption feature is created at 1216 or 1026 angstroms *at that moment in time*. The end result is that you build up a lot of absorption features as the spectrum gets redshifted while the light ray travels to us. These collections of absorption features are known as the Lyman alpha and Lyman beta forests, and they provide us detailed information about the distribution of matter and large-scale structure in our universe.

You will note in this movie that the light ray does not take a single, straight path. In reality, the light ray *does* take a roughly straight trajectory (accounting for gravitational effects) from the quasar to us, however, our simulation volume is not large enough due to computational restrictions. So we sample this volume over several consecutive outputs from the simulation, representing how this chunk of the universe would look at different times from when the light was emitted from the quasar to when it finally reaches us. Each time the light ray passes across our simulation volume, we randomize its location and trajectory, so as to assure that the light ray doesn't pass through the exact same filamentary structures from the same direction (because this would never happen if we had a larger simulation volume better representing the universe). The result is a sort of disjointed sightline as the light ray passes across time and space on its trajectory to us, but that's OK, as the output spectrum is representative.

Details: This simulation was run using the Enzo hydrodynamic simulation code (Bryan et al. 2014) for a 100 Mpc box using WMAP-9 cosmological parameters. It's relatively low-resolution was by design to simplify its visualization. Images were rendered using the yt analysis suite (Turk et al. 2010). Spectrum was calculated using the Trident code (Hummels et al. 2017).

Citation Information

All movies are licensed under Modified BSD License. Feel free to use these movies, but please cite me as the source of the movie as indicated.

Public Education

One of our duties as scientists is to share our knowledge of the natural world with our communities. At Caltech, I have assumed the ad hoc position of Director of Outreach in the Astronomy Department, leading a series of efforts to increase public education of science in the LA area. I organize the Caltech Astronomy Stargazing and Public Lecture Series to showcase some of the amazing science done at Caltech and provide free opportunities for members of the public to view the heavens through telescopes. I also lead the Los Angeles chapter of Astronomy on Tap, an opportunity for the public to hear short informal science-based talks in a bar while drinking and visiting with world-class astronomers. In addition, I organize a number of guerilla science events aimed at engaging with the public in non-traditional ways like Science Train and Sidewalk Stargazing. Look at some of the events I've organized here.

As a graduate student, I also acted as Director of Outreach for six years, organizing public education events in the Astronomy Department at Columbia University. During my time there, we built up the program to one of the premier astronomy public education programs in the country. I organized, lectured, and volunteered at most of the Columbia Astronomy Outreach events including our biweekly public lecture series and stargazing, Harlem Sidewalk Astronomy, Science Fiction vs Science Fact Film Series, Family Astro Events, and many school group visitations. In addition, I am one of the founding members of the Rooftop Variables scientific mentoring program, whereby graduate students mentor local high school science teachers and help them in designing astronomy curricula and in starting up astronomy clubs at their respective schools around New York City.

During the International Year of Astronomy (2009), I was awarded the position of NASA Student Ambassador to New York State & City. As part of this role, I organized an outdoor astrophotography exhibition in the middle of Columbia's campus, which brought more than 10,000 attendees from around the city and state. I also helped to design and record several educational podcasts as part of the 365 Days of Astronomy project.

I've been featured in numerous media discussing astronomy and education including: National Public Radio, Science Careers Magazine, The Village Voice, and the Brian Lehrer Show. It is my hope that by continuing to bring the beauty of science to a larger audience, we will not only touch individual lives, but aid in improving society as a whole.

Personal Pursuits

Outside of astronomy, I enjoy trail running, experiencing new cultures, learning new languages, long-distance bicycle tours, flying aircraft, brewing beer, making muppet-themed halloween costumes, and racing in ultramarathons and other endurance sports.

I spend much of my free time outside, giving me a chance to think while appreciating the natural world. I regularly train and participate in endurance sports like cycling races, ultra-marathons, and triathlons, achieving some success at Ironman Arizona, the Boston Marathon, and qualifying to represent Team USA at Triathlon Worlds. I have conducted several long-distance bicycle tours, traveling from New York City to Niagara Falls and biking along the American West Coast.

I love international travel since it means I get to meet new people, experience new cultures, and learn new languages. In preparing for these travels, I have tried my hand at numerous languages both formally and informally. It opens so many new opportunities to be able to speak with people in their native tongue and it provides a lot more insight into how different cultures operate.

In line with my love of astrophysics, travel, public education, and challenging experiences, I have applied to NASA to become an astronaut. I hope that one day I will have the opportunity to aid in humanity's exploration and understanding of our little neighborhood in this vast Universe.