Research

AGN Circumnuclear Obscuration

Broad Strokes: Active Galactic Nuclei (AGNs) are among the most luminous objects in the Universe. At the center of every major galaxy, there lives a Supermassive Black Hole (SMBH) weighing in up to a billion times the mass of our Sun. When you think of black holes, you likely have in your head some kind of monster that mercilessly eats everything in its path. However, most black holes (even of the supermassive variety) don't do all that much; they just sit around all day disappointing their mothers. (The SMBH at the center of our galaxy fits this description.) However, the Hollywood variety of SMBH that gobbles up a whole bunch of matter does exist, and it turns out that SMBHs are very messy eaters. They spew out* all kinds of nasty radiation and oftentimes outshine their entire host galaxy. This is an Active Galactic Nucleus ("Active" because the SMBH is doing something, and "Galactic Nucleus" because it sits at the center of its galaxy).

(*Actually, this energy does not come from the black hole itself; it is instead the kinetic and gravitational energy of objects falling into the black hole that we see. Black holes do  release some energy by a process known as Hawking Radiation due to weird consequences of quantum mechanics, but this effect is so tiny that we will almost certainly never be able to observe it.)

Zooming In:  AGNs are not isolated in space. They are surrounded by an intricate structure of gas and dust described by the Unified Model of AGNs. These structures include a swirling ring of plasma falling into the SMBH (the accretion disk) the size of our solar system, and a larger doughnut-like structure of order 30 lightyears in radius consisting of clouds of gas and dust (the obscuring torus). 

In my research, I am mostly interested in the shape of the obscuring torus, but other structures exist too, such as clouds above and below the obscuring torus (broad/narrow-line regions) and relativistic jets that point perpendicular to the disk. AGNs are too far away for these structures to be resolved by even the best telescopes, but using models we can deduce certain properties about the torus based on spectroscopic information. This work can be done in either the infrared (IR; less energetic than visible) where we mostly gain information about dust in the torus, or the X-ray (more energetic than visible) where we mostly gain information about gas.

My Current Research: IRAS 09104+4109 is one of the Universe's most powerful AGNs with a luminosity on order one trillion times that of our Sun. It is also (relatively) nearby with z=0.442 (~5 billion light years), and it is heavily obscured, potentially even "Compton-thick" meaning that as a photon travels through the torus it would be expected to bump into at least one atom on average. This makes it a very interesting target for geometric analysis as objects so luminous and obscured are rare, and most of them are so far away from us that we cannot get data with a strong enough signal to do the type of analysis that we want. IRAS 09104+4109 had previously been subject to geometric analysis in a 2016 study that successfully constrained geometry based on IR data, but was unable to constrain the geometry based on X-ray data. Since then, IRAS 09104+4109 has been observed again for 120,000 seconds by the NuSTAR,  a "hard" X-ray space telescope, and I used these data to update the results of the 2016 study. What I found is that: 

You can access the models used for my analysis on my Downloads page.

The Cosmic Web and Machine Learning

Broad Strokes: On the largest scale, the Universe is organized as a "cosmic web". This web is a network-like structure with dense nodes (galaxy clusters). These clusters are connected by bridges (filaments), the largest structures in the Universe, spanning tens of millions of light years through the emptiness of the void. (Interestingly, there are similarities between this structure and that of the human brain on the cellular level! However, it is unlikely that the Universe is sentient, despite some rather... outré theories...) This structure formed due to the gravitational collapse of tiny primordial quantum fluctuations which eventually grew into everything.    

Zooming In: Much is known about galaxy clusters, such as environmental impacts on galaxy morphology  or quenching of star formation. As the densest large-scale regions in the Universe, they are interesting, well-defined, bright, and generally small enough to be observed in a single telescope pointing. Besides being interesting, filaments do not share these properties and are generally more difficult to study. Because they are so huge (resolved or extended even at great distance), they cannot be observed in one go with a telescope. Because they are less dense, they are not as bright as a cluster. Because they are much more vague than a cluster (generally defined by a characteristic distance), there is significant deviation between definitions used between studies. Consequently, there is less work that has been done so far on filament populations. However, there are many reasons we would want to study filaments: They contain approximately half of the matter in the Universe; they feed into clusters, affecting cluster demographics through an effect known as pre-processing; and they form the edges of great cosmic voids which can serve as probes of general relativity and cosmology.

My Current Research: Fortunately, new studies of filaments such as the WEAVE Wide Field Cluster Survey (WWFCS) and the 4MOST CHileAN Cluster galaxy Evolution Survey (CHANCES) will give us the ability to observe filaments with greater numbers than ever before. However, our current methods of determining which galaxies are filamentary is insufficient in these observations. The current standard algorithm used to extract filamentary structure (DisPerSE) determines structural significance of structures using Morse Theory, a subfield of a branch of math called topology. DisPerSE is able to do these calculations in 3D (i.e., on full positional data) and in 2D (i.e., on the positional data wewould get from an observation). However, these do not produce the same results, and of the galaxies classified as filamentary using the 2D extraction, half of them are not truly filamentary. However, there is hope! Recently, a machine learning method known as a random forest was applied successfully in a similar problem. Using kinematic information and halo mass, we built a random forest on top of DisPerSE to classify filament haloes in a dark matter simulation. Although there is still work to be done to apply our algorithm to observations, we have seen some promising results including: