Researchers examine galactic formation and evolution


Last Tuesday, as a part of Bennett-McWilliams Lecture, Carnegie Mellon alumni invited Lars Hernquist, a theoretical astrophysicist and Mallinckrodt Professor of Astrophysics at the Harvard-Smithsonian Center for Astrophysics, to Carnegie Mellon’s campus to give a lecture on galaxy assembly and evolution.
Hernquist talked about Illustris, a project that uses a new computational model developed by Hernquist and his team, which consists of Volker Springel, a professor in the Department of Physics and Astronomy at Heidelburg University; Mark Vogelsberger, an assistant professor in the Department of Physics in MIT; and a contingent of students and faculty.
Galaxies are the building blocks of the universe’s large-scale structure. Due to the rapid development of technology and cosmology over the past centuries, there has been a recent explosion in robust observational literature on galaxies, as well as several studies of individual galaxies. In scientists’ effort to explore the universe, it seems necessary to have a predictive theory of galaxy-assembly with which to interpret this massive amount of data and to ultimately understand the origin of stars.
This question has become a very long-standing problem. Starting in the 1920s, various solutions have been put forward, but there’s still no fully predictive theory. The major complications are that the theory needs to account for normal physics — as opposed to anti-matter and the like. The theory will also have to handle galactic scale — from black hole accretion disks to star-forming regions to halos, all at the same time. Also, many processes relevant to galaxy formation and evolution are still poorly understood. The good news, however, is that researchers know where to begin.
The Illustris project description states that: “The Lambda Cold Dark Matter (Lambda-CDM) paradigm of cosmology, currently favored by observations of the large-scale distribution of galaxies in space, implies that the cosmos is filled with three distinct components: normal matter, dark matter, and dark energy.
The mathematical models which govern the physical behavior of these components are sufficiently complex that they can only be solved exactly for very particular, simplified ‘test’ problems.”
Hernquist concludes that fundamentally, the situation is a computational problem, which requires a numerical approach that is adaptive in space and time.
“A number of fundamentally different methods exist for simulating gas on a computer,” Hernquist said.
In astrophysics, most researchers have used one of two approaches. The first is smoothed particle hydrodynamics (SPH), where the mass of the gaseous fluid is parceled out to a discrete number of particles. These particles move in response to the combined forces of gravity and hydrodynamics, and their position at any time indicates where the gas is, as well as how it is moving.
Hernquist continued that the second approach of Eulerian, or mesh-based methods typically utilize a scheme called adaptive mesh refinement, or AMR. Through this method, space itself is divided up into a grid, and the flow of gas between neighboring cells of this grid is computed over time.
The Illustris simulation uses a different approach, employing a “moving, unstructured mesh.” Moving mesh treatment of hydrodynamics offers many advantages over other approaches for cosmological simulations of galaxy formation. Like in AMR, the volume of space is discretized into many individual cells, but as in smooth particle hydrodynamics, these cells move with time, adapting to the flow of gas in their vicinity.
As a result, the simulation has particle-based strengths that include geometrical flexibility, accurate and efficient gravity solvers, and continuous refinement, as well as grid-based strengths that include resolution of discontinuities, relatively less diffusive, and well-defined convergence criteria.
Several projects are now being carried out by the team members. With Ilustris, researchers are able to observe fine structures of galaxies and interacting galaxies. Research by Vogelsberger in 2014, and Greg Snyder in 2015, regarding galaxy morphologies and colors shows approximately observed abundance of morphologies of galaxies: 30 percent of spirals, 30 percent of elliptical, and 10 percent of irregulars at z=0.
Illustris can also help in tracking galaxies in redshift, or progressing farther away, such that the wavelengths that actually reach observers appear more red, though with some variations in simulation forwards and backward because galaxy number is not conserved in our universe due to the fact that galaxies merge at an unpredictable rate.
By measuring the slope of stellar density profiles in halos, Illustris shows that the slope correlates with the mass of the dark matter halo, thus researchers can estimate dark matter halo mass from stellar halo.
Researchers can also estimate galaxy-galaxy merger rates directly from Illustris, which shows that the rates have a strong dependence on z (redshift). This result is in reasonable agreement with data, but astonishingly opposes results from semi-analytical models done in the past using crude methods, which show that merging rates have less to do with z, but a lot with mass.
Also, by dragging the time of Illustris to the future, it shows that matter in the universe is moving more rapidly apart, meaning galaxies are more unlikely to merge.
Illustris shows promise, and Hernquist estimates that in the near future, in approximately two months, the team will improve Illustris with refined feedback models. The team aspires to improve performance and scalability.