Today’s colloquium was given by Brad Gibson of the University of Central Lancashire, a place you couldn’t blame us for not having heard of before. It is, among other things, the site of England’s tallest church (not counting cathedrals, of course), the home town of the actor who played R2-D2 in Star Wars, and the residence for six months of Charles Dickens while he wrote his novel Hard Times. It is also the new home of the Jeremiah Horrocks Institute, a relatively new center for computational physics where they are doing some very interesting things.
More to the point, our speaker told us a story about computer simulations of the formation and evolution of galaxies. A lot of this subject has to do with the hierarchical merging of galaxies as they collide, and it has a longer history than many people might think. The most-cited paper is Toomre & Toomre (1972), but an extraordinary and little-known effort in analog computation was made by Holmberg (1941): a contraption involving light bulbs and a photosensor, moved painstakingly around the floor of a dark room. The assumption here seems to have been that mass follows light, so that you can predict the gravitational forces inside and around a galaxy by looking at its shape in astronomical images. At the time this was not too bad an assumption given that most of the detailed strong evidence for dark matter hadn’t yet surfaced, and even now it is still true to within a constant of proportionality (which depends on the galaxy’s mass). All the features familiar to those who look at interacting galaxies, such as long tidal tails created in the galaxy interaction and the loss of kinetic energy and angular momentum through tidal interactions, appear even with this analog computation method. Gibson said he has been mentioning this in all his talks since 2006, and the citation rate for that paper has since spiked (you can look at it through the ADS link).
The story he told afterwards can be summarized as follows: Most contemporary simulations of interacting galaxies seem to be too “bulgy” to fit observations. Spiral galaxies such as our own Milky Way are typically made up of a flat, rotating disk and a bright central bulge, and the relative prominence of these features is one way we classify galaxies. There are many examples of spiral galaxies which are virtually all disk and no bulge, however:
The trick to fix this, or so he and his students seem to believe, has to do with a set of phenomena called “feedback” in the field. Although the general behavior of gas is to radiate its energy away, cool down, lose pressure and contract under its own gravity to form stars, many things happen to dump energy back into the surrounding medium and inhibit star formation or monolithic collapse of the gas: formation of young, massive, bright stars which ionize the surrounding gas, supernovae which also ionize and may push the surrounding gas outwards, and formation of active galactic nuclei from black holes in the centers of galaxies. The simulations are usually not sufficiently detailed to describe all of these things happening, because like supernova explosion simulations I’ve talked about before, you’d need to describe in detail things smaller than our solar system in a volume as big as the Milky Way, and no computer even has enough memory to represent that properly, let alone the CPU horsepower needed to evolve the system forward in time. So you have to cut corners and describe the “microphysics” happening inside things much smaller than a galaxy in an approximate way, and the way you choose to cut corners may have a big impact on the results you get out.
Anyway, the feedback of energy into the gas inside most galaxies happens with some efficiency. Most scientists studying galaxy formation have assumed that that efficiency is low, so that only a small fraction, maybe 10%, of all the energy available actually gets deposited into the gas rather than just radiated away into space and lost. What Gibson and his group showed was that if you make it harder to form stars in the first place (requiring gas densities up to 1000 times higher than previously assumed), and also assume high feedback efficiencies (so that close to 100% of the available energy goes into heating up the surrounding gas), you can make galaxies that have very small bulges. Moreover, the simulated galaxies end up satisfying a lot of the empirical scaling relations common for observed spiral galaxies in the local universe, such as the baryonic Tully-Fisher relation, surface brightness of galaxies with radius, “rotation curves” (angular speed of the matter orbiting around the galaxy’s center vs. distance from the center), etc. They still have trouble reproducing the gradients of chemical composition from the center of a galaxy to its edge, but that just gives them something else to work on.