24 Sep 2015 Haverford - The brightest galaxies in the universe, known as submillimetre galaxies (SMGs), aren't visible to the naked eye. But look through an infrared telescope, and they light up the sky. These galaxies are old, dating back probably 12 or 13 billion years, and what makes them so luminous is that they form stars very quickly. In our Milky Way galaxy, for example, one or two stars per year are formed on average; these SMGs form about 1000. The origins of these types of galaxies have, since their discovery in the 1990s, been uncertain, but a new paper inNatureby researchers led by Haverford Assistant Professor of Astronomy Desika Narayanan offers the first viable model.
The extreme properties of SMGs have presented a challenge to existing models of galaxy formation. Two general theories have been proposed: one suggesting that collisions between two galaxies may have driven a short-lived but spectacular burst of star formation; the other arguing that SMGs are long-lived objects that slowly accrete mass. However, neither scenario has been able to fully reproduce the observed physical properties of SMGs.
"People have hacked together different kinds of models, but they always violated some observed constraint", stated Desika Narayanan. "What we've done is develop the first model where we've been able to match the range of physical constraints that we know exist. So that is a pretty exciting result."
The simulation that Desika Narayanan and his colleagues have created indicate that SMGs are not transient events but natural, long-lasting phases in the evolution of massive galaxies, sustaining star formation rates of 500 to 1,000 solar masses per year for a billion years. Additionally, this new paper posits that these fertile star-formation rates aren't caused by galaxies banging together, as once thought.
The new simulation, published byNature, is based on more realistic and complete physics, including the energy input of stars to their surrounding gas. It also accounts for the transfer of energy from starlight to heated dust to better estimate how brightly the galaxy glow with long-wavelength light. When played out for several billion years - computations that ran for weeks even on supercomputers with large numbers of computing cores - the simulation generated galaxies that resemble astronomical observations.
The team found that no large mergers were needed to create submillimeter-bright galaxies, just large galaxies with great amounts of gas within a massive halo of dark matter.
"The basic idea is that what we thought were clear-cut cases for major galaxy mergers are actually probably collections of very gas-rich galaxies that collectively are forming tons of stars and are very bright", stated Desika Narayanan.
ThisNaturepaper is the result of roughly 18 months of work by Desika Narayanan and his team. Now that they have solidified a method for their simulations, their next step is to model many additional galaxies for a statistically significant sample. Additionally, they hope to use these findings to begin to uncover how these galaxies relate to other different galaxy populations of the early universe.
"A lot of this work wouldn't have been feasible without the cluster computing we have at the KINSC, and without the support of Joe Cammisa, who manages the cluster", stated Desika Narayanan, who began the work when he joined Haverford's faculty last January. "He really was integral in getting the cluster set up and the software together."