How do galaxies evolve, and is ours ‘normal’?

Project Leader: Dr Claudia Lagos, ICRAR - UWA

Summary

Understanding how galaxies evolve is fundamental to understanding our place in the Universe, as our Milky Way is quite different to even our closest galactic neighbours such as Andromeda and the Magellanic Clouds.  But spanning the reaches of both time and space to study the evolution of different galaxies is beyond the power of even our best telescopes.

Dr Claudia Lagos, an ASTRO 3D Senior Research Fellow at the International Centre for Radio Astronomy Research at UWA is combining numerical modelling of the physical Universe with telescope observations to work out the details we can’t see directly.

 
Core hours
 
Publications from this project
 
Euro (MERAC) Research Award
 
Team members 5 PhD masters students
Partner Institution: The University of Western Australia System: Magnus, Zeus Areas of science: Astronomy Applications used: Boost, GSL, Cmake, Python, Scipy, Matplotlib, HDF5, R

The Challenge

Within our Milky Way, we can directly observe individual stars, and study their masses, colours, ages and distributions to learn a lot about the structure of our galaxy.  But we can only see what is still here, as many early stars have gone supernova and no longer exist.  “We can only see the later portions of everything that has happened in our galaxy,” explains Dr Lagos, “so to learn about galaxy evolution we have to look at progressively younger galaxies, which are progressively further away.”

The problem is, our telescopes can only observe the generalised and cumulative signals coming from distant galaxies, and not the detail of their individual stars.  To study the aggregated characteristics of galaxies 10 billion years ago, statistics and modelling is needed to make sense of what you’re seeing.

Dr Lagos elaborates: “Because there are limitations on what we can see, we can model galaxy formation, and then compare the cumulative outputs of the model with the combined signals we get from our telescopes.”

Unfortunately, galaxy formation can’t be simulated just by solving all of the underlying physical laws from first principles.  “We can simulate processes on a small scale, like how black holes accrete matter at the sub-parsec scale, but not at the same time as we simulate the scales at which galactic structure emerges, as that’s eight or nine orders of magnitude larger.”

“Since we can’t model all of the physical processes at all of these scales, we make approximations at the smaller scales about how individual stars form, and then see how those assumptions affect the developing galactic structures within the model.”

Working out which assumptions and approximations are most accurate then relies on comparing the simulation outputs with real telescope observations.

“Can we accurately predict the variety of galaxies that we can see, their colours, masses, sizes, and population frequency throughout the Universe?  If we can, then the assumptions we have made in creating the model tell us a lot about the underlying structure of the Universe and how those galaxies must have evolved.”

The Solution

Supercomputing is needed to run the model, because it needs to simulate a representative volume of the universe.  “A characteristic scale of the universe is a cube with sides 300 megaparsecs long,” says Dr Lagos.  “In comparison, our local galaxy group containing the Milky Way and Andromeda is a cube of roughly 1 megaparsec.”

Over thirty million CPU hours are needed to calculate the movement and interactions of a billion particles under gravity over time, and up to 40 million CPU hours if the hydrodynamics are considered, to evolve the skeleton structure where galaxies will form.  Then thousands of simulations are run on this skeleton to see how different assumptions and physical parameters affect how the skeleton is populated with galaxies, for comparison with direct telescope observations.

Outcome

Comparing the model with the real Universe is painstaking as it is important to also reproduce the observation conditions, from the detection limits of the telescope to the effects of interstellar dust as it absorbs and re-emits starlight passing through.  But Dr Lagos’ model is already ‘shedding light’ on the formation of very bright, distant galaxies formed about 10 billion years ago, to fill in some of the early history of our own galaxy.

As the evidence mounts that the modelling is accurately representing our known Universe, it becomes more and more useful.  Dr Lagos enthuses: “We’re now starting to use the model to connect observations that represent different epochs in cosmic history – if I look at galaxies similar to the Milky Way, what did they look like five billion years ago?  We’re building a model that will effectively let us run the Universe in reverse, which will give us unique insights into where we’ve come from.  And as telescopes become ever more powerful with the development of the Square Kilometre Array and its precursors, we’ll be able to check our understanding of galaxy evolution against what is really out there progressively further back in time.”

“We’re now starting to use the model to connect observations that represent different epochs in cosmic history – if I look at galaxies similar to the Milky Way, what did they look like five billion years ago?  We’re building a model that will effectively let us run the Universe in reverse, which will give us unique insights into where we’ve come from. 
Dr Claudia Lagos, ICRAR - UWA,
Project Leader.