34 AUSTRALIAN SKY & TELESCOPE July 2017
HUBBLE: NASA / ESA / STSCI / MAST / ILLINGWORTH ET AL. 2013, SIMULATION: ILLUSTRIS COLLABORATIONthe Big Bang until the present day and simulate a volume
roughly 300 million light-years on a side. That’s big enough to
house tens of thousands of galaxies. However, it’s at a lower
resolution than zoom-in simulations, and it’s still a fraction
of the roughly 14 billion-light-year-wide box of Outer Rim
(one of the biggest dark-matter-only simulations).
Even BlueTides’ impressive 1.9 billion-light-year-wide box
housing 697 billion particles comes with a caveat: Although
taking the equivalent of one normal computer running for
300 million hours to complete, the simulation only looks at
the first billion years of the universe. “Every simulation has
to make some compromises, because we are fundamentally
limited by computing resources,” notes BlueTides team
member Yu Feng (University of California, Berkeley).
“BlueTides takes the unique approach of focusing on the
earlier universe, when we have strong evidence the physics is
cleaner,” meaning primaeval galaxies are mostly pure gas and
unfettered by complex histories and mergers.
As a result, although BlueTides makes predictions — for
example, that there should be a million galaxies detectable a
billion years after the Big Bang with NASA’s planned WFIRST
survey — those forecasts remain unverified. Confirmations are
held up by the fact that even the most advanced telescopes have
struggled to see far enough back in time to the era in which the
simulation evolves. Yet the promise of new telescopes, including
the James Webb Space Telescope — scheduled for launch in
2018 — offers hope of observing the very first galaxies, and
therefore potential proof of BlueTides’ predictions.
In contrast, Illustris and EAGLE have been able to match
many of the observed features of the present-day universe in
remarkable detail, driving new insights and predictions. For
instance, the EAGLE simulations uncovered the reason galaxies
fall into two distinct types: ‘blue-sequence’ low-mass, usually
disk galaxies (with less than 10 billion Suns’ worth of stars,
similar to the Milky Way) that continue to rapidly form young
stars; and ‘red-sequence’ massive, usually elliptical galaxies in
which star formation has almost completely ceased.
Essentially, they found that the central black hole of a
galaxy competes with stars for gas. If the galaxy is above a
critical mass, gas concentrates in the galactic centre and
the black hole rapidly grows, heating the galaxy’s gas and
preventing it from cooling and forming stars, leading to
a red-sequence galaxy. If, however, the galaxy is below thecritical mass, black hole growth is suppressed by stellar winds
and supernovae, whose outflows prevent gas build-up near
the black hole and keep it available for more star formation.
That leads to a blue-sequence galaxy with a ‘small’ black hole
— one maybe a tenth as big as astronomers would expect for
a red galaxy of the same mass. Observations suggest that this
prediction is in fact true.
Meanwhile, Illustris’ main achievement is its close
agreement with reality, accurately mimicking observationalSREAL OR FAKE? Shown here side-by-side are the real Hubble
Extreme Deep Field observations and a ‘mock’ observation of the Illustris
simulation’s results. Can you find the dividing line?MOCK MILKY WAY The Latte simulation produces from scratch
a Milky Way–like galaxy (top) that looks incredibly similar to our
own galaxy (bottom), even down to the dust lanes. The simulation
is positioned as seen from a star about as far from the galaxy’s
centre as Earth is from the Milky Way’s.
SIMULATED: FIRE TEAM, REAL MILKY WAY: ESO / S. BRUNIERUNIVERSE 2.0