Missing Dwarf Galaxies Never Were

Astronomers may have solved the case of the Milky Way’s missing satellite galaxies. The solution? The galaxies never existed.

Simulated galaxy from FIRE cosmological simulation

This galaxy is a simulation from the Latte Project, which follows the growth of a Milky-Way-mass galaxy across cosmic time. Despite violent outflows in early times, triggered by stellar feedback, a fairly thin disk of young stars still forms. Such disks were a major challenge to produce in previous computer simulations.
Latte Project

New computer simulations may put to rest a long-standing cosmological mystery. Until now, high-fangled calculations of galaxy formation predicted that the Milky Way should have thousands — perhaps even millions — of little satellites swarming around it. These would be clumps of dark matter and, for the larger ones, gas and stars that had coalesced as gravity pulled matter into making structures in the early universe.

Yet astronomers have detected only a few dozen dwarf galaxies around the Milky Way. That’s it. Recent detections of so-called ultra-faint dwarfs have raised speculations that it’s all a matter of how faint a thingamajig we can currently detect. If so, an imminent unveiling of a gaggle of dark dwarfs will soon uphold the simulations and prove cosmologists right.

But a related problem makes that a dubious prospect. If the satellites are all there but just dark, then the couple dozen that astronomers do see would presumably be the biggest ones, full of the most stars and gas and therefore easiest to detect. But when astronomers compare those dwarfs with the largest ones predicted from simulations, the observed satellites are too small. This second conundrum is called the “too big to fail” problem.

Now, new work suggests that the problem is the simulations, not the observations.

Galaxies on FIRE

Up to a few years ago, all simulations had an Achilles’ heel: they only included dark matter. It’s not some sneaky plot; it’s just much easier to calculate the evolution of structure when you’re only dealing with gravity (which is the only force dark matter notices). And such simulations have done a fantastic job of recreating the large-scale cosmic structure we observe.

But a dark-matter-only cosmos is not reality. The real universe contains gas and stars. Astronomers had assumed that dark matter would drag the gas with it, and stars would form fairly peaceably. But they knew that wasn’t the whole story: stars are brutal characters — they lash out with winds and explode catastrophically. The combined supernovae can blast gas out of the galaxy at hundreds of kilometers per second (millions of miles per hour) and even unleash more energy than the gravitational energy holding the whole galaxy together.

Andrew Wetzel and Philip Hopkins (Caltech) and their colleagues have been working to determine how this violence affects galaxy formation with the Feedback in Realistic Environments (FIRE) simulations. These calculations integrate an impressive variety of gas and stellar feedback physics, and their previous results have suggested that, when supernovae go off in dwarf galaxies, the dwarfs are too puny to hold onto the launched gas, killing star formation.

At this week’s American Astronomical Society meeting in San Diego, Hopkins presented the latest results from FIRE’s next iteration, called “the Latte Project: the Milky Way on FIRE.” (Yes, the pun is intentional.) Latte has the highest resolution yet of any simulation of its kind, zooming in to a mere few light-years as it follows the development of a Milky-Way-size galaxy and its legion of dwarfs.

FIRE simulation

These three shots are from two Latte simulations. On the far left is the dark-matter-only simulation, which produced thousands of little satellites around the central galaxy’s dark matter halo. The center and right shots are from the simulation that included both dark (center) and normal (right) matter. That version produced roughly a tenth as many dark matter satellites (little blue clumps), and of those subhalos only nine hosted a satellite galaxy (stars in yellow at right).
A. R. Wetzel et al. / Latte Project

Watch what happens to gas in a galaxy like the Milky Way as it evolves over cosmic time, from nearly the start of the universe to today (redshifts of 100 to 0). Stellar feedback, especially from supernovae, drive massive outflows, visible in the movies. As you can see from the movies, these outflows have a dramatic impact on the material surrounding the galaxy. At later times, things calm down and a recognizable disk starts to form—but not until roughly a redshift of 0.7, or about 6 billion years ago.The box is 160,000 light-years on a side, so at the beginning (the video starts about ten million years after the Big Bang), the galaxies only occupy a small fraction of the space.

(Curious what the stars look like? See the same simulation from a starlit point of view.)

The team ran two versions of the simulation: one with just dark matter, and one with dark matter plus gas physics plus stellar feedback. The researchers found that the souped-up version produced about 700 dark-matter satellite clumps near the disk of the big central galaxy, only a tenth as many “subhalos” as in the dark-matter-only version. And of these 700, only nine managed to form a stars-and-gas dwarf galaxy inside it.

That’s on the order of a thousandth as many dwarf galaxies as what astronomers assumed would exist, based on dark matter calculations alone.

The reasons for this drop are both exterior and interior. The big galaxy itself actually keeps a lot of dwarfs from forming, Wetzel explains. The budding satellites fly through the hot halo of gas that surrounds the central galaxy, and as they do so their own gas is ripped out of them by the ram pressure. (Other theorists have previously suggested ram-pressure stripping could solve the missing satellite problem.) No gas means no stars, which means no dwarf galaxy.

But the central, Milky-Way-like galaxy also destroys the invisible dark-matter clumps in which the dwarfs would have formed. As the satellites orbit the central stellar disk — which is completely absent in the dark-matter-only simulation, Wetzel emphasizes — the disk’s tidal forces yank on the satellites. If the satellites fly within two or three times the disk’s radius, the tides will wrench them apart.

Fake galaxies from FIRE

This slide, from a presentation by Phil Hopkins (Caltech), has both real and simulated galaxies from the Latte Project. The simulations are now sophisticated enough that they’re producing “fake” galaxies that can fool observational astronomers.
Latte Project

Not only does Latte fix both the missing satellite and too-big-to-fail problems, but it also creates galaxies that are eerily real. Hopkins put up a slide of Latte’s simulated disk galaxies during a press conference. “I like showing this at conferences now in the last few weeks and trying to trick observers into guessing which ones are simulated and which ones are real galaxies,” he said. “Some of them can still [distinguish the real ones]. But some of them, I’m starting to actually be able to trick people — which is a big victory in this field.”

Latte explored satellites with down to about 1 million solar masses in stars, or about a hundredth the mass of the Small Magellanic Cloud. Wetzel is now working on an even-higher resolution version (eight times better, in terms of how lightweight a satellite it can detect) that will enable them to see satellites with only a hundredth as much mass in stars. “This mass range gets interesting,” Wetzel says, “because such dwarf galaxies are so faint that we do not yet have a complete observational census of how many there are around the Milky Way.” These are the runt satellites astronomers are now uncovering with projects like the Dark Energy Survey. “With this next simulation, we can start to predict how many there should be for observers to find.”


“Failures No More: The Radical Consequences of Realistic Stellar Feedback.” American Astronomical Society 228th meeting. Abstract 304.05. (Watch the press conference.)

A. R. Wetzel et al. “Reconciling Dwarf Galaxies with LCDM Cosmology: Simulating a Realistic Population of Satellites Around a Milky-Way-Mass Galaxy.” arXiv.org. Posted February 18, 2016.

One thought on “Missing Dwarf Galaxies Never Were

  1. bwanabwana

    Getting to the point of “which ones are simulated and which ones are real galaxies”…

    Neat stuff! Pretty soon we’ll realize we’ve actually been doing the same thing when we stare out into the cosmos and think we see the real thing!?

All comments must follow the Sky & Telescope Terms of Use and will be moderated prior to posting. Please be civil in your comments. Sky & Telescope reserves the right to use the comments we receive, in whole or in part, and to use the commenter’s username, in any medium. See also the Terms of Use and Privacy Policy.