≡ Menu

Galactic Interaction: Rivers of Stars

Discovered as recently as 1994, the Sagittarius dwarf spheroidal galaxy is a satellite of the Milky Way, and one with an interesting history. One of the nearest of the dwarf galaxies, the Sagittarius dwarf lies 25 kiloparsecs (roughly 82,000 light years) from the center of the Milky Way, and has passed through the disk of the parent galaxy more than once. The result: We see what a new paper on this object calls a ‘stream of tidally stripped stars’ that wraps completely around the celestial sphere. Our own Sun, in fact, is close enough to the Sgr galaxy’s orbital plane that it lies within the width of what can be called the debris tail.

What astronomers would like to do is to reconstruct the orbital history of this interesting dwarf galaxy, something Marion Dierickx (Harvard-Smithsonian Center for Astrophysics), working with her PhD advisor Avi Loeb (Harvard) have now managed through computer simulations. Dierickx and Loeb simulated the movements of the Sgr dwarf for the past 8 billion years, varying factors like the initial velocity and angle of approach to the Milky Way. As expected, these variables have a major effect on the orbit and the resulting stellar stream. Each passage around the Milky Way has pulled the Sgr dwarf apart, costing it component stars.

Matching their models with current observations, the researchers have learned that over time, the dwarf galaxy, which begins the simulations with 10 billion times the mass of the Sun (about one percent of the mass of the Milky Way), loses a third of its stars and nine-tenths of its dark matter. The stripping of this material from the Sgr dwarf produced three streams of stars reaching as far as one million light years from the center of the Milky Way. These streams have become one of the largest structures observable on the sky.

From the paper:

The simulation presented here includes stellar overdensities at distances of up to ∼ 250 − 300 kpc, extending beyond the MW virial radius [the radius within which the density perturbation that will become a galaxy is collapsing]. We provide their predicted positions on the sky, heliocentric distances and line of sight velocities for possible future observational searches. The most distant known stars of the MW coincide with our predicted streams in both position and small radial velocities. If verified observationally, the distant branches of the Sgr stream would be the farthest-ranging stellar stream in the MW halo known to date.


Image: In this computer-generated image, a red oval marks the disk of our Milky Way galaxy and a red dot shows the location of the Sagittarius dwarf galaxy. The yellow circles represent stars that have been ripped from the Sagittarius dwarf and flung far across space. Five of the 11 farthest known stars in our galaxy were probably stolen this way. Marion Dierickx / CfA.

The cosmological landscape that can emerge from work like this is striking. As the paper notes, we should be able to use the Panoramic Survey Telescope & Rapid Response System (Pan-STARRS) datasets to drill deeper into the location of these streams, and future data from the Large Synoptic Survey Telescope (LSST) will offer mapping of the outer galactic halo at visible wavelengths. The Wide Field Infrared Survey Telescope (WFIRST) will produce useful new data in the infrared. With all of these tools, we can probe the outer edges of the Milky Way, in conjunction with existing missions like Gaia. As the paper notes:

The Gaia mission will accurately map a large volume of the MW at smaller Galactocentric radii. With a complete picture of the MW mass distribution from the solar neighborhood to the outskirts of the halo, we will be able to place our Galaxy and the Local Group in a cosmological context.

Note that the 11 farthest known stars in the Milky Way are about 300,000 light years out, well beyond the galaxy’s spiral disk. About half of these are evidently stars captured from the Sagittarius dwarf galaxy, while the rest may have been pulled from a different dwarf galaxy. Says Dierickx, “The star streams that have been mapped so far are like creeks compared to the giant river of stars we predict will be observed eventually.”

The paper is Dierickx & Loeb, “Predicted Extension of the Sagittarius Stream to the Milky Way Virial Radius,” accepted at the Astrophysical Journal (preprint).


Comments on this entry are closed.

  • kzb January 16, 2017, 13:58

    The BBC’s Sky at Night this week featured GAIA. They have discovered a stellar bridge joining the two Magellanic Clouds (only a gas bridge was known before GAIA).
    It was also stated, when looking towards the galaxy central areas, that “there are twice as many stars as we thought”. That is an interesting statement in itself, but it wasn’t expanded on during the programme. Can anyone on here expand on this comment?

  • Joe January 16, 2017, 14:14

    I’ve often wondered about the orbital characteristics of galaxies and dwarf galaxies. This is especially important for the Breakthrough Starshot Initiative. During the 20+ years it will take for a starchip to reach Proxima Centauri, the star will have moved, and all the while Proxima B will be orbiting the star. It will take some extremely good navigation software on the starchip to rendezvous with Proxima B and send data back to our solar system (which is also moving.)

    • Michael January 16, 2017, 16:45

      With a radial velocity of ~20 km/s it will have moved about 12.5 billion km in 20 years. But we can compensate for a fair amount by shooting it ahead of AC, GAIA should give a greater accuracy to aid us.

  • Bruce Mayfield January 16, 2017, 14:18

    This shows that the space between galaxies such as our Milky Way and Andromeda won’t be as empty as it at first seemed. There must be many other tidially striped, widely dispersed remnants of the many dwarf galaxies the MW and Andromeda have canabalized over universal history.

    Isn’t it neat that the always attractive force of gravity still manages to hurl things apart!

  • Alex Tolley January 16, 2017, 16:09

    I’m assuming from the text that the simulations use the standard formula for gravity and adds dark matter. Would the simulations arrive at any significant difference is we assume various MOND formulae for gravity instead and avoid the use of dark matter? Could such simulations help to determine whether dark matter really exists or not?

    • Michael January 16, 2017, 16:51

      I am not sure hydrogen could be that hidden matter, this galaxy would be a strange and lonely place.


    • Robert G January 17, 2017, 3:47

      I’m no expert in MOND (or lamCDM either), but would imagine the following. If we have good 3d position and velocity vectors for large numbers of stars (such as are starting to come through the pipe now), we should be able to integrate Newton’s equations of motion backwards in time, calculating how everything has evolved to get to its current configuration. We can add into this the postulated dark matter and see if anything in the earlier configuration looks nonsensical or contradicts known observations. Then we do the same calculation with modified Newtonian physics and no dark matter, and compare again with experiment. With any luck we will find something that trips up at least one of the two models.

      • Alex Tolley January 17, 2017, 17:25

        I may be wrong, but I don’t think you can model time in reverse for an n-body problem to infer starting conditions.

    • Ron S January 17, 2017, 11:32

      Alex, you have this backwards. The presence of dark matter was inferred from the empirical rotation data and gravitation theory. So of course any gravitation theory, including the countless varieties of MOND, fit the data!

      Also, dark matter is merely a moniker for matter that isn’t seen. Originally anything was deemed possible within the realm of known physics and galactic astrophysics. Since then observations have pretty thoroughly eliminated all foreseeable distributions of baryonic matter, MACHOs and other (more or less) conventional sources. Hence the hypothesis and subsequent search for non-baryonic matter to explain all the data, and not exclusively galactic rotation curves.

      • Alex Tolley January 17, 2017, 17:23

        I don’t think I have it backwards at all. If you are modeling a collision of galaxies, you would have started with just gravitational forces. But since you know that galaxies don’t appear to work like that based on the visible material, you then add in dark matter that will modify the apparent forces.

        The simulation will therefore show the trajectories of the stars and the changes in the shape of the presumed dark matter.
        OTOH, you could run the simulation with just the visible matter if you change the gravitation equation so that you can ignore the dark matter.

        My question is whether teh simulation outcomes look different, and if they do, can that be used as a guide to observations to determine whether dark matter relly exists or not.

        • Ron S January 18, 2017, 21:07

          Alex, my initial reaction was to ignore your comment since the scenario you propose is so convoluted and confounded that such a simulation would produce nonsense. Galactic dynamics require dark matter, or some MOND variation, to match the empirical data. And you want to turn that off when galaxies interact? That’s fundamentally inconsistent. You don’t cavalierly turn bits of physics on and off in different parts of a model and expect to learn anything at all. I maintain my previous position.

          In any case there are simulations of galactic interactions that include dark matter and that reasonably well match the fit of dark matter, not MOND, that is weakly interacting with baryonic matter and with itself. By “match” I mean resulting trajectories of visible and dark matter as shown by direct observation, emissions of radio, xray etc of colliding gas masses and the lensing by the totality of matter and dark matter.

  • Sanford Zane Meschkow January 16, 2017, 18:19

    It’s too bad the stripping of material from the Sgr dwarf galaxy sounds like there would be small chance of conditions there to be anything like the conditions postulated in Poul Anderson’s story “Starfog”. A star cluster diving through the gas and dust clouds of the Milky Way ends up capturing so much matter that it becomes a runaway star nursery loaded with novas and supernovas and systems with such a high percentage of heavy elements that one could mine valuable heavy elements on almost any planet. There is at least one habitable planet in the cluster, which I think is pretty unlikely, but maybe the dangerous region with all the supernovas in it is in the leading edge of the cluster as it plows through our galaxy. Could a star cluster quite a bit less massive than a dwarf galaxy collect material rather than being stripped of it? What effect would the velocity of the passage have?

  • kzb January 17, 2017, 8:58

    Alex Tolley;
    The problem is, the distribution of dark matter is almost a free parameter. It has the distribution that fits the data. That is the way its distribution is measured. (Yes there is weak lensing, but I don’t think that can give the fine details of its distribution.)
    The only way this would work is if the derived DM distribution came out clearly non-sensible.