Some of my own science, for a change

There’s not much point in running a science blog if you don’t discuss your own research now and again. The reason I had thought of writing about it is that, last week, I was at a meeting I had helped organise. The meeting was about discs around young stars; what we call protoplanetary, or protostellar, discs. The basic picture is that stars form from clouds of gas and dust that collapse under the influence of their own gravity. However, conservation of angular momentum means that if this cloud has any initial rotation, it will spin faster and faster as it collapses. Therefore material cannot simply fall onto the protostar in the centre, as it would end up spinning so fast that it would break apart. Instead, most of the material falls roughly perpendicular to the rotation axis to form a thin, disc-like structure – the protostellar/protoplanetary disc. Various processes in this disc then act to transport angular momentum outwards, allowing mass to flow onto the central protostar.

discgapThese disc are, however, also the sites of planet formation, and that’s what I was going to mainly focus on here. One thing we know is that big, gas giant planets – like Jupiter and Saturn – must form before the disc has dissipated, otherwise there would be no gas available to form their relatively massive gaseous atmospheres/envelopes. If a gas giant planets does form, then we would expect it to start opening a gap in the surrounding disc, as illustrated in the figure on the right.

In the early 2000s there was some evidence for systems with cavities/holes in the inner regions of their discs, which was interpreted as being evidence for a gap-opening planet. However, in some cases these systems appeared to still have mass accreting onto the central star, which seemed a bit odd if there was a cavity in the inner parts of the disc. However, what was really being detected in these observations was emission from dust grains, not from the gas itself, and so the cavity might have simply been a cavity in the dust disc, rather than in the gas and dust disc.

There had already been a couple of papers pointing out that, in the presence of a planet, the dust could respond quite differently to the gas, and that the response of the dust would depend on the size of the dust grains. Myself and some colleagues then published a paper suggesting that if a planet started opening a cavity in the disc, the dust could undergo a form of filtration. The really small dust grains (micron-sized) would be tightly coupled to the gas and, hence, if any gas was flowing through the cavity into the inner regions of the disc, it would also drag these small grains into the inner disc. Larger grains (mm-sized) would, on the other hand, be prevented from flowing through the gap into the inner disc. This process could therefore reduce the dust-to-gas ratio in the inner regions of the disc, which could explain why these young stars appeared to still be accreting despite there appearing to be an inner cavity/hole.

Credit: Christoph Malin, ESO

ALMA antennae (credit: Christoph Malin, ESO)

After publishing the paper above, I started working on somewhat different projects and didn’t really think about it too much. A couple of years ago, however, I noticed that the paper was starting to collect many more citations than it had been in previous years. Then, at last week’s meeting there were a number of talks discussing observations of discs with possible cavities using the ALMA observatory. The ALMA observatory uses multiple anntennae to make high-resolution observations. It can also observe at different wavelengths, so can probe – in the case of discs around young stars – the structure in dust grains of different sizes.

What the ALMA observations appear to be showing is that in systems with inner cavities, there is indeed often a difference between the small and large grains, with (as shown in the figure below) the smaller grains extending closer to the star than the larger grains. This is largely what we had suggested in our 2006 paper, and so it was fascinating to see something that you had been involved in predicting 10 years ago, possibly being confirmed by observations that couldn’t have been done at the time.

Credit: Garufi et al. (2016)

Credit: Garufi et al. (2016)

As I said above, we weren’t the only group to highlight that gas and dust may behave differently in the presence of a planet. I think, however, that we were the first to propose the basic filtration process. It’s recently been pointed out, though, that dust filtration alone is unlikely to be sufficient, and that other processes may also be operating. This is no great surprise as what we did was pretty simple, and there are almost certainly other processes operating (grain growth, for example).

However, I think this is an still interesting illustration of how science often works. Early observations lead to models/theories that may then require a new generation of observations to actually confirm. By this time, however, the models/theories may have become even more complex and would require an even newer generation of observations to confirm. However, at each stage our understanding improves and, ideally, we converge towards some kind of consistent picture in which the models/theories and observations largely agree. It can, however, take time and it is great to be part of the process, even if your contribution is reasonably modest.

Advertisements
This entry was posted in Personal, Science and tagged , , , , , . Bookmark the permalink.

32 Responses to Some of my own science, for a change

  1. Vinny Burgoo says:

    Cool!

  2. Fascinating.

    Another post I would be interested in would be how stable planetary systems are and why they are stable. Is that still within your expertise?

    I think this is an still interesting illustration of how science often works. Early observations lead to models/theories that may then require a new generation of observations to actually confirm. By this time, however, the models/theories may have become even more complex and would require an even newer generation of observations to confirm. However, at each stage our understanding improves and, ideally, we converge towards some kind of consistent picture in which the models/theories and observations largely agree. It can, however, take time and it is great to be part of the process, even if your contribution is reasonably modest.

    That is then were we are back to the climate debate, where the mitigation sceptics are dreaming of a Galileo paper than will show their political preference right. Possibly that paper is already written, but suppressed by the evil Michael Mann.

  3. Victor,
    I’m doing some work on planetary stability at the moment, but I don’t think I have some kind of post in mind. If I do, I’ll write one.

  4. Tom Curtis says:

    Very interesting, and congratulations on the successful prediction 🙂

  5. John Hartz says:

    You and Phil Plait seem to be cut from the same cloth. Are you twins? 🙂

  6. geoffmprice says:

    Fun, thanks for sharing

  7. Magma says:

    A very basic question: do accretion models include electrostatic effects and if so, are such effects strongly size-dependent?

  8. Magma,
    No, not really. One of the main accretion processes is a magneto-rotational instability and some have worked on how this would operate in the presence of dust grains, but it’s a very complex problem. During the meeting last week we had a discussion session about to improve models of discs with embedded dust and one of the observers asked whether or not we were planning to introduce charged dust grains and magnetic fields. The theorists all just groaned 🙂

  9. BBD says:

    ALMA observations – a pretty picture:

    Source

  10. Yes, that’s HL Tau. It was also discussed quite a lot at the meeting last week. Again, just under 10 years ago, myself and some colleagues published a paper pointing out that VLA observations of HL Tau, showed an excess of emission in the cm at about 65 AU from the central protostar, and suggested that it might be a planet in formation. Where we observed this excess emission is right about in one of the gaps in the above image. However, the ALMA observations are in the mm and sub-mm, and don’t show any evidence for any emission where we observed the excess in the cm. Still not sure what it is, but I think our observations in the cm have been confirmed in other observations. So, something is causing excess emission at cm bands that isn’t seen in mm-bands and sub-mm bands.

  11. BBD says:

    Alien nanostructures 🙂

  12. We haven’t suggested that….yet 🙂

  13. verytallguy says:

    Interesting.

    A technical question.

    Many, many years ago I had a job which required the use of ACSL to model a system. It’s just a way to translate differential equations into code simply, nothing very sophisticated.

    https://en.wikipedia.org/wiki/Advanced_Continuous_Simulation_Language

    In order to learn ACSL, I mocked up a (very crude and crap) simulation of the solar system, and that made me realise how difficult it is to do to any level of accuracy, due to the “stiff” nature of the equations. (Stiff in this context means having very different time constants, eg orbits of moons vs orbits of outer planets)

    That makes it extremely difficult to, for instance, put a comet into the system, as the time steps while it passes a planet close by have to be *very* small for any level of accuracy, but that makes the whole thing computationally intractable over long times.

    From this extremely superficial experience, it strikes me that modelling the formation of a planetary system is fiendishly difficult to do, even neglecting non-gravitational processes such as discussed above. And that the outcome could easily depend more on the computational methods used than the physics.

    Interested in any comment you have.

  14. Andrew Dodds says:

    More propaganda from the round-earth brigade. Try publishing a paper about the Great World Turtle and see how the ‘Round Earth Team’ suppress it..

  15. vtg,
    Interesting comment. You’re right that the different time constants make it difficult to do very detailed computations. Sometimes you can get away with different timesteps for different bodies, so you’re not updating all of them all the time. However, even this doesn’t work if the difference is very large (a very close encounter between two bodies can slow the simulation down so much that it essentially doesn’t progress). There are some clever methods for analytically determining the evolution of close encounters so that you don’t integrate these directly, which can help.

    However, noone has ever done a computation that includes all stages of planet formation. Typically people consider different aspects of the process. For example, there are people who look at gas discs and how the solids evolve in such discs. These simulations, however, typically don’t actually consider how the solids grow, just how different sizes might evolve in such a disc. Others look at discs of solids and how they collide and grow; but even here they typically assume that they’re already quite massive (moon-sized). Others consider the growth of really small dust grains to see how you might actually build large planetesimals. Essentially our understanding is based on a combinations of different types of computations, rather than one single computation that tries to do everything. It would be great if we could do this, but I don’t think we have the computational resources, or methods, to do so yet.

  16. verytallguy says:

    There are some clever methods for analytically determining the evolution of close encounters so that you don’t integrate these directly, which can help.

    OK, that makes sense.

    Wonder what you make of this

    I assume these simulations model galaxies as lots of point masses.

    Maybe the reality of galactic structures is that the distances between stars are so large that close encounters can be neglected as very rare in such a simulation.

  17. vtg,
    I think that is a simulation that doesn’t actually have any gas; the galaxies are just represented by a bunch of particles. Also – as you say – I think that the likelihood of a close encounter is actually quite small. I’ll try and look it up.

  18. Kevin Boyce says:

    My recollection of my one astrophysics class was that indeed, in a collision between two Milky Way size galaxies there was likely to be approximately one hard collision (defined as more than a few degrees change of direction) between stars.

    We also had a guest lecture from Alar Toomre, who showed us his (first ever) modeling of colliding galaxies, made with his brother in 1972. It contained ~500 point masses and took two weeks to simulate on a PDP-11. 20 years later of course larger simulations were being done with spare cycles as screen savers.

    I expect the inclusion of gas would change the dynamics somewhat, although even Toomre and Toomre’s original simulation gave results that look very like actual colliding galaxies.

    You can see one of the simulations at http://kinotonik.net/mindcine/toomre/

  19. Andrew Dodds says:

    vtg –

    Yes.. Bear in mind that a light year is c. 60,000 AU and the sun is a lot smaller than an AU, meaning that stars are incredibly well spaced out. Might as well be points as far as the simulation goes.

    (And the space turtles swim out of the way, of course)

  20. dikranmarsupial says:

    Fascinating, thank you for posting this! I haven’t read the paper (yet) but it sounds like this is an example of gaining understanding of a physical system through computational simulation rather than analysis (as mathematical analysis would be intractable), which is similar to the way climate models are used in climatology?

  21. Andrew Dodds says:

    dikranmarsupial –

    Yes, it’s the kind of thing I did in my brief academic career (but done better). Analytic solutions don’t realistically exist for most macroscopic real world problems like this.

    The question I always want answered from this kind of work is the big one – does it imply that Earth-like planets in Earth-like orbits are common? Probably a bit early to decide.

  22. Kevin,
    Thanks for the comment

    My recollection of my one astrophysics class was that indeed, in a collision between two Milky Way size galaxies there was likely to be approximately one hard collision (defined as more than a few degrees change of direction) between stars.

    Sounds reasonable. In a simulation, however, you can’t include all the stars and so an actual close encounter in the simulated collision is probably still very unlikely.

    I expect the inclusion of gas would change the dynamics somewhat, although even Toomre and Toomre’s original simulation gave results that look very like actual colliding galaxies.

    I don’t know if adding gas would make much different it would. The mass is predominantly dark matter and the stars aren’t really influenced by the gas. That’s probably why even their early simulations gave pretty good results.

    Dikran,

    it sounds like this is an example of gaining understanding of a physical system through computational simulation rather than analysis (as mathematical analysis would be intractable), which is similar to the way climate models are used in climatology?

    Yes, that’s probably about right.

  23. Andrew,

    The question I always want answered from this kind of work is the big one – does it imply that Earth-like planets in Earth-like orbits are common?

    Indeed, many would like to know the answer to this. I think the current view is that it would be surprising if they weren’t quite common, but we really don’t know yet. The one region of parameter space that we can’t properly probe is Earth-like planets in Earth-like orbits around Sun-like stars. The next generation of surveys will probably start to be able to probe this region, though.

  24. I think this is the paper for the galaxy collision video. The key paragraph is

    We ran collisionless N-body simulations of the MW-M31-M33 system, including only stars and dark matter. The calculations were performed with the N-body smoothed particle hydrodynamics code, GADGET-3 (Springel 2005). Typical numbers of particles used for the simulations are listed in Table 1.

    In each of the galaxies, the gas comprises only a small fraction of the total galaxy mass. We therefore chose not to include the gaseous components of the galaxies in the simulations.

    You can check Table 1 for the number of particles in the simulation, but it was about 1 million for each of the Milky Way and M31 and about 100000 for M33. Also, don’t get confused by the term “hydrodynamics” in the quote above; Smoothed Particle Hydrodynamics is a Lagrangian Hydrodynamics formalism in which particles are used to represent a fluid. Consequently, in astrophysics it is sometimes used with the hydro turned off, in which case it simply becomes an N-body code. The advantage is it that it uses a TREE to calculate the gravitational forces between particles, which is considerably faster than a direct gravity calculation and allows one to use many more particles than if the gravity was being calculated directly for each particle pair.

  25. verytallguy says:

    Thanks ATTP, that’s pretty much what I’d assumed.

    I won’t ask what a TREE is…

  26. I won’t ask what a TREE is…

    I know you didn’t ask :-), but essentially it takes the simulation domain and divides it into a nested series of volumes, each of which is assigned a position and mass based on the particles that fall within it. Hence when you’re calculating the gravitational force on a particle due to all the others, you can treat a region with lots of particles that is far from the particle being considered as a single mass, while as you get closer to the particle being considered you use ever decreasing volumes until you’re doing direct gravity calculations for the particles very close to the particle being considered. Hence you do far fewer calculations than if you were calculating gravity directly for each particle pair. It goes as N logN, rather than N^2.

  27. Wonderful, thank you. Interesting how the increase of knowledge builds understanding (to state the blindingly obvious, but one could wish that were true elsewhere!).

  28. BBD says:

    Indeed, and where are all the odd bods telling ATTP that he’s wrong and the experts don’t know nothing and complaining about teh modulz?

  29. The Very Reverend Jebediah Hypotenuse says:


    …he’s wrong and the experts don’t know nothing…

    Pluto was once a planet. Now it’s not. Obviously, the science is not settled.

  30. Brandon Gates says:

    Anders, fascinating post, thanks for sharing your work with us.

    BBD, or bellyaching about what a waste of public funds pure science is.

  31. Brandon Gates says:

    PS,

    … [TREE] takes the simulation domain and divides it into a nested series of volumes, each of which is assigned a position and mass based on the particles that fall within it.

    Had to read that one twice before it clicked. Clever and elegant.

  32. Pingback: Three years! | …and Then There's Physics

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s