0:00:00.000,0:00:19.871
36c3 preroll music
0:00:19.871,0:00:22.920
Herald: Ok, I have to say, I'm always[br]deeply impressed about how much we already
0:00:22.920,0:00:31.430
learned about space, about the universe[br]and about our place in the universe,
0:00:31.430,0:00:37.230
our solar system. But the next speakers[br]will explain us how we can use
0:00:37.230,0:00:44.320
computational methods to simulate the[br]universe and actually grow planets. The
0:00:44.320,0:00:49.530
speakers will be Anna Penzlin (miosta).[br]She is PHC student in computational
0:00:49.530,0:00:55.230
astrophysics in Tübingen and Carolin[br]Kimmich (caro). She is a physics master's
0:00:55.230,0:01:02.630
student at Heidelberg University. And the[br]talk is entitled "Grow Your Own Planets
0:01:02.630,0:01:07.691
How Simulations Help us understand the[br]universe." Thank you!
0:01:07.691,0:01:14.743
applause
0:01:14.743,0:01:24.720
caro: So hi, everyone. It's a cool[br]animation right? And the really cool thing
0:01:24.720,0:01:28.729
is that there's actually physics going on[br]there. So this object could really be out
0:01:28.729,0:01:35.380
there in space but was created on a[br]computer. So this is how a star is
0:01:35.380,0:01:41.680
forming, how our solar system could have[br]looked like in the beginning. Thank you
0:01:41.680,0:01:47.440
for being here and that you're interested[br]in how we make such an animation. Anna and
0:01:47.440,0:01:54.060
I are researchers in astrophysics. And[br]we're concentrating on how planets form
0:01:54.060,0:01:58.789
and evolve. She's doing her PHD and in[br]Tübingen and I'm doing my masters in
0:01:58.789,0:02:04.030
Heidelberg. And in this talk, we want to[br]show you a little bit of physics and how
0:02:04.030,0:02:13.390
we can translate that in such a way that a[br]computer can calculate it. So, let's ask a
0:02:13.390,0:02:19.420
question first. What is the universe or[br]what's in the universe? The most part of
0:02:19.420,0:02:23.650
the universe is something we don't[br]understand, yet. It's dark matter and dark
0:02:23.650,0:02:28.680
energy and we don't know what it is, yet.[br]And that's everything we cannot see in
0:02:28.680,0:02:35.120
this picture here. What we can see are[br]stars and galaxies, and that's what we
0:02:35.120,0:02:39.980
want to concentrate on in this talk. But[br]if we can see it, why would we want to
0:02:39.980,0:02:48.590
watch a computer? Well, everything in[br]astronomy takes a long time. So each of
0:02:48.590,0:02:54.299
these tiny specs you see here are galaxies[br]just like ours. This is how the Milkyway
0:02:54.299,0:02:59.560
looks like. And we are living in this tiny[br]spot here. And as you all know, our earth
0:02:59.560,0:03:04.080
takes one year to orbit around the sun.[br]Now, think about how long it takes for the
0:03:04.080,0:03:10.459
sun to orbit around the center of the[br]galaxy. It's four hundred million years.
0:03:10.459,0:03:16.159
And even the star formation is 10 million[br]years. We cannot wait 10 million years to
0:03:16.159,0:03:23.520
watch how a star is forming, right? That's[br]why we need computational methods or
0:03:23.520,0:03:29.730
simulations on a computer to understand[br]these processes. So, when we watch to the
0:03:29.730,0:03:35.970
night sky, what do we see? Of course we[br]see stars and those beautiful nebulas.
0:03:35.970,0:03:42.319
They are a gas and dust. And all of these[br]images are taken with Hubble Space
0:03:42.319,0:03:51.060
Telescope. Oh, so there's one image that[br]does belong in there. But it looks very
0:03:51.060,0:03:56.810
similar, right? This gives us the idea[br]that we can describe the gases in the
0:03:56.810,0:04:04.840
universe as a fluid. It's really[br]complicated to describe the gas in every
0:04:04.840,0:04:10.069
single particle. So, we cannot track every[br]single molecule in the gas that moves
0:04:10.069,0:04:15.489
around. It's way easier to describe it as[br]a fluid. So remember that for later, we
0:04:15.489,0:04:22.310
will need that. But first, let's have a[br]look how stars form. A star forms from a
0:04:22.310,0:04:29.500
giant cloud of dust and gas. Everything[br]moves in that cloud. So, eventually more
0:04:29.500,0:04:38.699
dense regions occur and they get even[br]denser. And these clams can eventually
0:04:38.699,0:04:47.160
collapse to one star. So, this is how a[br]star forms. They collapse due to their own
0:04:47.160,0:04:53.810
gravity. And in this process, a disc[br]forms. And in this disc, planets can form.
0:04:53.810,0:04:59.710
So why a disc? As I said, everything moves[br]around in the cloud. So it's likely that
0:04:59.710,0:05:05.120
the cloud has a little bit of an initial[br]rotation. As it collapses, this rotation
0:05:05.120,0:05:11.840
gets larger and faster. And now you can[br]think of making a pizza. So when you make
0:05:11.840,0:05:16.870
a pizza and spin your dough on your[br]finger, you get a flat disc like a star,
0:05:16.870,0:05:25.110
like a disc around a star. That's the same[br]process, actually. In this disc, we have
0:05:25.110,0:05:31.330
dust and gas. From this dust in the disc[br]the planet can form. But how do we get
0:05:31.330,0:05:40.759
from tiny little dust particles to a big[br]planet? Well, it somehow has to grow and
0:05:40.759,0:05:46.320
grow even further and compact until we[br]have rocks. And even grow further until we
0:05:46.320,0:05:53.390
reach planets. How does it grow? Well,[br]that dust grows we know that. At least
0:05:53.390,0:06:00.720
that's what I observed when I took those[br]images in my flat. Well, so dust can grow
0:06:00.720,0:06:06.680
and grow even further and compact. But[br]when you take two rocks, we're now at this
0:06:06.680,0:06:11.729
in this stage. When you take two rocks and[br]throw them together, you don't expect them
0:06:11.729,0:06:21.060
to stick, right? You expect them to crash[br]and crack into a thousand pieces. So,
0:06:21.060,0:06:28.300
we're standing on the proof that planets[br]exist. How does this happen? And it's not
0:06:28.300,0:06:34.699
quite solved yet in research. So, this is[br]a process that is really hard to observe
0:06:34.699,0:06:39.379
because planets are very, very tiny[br]compared to stars. And even stars are only
0:06:39.379,0:06:45.380
small dots in the night sky. Also, as I[br]said, planets form in a disc. And it's
0:06:45.380,0:06:52.389
hard to look inside the disc. So this is[br]why we need computation to understand a
0:06:52.389,0:06:58.670
process that how planets form and other[br]astronomical processes. So let's have a
0:06:58.670,0:07:09.530
look at how this simulated on a computer.[br]miosta: OK. So, somehow we have seen
0:07:09.530,0:07:15.949
nature. It's beautiful and it's just like[br]a tank of water and a bubbly fluid we
0:07:15.949,0:07:21.000
already have. So, now we have this bubbly[br]fluid and here in the middle demonstrated.
0:07:21.000,0:07:25.819
But now we have to teach our computer to[br]deal with the bubbly fluid. And that's way
0:07:25.819,0:07:31.601
too much single molecules to simulate[br]them, as we already said. So there are two
0:07:31.601,0:07:37.759
ways to discretize it in a way that we[br]just look at smaller pieces. One is the
0:07:37.759,0:07:47.080
Lagrangian description, just like taking[br]small bubbles or balls of material that
0:07:47.080,0:07:52.370
have a fixed mass. They have a certain[br]velocity that varies between each particle
0:07:52.370,0:07:57.180
and they have, of course, a momentum[br]because they have a velocity and a mass.
0:07:57.180,0:08:01.629
And we've created a number of those[br]particles and then just see how they move
0:08:01.629,0:08:08.260
around and how they collide with each[br]other. That would be one way. And that was
0:08:08.260,0:08:12.639
described last year in a very good talk. I[br]can highly recommend to hear this talk if
0:08:12.639,0:08:18.099
you're interested in this method. However,[br]there's a second way to also describe
0:08:18.099,0:08:23.169
this. Not just going with the flow of the[br]particles, but we are a bit lazy, we just
0:08:23.169,0:08:30.039
box it. So we create a grid. And as you[br]see down here in this grid, you have the
0:08:30.039,0:08:38.789
certain filling level, a bit of a slope.[br]So, what's the trend there? And then we
0:08:38.789,0:08:44.910
just look for each box, what flows in what[br]flows out through the surfaces of this
0:08:44.910,0:08:51.280
box. And then we have a volume or a mass[br]filled within this box. And this is how we
0:08:51.280,0:08:57.230
discretize what is going on in the disc.[br]And actually, since we are usually in the
0:08:57.230,0:09:04.220
system of a disc, we do not do it in this[br]nice box way like this. But we use boxes
0:09:04.220,0:09:09.710
like those because they are already almost[br]like a disc and we just keep exactly the
0:09:09.710,0:09:14.780
same boxes all the time and you just[br]measure what goes through the surface in
0:09:14.780,0:09:22.890
these boxes. So, this is how these two[br]methods look like if you compute with both
0:09:22.890,0:09:30.030
of them. So, one was done by me. I'm[br]usually using this boxing method and the
0:09:30.030,0:09:35.960
other was done by my colleague. You see[br]this like when you look at them, at the
0:09:35.960,0:09:40.490
colors, at the structure here, you have[br]the slope inwards, you have the same slope
0:09:40.490,0:09:46.600
inwards here. You have even this silly[br]structure here. The same here. But what
0:09:46.600,0:09:52.640
you notice is you have this enlarge dots[br]that are really the mass particles we saw
0:09:52.640,0:09:57.980
before, these bubbles. And here you have[br]this inner cutout. This is because when
0:09:57.980,0:10:05.450
you create this grid, you have the very[br]region at the inner part of the disc where
0:10:05.450,0:10:11.410
the boxes become tiny and tinier. And[br]well, we can't compute that. So, we have
0:10:11.410,0:10:19.130
to cut out at some point in inner part So, here[br]when you go to low densities, these
0:10:19.130,0:10:24.580
bubbles blow up and distribute their mass[br]over a larger area. So, it's not very
0:10:24.580,0:10:30.840
accurate for these areas. And here we have[br]the problem we can't calculate the inner
0:10:30.840,0:10:38.590
area. So both methods have their pros and[br]cons. And are valid. But now, for most we
0:10:38.590,0:10:51.460
will focus on this one. Just so we have[br]this nice stream features. So, again,
0:10:51.460,0:11:00.150
going back to the boxes, we have to[br]measure the flow between the boxes. This
0:11:00.150,0:11:06.580
flow, in physics we call it flux, and we[br]have a density row one, density row too.
0:11:06.580,0:11:12.280
And the flux is the description of what[br]mass moves through the surface here from
0:11:12.280,0:11:22.100
one box to the next. So, if we write this[br]in math terms, it looks like this. This
0:11:22.100,0:11:36.560
says the time derivative of the density,[br]meaning the change over time. So how much
0:11:36.560,0:11:43.540
faster or slower, the velocity would be a[br]change in time. And then this weird
0:11:43.540,0:11:50.440
triangle symbol it's called nabla is a[br]positional derivative. So, it's like a
0:11:50.440,0:12:00.620
slope. So, how do we change our position,[br]actually. So, if we change, look at the
0:12:00.620,0:12:10.200
density over time, it should correlate to[br]what inflow we have over position. That is
0:12:10.200,0:12:16.100
what that says. So and then we have in[br]physics a few principles that we have
0:12:16.100,0:12:21.920
always to obey because that is just almost[br]common sense. One of them is, well, if we
0:12:21.920,0:12:29.760
have mass in a box. Well, like this, the[br]mass should not go anywhere unless someone
0:12:29.760,0:12:35.200
takes it out. So, if we have a closed box[br]and mass in that box, nothing should
0:12:35.200,0:12:42.620
disappear magically. It should all stay in[br]this box. So, even if these particles jump
0:12:42.620,0:12:48.050
around in our box with a certain velocity,[br]it's the same number of particles in the
0:12:48.050,0:12:57.870
end. That's again, the same equation just[br]told in math. So, a second very
0:12:57.870,0:13:04.220
rudimentary principle is if we have energy[br]in it, in a completely closed box. So, for
0:13:04.220,0:13:10.480
example, this nice chemicals here and we[br]have a certain temperature. So, in this
0:13:10.480,0:13:18.150
case, our temperature is low, maybe like[br]outside of around zero degree Celsius. And
0:13:18.150,0:13:24.070
then we have this nice chemicals down here[br]and at some point they react very heavily.
0:13:24.070,0:13:30.770
We suddenly end up with much less chemical[br]energy and a lot more thermal energy. But
0:13:30.770,0:13:36.940
overall, the complete energy summed up[br]here, like the thermal and the chemical
0:13:36.940,0:13:47.290
energy, also the energy of the movement[br]and the energy of potential added up to
0:13:47.290,0:13:53.720
this variable "U". That should not change[br]over time if you sum up everything.
0:13:53.720,0:13:59.500
Because our energy is conserved within our[br]clothed box. And then the third thing is I
0:13:59.500,0:14:09.860
think you all know this. If you have like[br]a small mass with a certain velocity, a
0:14:09.860,0:14:14.181
very high velocity in this case and it[br]bumps into someone very large, what
0:14:14.181,0:14:21.330
happens? Well, you get a very small[br]velocity in this large body and the
0:14:21.330,0:14:28.260
smaller mass stops. And the principle here[br]is that momentum is conserved, meaning
0:14:28.260,0:14:36.680
that the velocity times the mass of one[br]object is the same as then later for the
0:14:36.680,0:14:42.700
other one. But since it's larger, this[br]product has to be the same. That doesn't
0:14:42.700,0:14:49.381
change. And we have also in our[br]simulations to obey these rules and we
0:14:49.381,0:14:54.730
have to code that in so that we have[br]physics in them. So you say, ok, this is
0:14:54.730,0:14:59.450
really simple, these rules, right? But[br]actually, well, it's not quite as simple.
0:14:59.450,0:15:03.880
So, this is the Navier-Stokes equation, a[br]very complicated equation is not
0:15:03.880,0:15:10.550
completely solved. And we have here all[br]that is marked red are derivatives. Here
0:15:10.550,0:15:16.230
we have our conservation law that was the[br]nice and simple part. But now we have to
0:15:16.230,0:15:25.700
take other physical things into accounting[br]for pressure, accounting for viscosity,
0:15:25.700,0:15:33.370
for compression. So squeezing. And like[br]how sticky is our fluid? And also gravity.
0:15:33.370,0:15:38.790
So, we have a lot of additional factors,[br]additional physics we also have to get in
0:15:38.790,0:15:45.470
somehow. And all of these also depend[br]somehow on the change of position or the
0:15:45.470,0:15:51.850
change of time. And these derivatives[br]aren't really nice for our computers
0:15:51.850,0:15:57.370
because they well, they don't understand[br]this triangle. So, we need to find a way
0:15:57.370,0:16:03.920
to write an algorithm so that it can[br]somehow relate with these math formula in
0:16:03.920,0:16:14.920
a way that the computer likes. And one of[br]the way to do this is, well, the simplest
0:16:14.920,0:16:24.690
solution actually is just we say, OK, we[br]have now this nasty derivatives and we
0:16:24.690,0:16:32.160
want to get rid of them. So, if we look[br]just at one box now and we say that in
0:16:32.160,0:16:42.170
this box, the new value for the density in[br]this box would be the previous density,
0:16:42.170,0:16:49.590
plus the flux in and out times the time[br]stepover which we measure this flux,
0:16:49.590,0:16:58.260
right? So, and we have to somehow get to[br]this flux and we just say, OK, this flux
0:16:58.260,0:17:06.220
now is if we start here and the slope of[br]this curve, the trends so to say, where
0:17:06.220,0:17:10.550
this curve is going right now, it would[br]look like this. So, in our next step, time
0:17:10.550,0:17:19.140
step, we would have a density down here.[br]And well, then we do this again. We again
0:17:19.140,0:17:25.620
look at this point, where's the trend[br]going, where's the line going? And then we
0:17:25.620,0:17:36.539
end up here. Same here. So, again, we just[br]try to find this flax and this is the
0:17:36.539,0:17:43.130
trend at this position in time. So, this[br]goes up here. And then if we are here now,
0:17:43.130,0:17:48.399
look at this point, it should go up here.[br]So this is what our next trend would be.
0:17:48.399,0:17:55.269
And we do this over all the times. And[br]this is how our simulation then would
0:17:55.269,0:18:02.929
calculate the density for one box over a[br]different time steps. So, that kind of
0:18:02.929,0:18:09.250
works. So, the blue curve is the[br]analytical one, the red curve, well it
0:18:09.250,0:18:17.740
kind of similar, it works. But can we do[br]better? It's not perfect, yet, right? So,
0:18:17.740,0:18:23.259
what we can do is we refine this a bit,[br]taking a few more steps, making it a bit
0:18:23.259,0:18:31.000
more computationally heavy, but trying to[br]get a better resolution. So, first we
0:18:31.000,0:18:36.309
start with the same thing as before. We go[br]to this point, find the trend in this
0:18:36.309,0:18:43.689
point. That point like the line would go[br]in this direction from this point. And
0:18:43.689,0:18:51.529
then we go just half a step now. Sorry![br]And now we look at this half a step to
0:18:51.529,0:18:57.650
this point now. And again, the same[br]saying, OK, where's the trend going now?
0:18:57.650,0:19:07.539
And then we take where this point would go[br]and added to this trend. So that would be
0:19:07.539,0:19:14.179
that. The average of this trend, of this[br]exact point and this trend, this dark
0:19:14.179,0:19:19.360
orange curve. And then we go back to the[br]beginning with this trend now and say this
0:19:19.360,0:19:24.260
is a better trend than the one we had[br]before. We now use that and go again and
0:19:24.260,0:19:34.700
search the point for half a time step. And[br]then again, we do the same thing. Now we
0:19:34.700,0:19:42.459
again try to find actually the trend and[br]average it with the arrow before. So it's
0:19:42.459,0:19:46.321
not exactly the trend. It's a bit below[br]the trend because we averaged it with the
0:19:46.321,0:19:51.880
arrow before. And now we take this[br]averaging trend from the beginning to the
0:19:51.880,0:19:57.080
top like this. Okay. This is already quite[br]good, but we can still do a little bit
0:19:57.080,0:20:02.570
better if we averaged with our ending[br]point. So, we go here, look, where is the
0:20:02.570,0:20:10.740
trend going that would go quite up like[br]this and we average this and this together
0:20:10.740,0:20:15.110
and then we end up with a line like this.[br]This is so much better than what we had
0:20:15.110,0:20:22.920
before. It's a bit more complicated, to be[br]fair. But actually it's almost on the
0:20:22.920,0:20:29.059
line. So, this is what we wanted. So, if[br]you compare both of them, we have here our
0:20:29.059,0:20:34.690
analytical curve. So, over time in one[br]box, this is how the densities should
0:20:34.690,0:20:39.909
increase. And now with it both of the[br]numerical method, the difference looks
0:20:39.909,0:20:46.050
like this. So, if we have smaller and[br]smaller time steps, even the Euler gets
0:20:46.050,0:20:55.749
closer and closer to the curve. But[br]actually the Runge Kutta this four step process
0:20:55.749,0:21:00.620
works much better and much faster.[br]However, it's a bit more computationally
0:21:00.620,0:21:08.370
and difficult.[br]caro: When we simulate objects in
0:21:08.370,0:21:15.039
astronomy, we always want to compare that[br]to objects that are really out there. So,
0:21:15.039,0:21:20.489
this is a giant telescope, well consisting[br]of a lot of small telescopes. But they can
0:21:20.489,0:21:27.010
be connected and used as a giant telescope[br]and it takes photos of dust in the sky.
0:21:27.010,0:21:34.159
And this is used to take images of discs[br]around stars. And these discs look like
0:21:34.159,0:21:41.049
this. So, these images were taken last[br]year and they are really cool. Before we
0:21:41.049,0:21:46.121
had those images, we only had images with[br]less resolution. So, they were just
0:21:46.121,0:21:52.120
blurred blobs. And we could say, yeah,[br]that might be a disc. But now we really
0:21:52.120,0:21:58.659
see the discs and we see rings here, thin[br]rings and we see thicker rings over here.
0:21:58.659,0:22:05.590
And even some spiraly structures here. And[br]also some features that are not really
0:22:05.590,0:22:11.990
radial symmetric like this arc here. And[br]it's not completely solved how these
0:22:11.990,0:22:24.259
structures formed. And to find that out a[br]colleague of mine took this little object
0:22:24.259,0:22:30.799
with the asymmetry here. And so, this is[br]image we just saw. And this is his
0:22:30.799,0:22:37.590
simulation. So, this is how the disc[br]looked like in the beginning, probably.
0:22:37.590,0:22:43.980
And we put in three planets and let the[br]simulation run. And so, what we see here
0:22:43.980,0:22:52.029
is that the star is cut out as Anna said.[br]So, the grid cells in the inner part are
0:22:52.029,0:22:56.690
very, very small. And it would take a long[br]time to compute them all. So, that's why
0:22:56.690,0:23:06.779
we're leaving out that spot in the middle.[br]And what we see here is three planets
0:23:06.779,0:23:16.309
interacting with the material in the disc.[br]And we can see that these planets can make
0:23:16.309,0:23:24.440
this thing here appear so that in the end[br]we have something looking very similar to
0:23:24.440,0:23:30.700
what we want to have or what we really[br]observe. So, we can say three planets
0:23:30.700,0:23:37.379
could explain how these structures formed[br]in this disc. It's a little bit
0:23:37.379,0:23:42.409
elliptical, you see that. That's because[br]it's tilted from our side of line. It
0:23:42.409,0:23:47.430
would be round if you watched at it face[br]on. But it's a little bit tilted. That's
0:23:47.430,0:23:55.269
why it looks elliptical.[br]miosta: So, we already saw we can put
0:23:55.269,0:24:02.080
planets in the gas and then we create[br]structures. One very exciting thing that
0:24:02.080,0:24:08.740
we found in the last year - or two years[br]ago it started but then we found more - is
0:24:08.740,0:24:15.690
this system PDS 70. In this system, for[br]the very first time, we found a planet
0:24:15.690,0:24:24.249
that was still embedded completely within[br]the disc. So, the gas and dust. Usually,
0:24:24.249,0:24:32.259
because the gas and dust is the main thing[br]that creates this signal of some radiation
0:24:32.259,0:24:37.749
because of heat. We only observe that and[br]then we can't observe the planet embedded.
0:24:37.749,0:24:41.629
But in this case, the planet was large[br]enough. And in the right position that we
0:24:41.629,0:24:48.940
actually were able to observe some[br]signature of accretion on this planet that
0:24:48.940,0:24:57.440
was brighter than the rest of the disc.[br]And then later, just this year, just a few
0:24:57.440,0:25:03.739
months ago, we actually found out well,[br]this is not the only object here. This is
0:25:03.739,0:25:10.850
very clearly a planet. But actually,[br]like this spot here is also something. So,
0:25:10.850,0:25:17.299
we can see it in different grains. Every[br]picture here is a different set of grains
0:25:17.299,0:25:26.950
observed. And we can see [br]this in five different kinds of
0:25:26.950,0:25:32.799
observations. So, there is a planet here.[br]And then there is also something we don't
0:25:32.799,0:25:37.710
know what it is yet, but its point like[br]and actually creates the feature that we
0:25:37.710,0:25:43.240
reproduce in different kinds of[br]observational bands or different kinds of
0:25:43.240,0:25:52.070
signals here. This is very interesting.[br]For the first time, we actually see a
0:25:52.070,0:25:58.030
planet forming right now within the disc.[br]And so a colleague of mine also is very
0:25:58.030,0:26:04.929
interested in the system and started to[br]simulate how do two planets in a disc
0:26:04.929,0:26:13.149
change the dynamics of a disc? So here we[br]have, of course, this disc is again tilted
0:26:13.149,0:26:20.230
because it's not phase on, it's like 45[br]degrees tilted, not like this, but like
0:26:20.230,0:26:27.289
this. And so he had it face on. This is[br]what a simulation looks like. So, there
0:26:27.289,0:26:33.880
are two planets: these blobs here, again,[br]as in this simulation. Here we have a
0:26:33.880,0:26:39.289
close up. You can actually see this little[br]boxes are actually our simulation boxes in
0:26:39.289,0:26:47.429
which we have our own densities. And then[br]he just looked at how the planets would
0:26:47.429,0:26:52.620
change the structure and the gas and also[br]how the gas would interact with the
0:26:52.620,0:26:59.249
planets, shifting them around. And it's[br]interesting. So, the planets tend to clear
0:26:59.249,0:27:05.259
out an area, open a gap, and within the[br]disk, that block has a lot of gas around
0:27:05.259,0:27:11.039
here. So, you have the brighter ring here[br]again and then clearing out more and more.
0:27:11.039,0:27:23.390
And at some point in the simulation you[br]saw they get a bit jumpy. So it's very nice.
0:27:23.390,0:27:29.570
You also see that planets induce in the[br]whole disc some kind of features like
0:27:29.570,0:27:36.740
spiral features. And so a single planet[br]will change the symmetry and the
0:27:36.740,0:27:40.989
appearance of a whole disc.[br]caro: So, the reason why the planet is
0:27:40.989,0:27:46.489
staying at this point is because we're[br]rotating with the planet. So it's actually
0:27:46.489,0:27:53.499
going around the disc, but the like camera[br]is rotating with the planet. So, it's
0:27:53.499,0:28:00.360
staying at that fixed place we put it in.[br]miosta: Exactly. But there's more because
0:28:00.360,0:28:04.600
as I already said, in the Navier-Stokes[br]equation, we have a lot of different kinds
0:28:04.600,0:28:08.970
of physics that we all have to include in[br]our simulations. One of the things, of
0:28:08.970,0:28:15.149
course, is we maybe don't have just a star[br]and a disc. We have planets in there and
0:28:15.149,0:28:20.600
maybe two stars in there. And all of these[br]larger bodies have also an interaction
0:28:20.600,0:28:27.360
between each other. So, if we have the[br]star, every planet will have an
0:28:27.360,0:28:32.600
interaction with the star, of course. But[br]then also the planets between each other,
0:28:32.600,0:28:40.381
they have also an interaction, right? So,[br]in the end, you have to take into account
0:28:40.381,0:28:48.820
all of these interactions. And then also[br]we have accretion just looking like this.
0:28:48.820,0:28:59.350
So, accretion means that the gas is bound[br]by some objects. It can be the disc, the
0:28:59.350,0:29:07.009
planet or the star that takes up the mass,[br]the dust or the gas and bounce it to this
0:29:07.009,0:29:14.840
object. And then it's lost to the disc or[br]the other structures because it's
0:29:14.840,0:29:22.309
completely bound to that. So, the[br]principle of this would be the simulation
0:29:22.309,0:29:29.279
I did last year and published, we have[br]here a binary star. So, these two dots are
0:29:29.279,0:29:38.809
stars. I kind of kept them in the same[br]spot. But every picture will be one orbit
0:29:38.809,0:29:42.759
of this binary, but since we have[br]interactions, you actually see them
0:29:42.759,0:29:48.539
rotating because of the interactions, with[br]each other. And then also we have here a
0:29:48.539,0:29:52.669
planet and here a planet. And the[br]interesting thing was that these two
0:29:52.669,0:30:00.361
planets interact in such a way that they[br]end up on exactly the same orbit. So, one
0:30:00.361,0:30:06.179
star's further out, the orange one, and then[br]very fast they go in. And they end up on
0:30:06.179,0:30:28.419
exactly the same orbit. If it now play nicely. [br]So, another thing is with the accretion here,
0:30:28.419,0:30:36.600
we actually see clouds from above dropping[br]down onto the new forming star here. So,
0:30:36.600,0:30:44.409
all of this, what you see here would be[br]gas, hydrogen. And it's a very early phase
0:30:44.409,0:30:49.499
so that disc is not completely flat. It[br]has a lot of material. And then we
0:30:49.499,0:30:55.779
actually have this infall from above[br]towards the star and then the star keeps
0:30:55.779,0:31:01.767
the mass. And we have to take this also[br]into account in our simulations. Another
0:31:01.767,0:31:07.220
thing we have to take into account up till[br]now, we just cared about masses and
0:31:07.220,0:31:12.929
densities. But of course what we actually[br]see is that stars are kind of warm,
0:31:12.929,0:31:21.759
hopefully. Otherwise, temperatures here[br]would also not be nice. And different
0:31:21.759,0:31:27.929
chemicals have different condensation[br]points. And this is also true in a system.
0:31:27.929,0:31:35.019
So, we start with the start temperature at[br]the surface of the star. We have a
0:31:35.019,0:31:41.479
temperature around 4.000 Kelvin. And then[br]we go a bit into the disc. And there is a
0:31:41.479,0:31:47.889
point where we for the first time reach a[br]point where we have any material at all.
0:31:47.889,0:31:52.169
Because it starts to condensate and we[br]actually have something solid like iron.
0:31:52.169,0:31:58.179
For example, at a 1500 Kelvin. And then if[br]we go further in, we reach a point where
0:31:58.179,0:32:07.690
we have solid water and this is at 200[br]Kelvin. This is what we then would need
0:32:07.690,0:32:12.590
actually to have a planet that also has[br]water on it. Because if we don't get the
0:32:12.590,0:32:18.889
water in the solid state, it will not fall[br]onto a terrestrial planet and be bound
0:32:18.889,0:32:24.899
there, right? So, this is important for[br]our Earth, actually. And then if we go
0:32:24.899,0:32:33.340
even further out, we have also other gases[br]condensating to solids like CO2 or methane
0:32:33.340,0:32:40.591
or things like that. And since we only get[br]water on a planet when we have a
0:32:40.591,0:32:48.009
temperature that is low enough so that the[br]water actually forms is solid and it's
0:32:48.009,0:32:54.269
important for us to think about where that[br]is in our forming disc. Where do we start?
0:32:54.269,0:32:59.769
We have a planet like Earth that could[br]have some water, right? But it's not just
0:32:59.769,0:33:07.570
the simple picture, where we have all these[br]nice ring structures, where we have a clear
0:33:07.570,0:33:13.619
line. Actually, it gets more complicated[br]because we have pressure and shocks. And
0:33:13.619,0:33:19.539
thermodynamics is a lot like pogo dancing,[br]right? You crash into each other. And it's
0:33:19.539,0:33:25.629
all about collisions. So, the gas[br]temperature is determined by the speed of
0:33:25.629,0:33:31.299
your gas molecules. Like you bouncing and[br]crashing into each other, exchanging
0:33:31.299,0:33:39.340
momentum. So, there's two ways to heat up[br]such dance. First thing is you get a large
0:33:39.340,0:33:45.944
amount of velocity from the outside like a[br]huge kick, a shock into your system. A
0:33:45.944,0:33:51.519
second way would be if we have a higher[br]pressure, like more molecules, then also
0:33:51.519,0:33:55.909
you of course have more collisions and[br]then a higher temperature. So, if you
0:33:55.909,0:34:02.529
change - because you have a planet now in[br]the system - the pressure at some point,
0:34:02.529,0:34:08.700
you actually get a higher temperature. So,[br]that is not then we lose this nice line
0:34:08.700,0:34:19.136
because suddenly we have different[br]pressures at different locations. And a
0:34:19.136,0:34:24.700
colleague of mine also simulated this. [br]So, this is the initial condition we
0:34:24.700,0:34:28.860
just assumed: OK, if we have no[br]disturbance whatsoever, we have our nice
0:34:28.860,0:34:36.890
planet here at 1au. So, same distance as[br]earth to the sun. Here, too. But here we
0:34:36.890,0:34:46.670
assume that less heat gets transferred[br]from the surface of the disc. And here we
0:34:46.670,0:34:52.030
have the planet far out like Jupiter or[br]something. And now we actually let this
0:34:52.030,0:34:59.590
planet change the structure of the disc.[br]And what happens is - we found these spirals
0:34:59.590,0:35:05.800
and within these spirals, we change[br]pressure. And with this actually, if you
0:35:05.800,0:35:11.590
see this orange, everywhere where it's[br]orange it's hotter than the iceline. So,
0:35:11.590,0:35:17.020
we don't have water where it's orange. And[br]where it's blue we can have water. And the
0:35:17.020,0:35:22.350
interesting thing is, even if we put a[br]planet out here like Jupiter, we still
0:35:22.350,0:35:32.569
form these regions in the inner system[br]where we have less water.
0:35:32.569,0:35:38.022
caro: One problem in astrophysical[br]simulations is that we don't always know
0:35:38.022,0:35:47.940
how to shape our boxes or how small these[br]boxes have to be. So, we use a trick to
0:35:47.940,0:35:54.670
reshape the boxes as we need them. It's[br]called adaptive mesh. And this is a
0:35:54.670,0:35:58.890
simulation of the red fluid flowing in[br]this direction and the blue fluid in the
0:35:58.890,0:36:06.581
other one. So, at the boundary, the two[br]fluid shear and they mix up somehow and we
0:36:06.581,0:36:12.990
don't know how in advance. So, we start a[br]simulation and as the simulation starts,
0:36:12.990,0:36:19.640
we reshape those boxes here. So, in the[br]middle we don't need much. We reshape
0:36:19.640,0:36:25.400
because it's not that complicated here.[br]It's just the flow. But at the boundary we
0:36:25.400,0:36:35.060
see those mixing up of the two fluids. And[br]so, we reshape the cells as we need them.
0:36:35.060,0:36:44.760
This is done in a program, in an[br]astrophysical program called AREPO. We
0:36:44.760,0:36:52.750
will later show you some more programs to[br]use for simulations. But another
0:36:52.750,0:36:59.020
simulation I want to show you first is[br]also done with AREPO and it's a simulation
0:36:59.020,0:37:04.710
of the universe. So, from here to here,[br]it's very big. It's 30 million light
0:37:04.710,0:37:12.210
years. So each of these dots you see here[br]is the size of a galaxy or even more. And
0:37:12.210,0:37:17.840
here you can actually see that at some[br]regions it's very empty. So, we're
0:37:17.840,0:37:23.420
rotating around this universe, this[br]simulated universe here. And these regions
0:37:23.420,0:37:28.990
here are empty. And we don't need a lot of[br]boxes there. The big boxes are enough
0:37:28.990,0:37:35.010
here. But in this dense regions where we[br]have a lot of material, we need smaller
0:37:35.010,0:37:42.380
boxes. And this method I showed you where[br]we reshape the boxes as we need them is
0:37:42.380,0:37:53.420
used for this simulation.[br]miosta: So, actually, you see the
0:37:53.420,0:37:56.340
beginning of the universe there.[br]caro: Yes!
0:37:56.340,0:38:01.000
miosta: Basically, the initial mass[br]collapsing to the first galaxies and first
0:38:01.000,0:38:07.030
supernovae starting. Very beautiful[br]simulation.
0:38:07.030,0:38:19.820
caro: So, there are different programs, as[br]I already mentioned, in astrophysics.
0:38:19.820,0:38:24.970
Three of them, those three are all open[br]source, so you can download them and use
0:38:24.970,0:38:31.090
them on your own machine, if you like. But[br]there are more, a lot more. Some of them
0:38:31.090,0:38:38.630
are open source, some of them are not.[br]Sometimes it's hard to get them. In the
0:38:38.630,0:38:43.700
following, we will present the tool[br]FARGO3D and PLUTO in a detailed version or
0:38:43.700,0:38:53.160
a more detailed vision than AREPO [br]because we usually use those two for our
0:38:53.160,0:38:58.380
simulations. What I want to show you with[br]this slide is that depending on what you
0:38:58.380,0:39:04.520
want to simulate, you need to choose a[br]different program. And one thing is that
0:39:04.520,0:39:10.250
in astrophysics we sometimes call the[br]whole program code. So, if I use the word
0:39:10.250,0:39:19.170
code. Sorry about that. I mean, the whole[br]program. So, let's have a look at FARGO3D.
0:39:19.170,0:39:27.870
It's a hydro dynamics code and what you[br]see here is an input parameter file. There
0:39:27.870,0:39:35.180
you define how the disc looks like. How[br]much mass does it have? How big is it? And
0:39:35.180,0:39:43.140
what planet? So, here at Jupiter, do you[br]see that? Jupiter is put in. And we also
0:39:43.140,0:39:51.280
define how big our boxes are. This[br]program is written in C, which is quite
0:39:51.280,0:39:57.500
nice because a lot of astrophysical[br]programs are still written in Fortran. So,
0:39:57.500,0:40:05.600
this is good for me because I don't know[br]any Fortran. We can run this and what's
0:40:05.600,0:40:11.010
typical for FARGO3D. So that's a compilation[br]actually on my computer. So, I don't need
0:40:11.010,0:40:18.840
a fancy computer. I just did it on my[br]small laptop and now we run it. Now,
0:40:18.840,0:40:26.130
typical for FARGO3D, as you will see are lot[br]of dots. So, here it will print out a lot
0:40:26.130,0:40:33.810
of dots and it will create at certain[br]times some outputs. And these outputs are
0:40:33.810,0:40:38.300
huge files containing numbers. So, if you[br]look at them they are not really
0:40:38.300,0:40:44.290
interesting. They just are a numbers in[br]something like a text file. So, a big part
0:40:44.290,0:40:50.430
of astrophysics is also to visualize the[br]data. Not only to create it but also to
0:40:50.430,0:40:57.080
make images so that we can make movies out[br]of them. For that, I prefer to use Python
0:40:57.080,0:41:01.600
but there are a lot of tools Python[br]Matplotlib, but there are a lot of
0:41:01.600,0:41:09.290
different tools to visualize the data. So,[br]this is actually that output. That first
0:41:09.290,0:41:16.350
one we just saw. The Jupiter planet in the[br]disc that I defined in this parameter file
0:41:16.350,0:41:23.280
and it's already started to do some[br]spirals. And if I would have let it
0:41:23.280,0:41:33.680
run further than the spirals were more[br]prominent. And yeah, now we have a planet
0:41:33.680,0:41:45.230
here on our computer.[br]miosta: OK, so we also have PLUTO. PLUTO
0:41:45.230,0:41:53.590
somehow has a bit more setup files. So,[br]what I need is three files here. Looks a
0:41:53.590,0:41:59.320
bit complicated to break it down. This[br]file defines my grid and initial values.
0:41:59.320,0:42:04.770
And this simulation time here we input[br]actually what physics do we want to need?
0:42:04.770,0:42:13.020
What is our coordinate system? So, do we[br]want to have a disc or just like spherical
0:42:13.020,0:42:20.660
boxes or like squared boxes? And how is[br]the time defined? And here we then
0:42:20.660,0:42:26.720
actually write a bit of code to say, OK,[br]now how do I want a gravitational
0:42:26.720,0:42:34.580
potential? So, what's the source of[br]gravity or what will happen at the inner
0:42:34.580,0:42:39.890
region where we have this dark spot? We[br]have somehow to define what happens if gas
0:42:39.890,0:42:45.090
reaches this boundary. Is it just falling[br]in? Is it bouncing back or something? Or
0:42:45.090,0:42:50.890
is it looping through the one end to the[br]next? This is also something we then just
0:42:50.890,0:43:01.530
have to code in. And if we then make it[br]and let run, it looks like this. So,
0:43:01.530,0:43:08.590
again, our nice thing we hopefully put in[br]or wanted to put in: the time steps, what
0:43:08.590,0:43:14.890
our boundaries were, parameters of[br]physics. Hopefully, the right ones and
0:43:14.890,0:43:21.396
then nicely we start with our time steps[br]and then we see this. It's hooray! It
0:43:21.396,0:43:27.240
worked actually. Because it's actually not[br]quite simple usually to set up a running
0:43:27.240,0:43:32.000
program. A running problem, because you[br]have to really think about what should be
0:43:32.000,0:43:38.170
the physics. What's the scale of your[br]problem? What's the timescale of your
0:43:38.170,0:43:44.990
problem? And specify this in a good way.[br]But in principle, this is how it works.
0:43:44.990,0:43:49.320
There are few test problems if you[br]actually want to play around with this to
0:43:49.320,0:43:56.390
make it easy for the beginning. And this[br]is how we do simulations. So, as I already
0:43:56.390,0:44:02.320
set, we can just start them on our[br]laptops. So, here this is my laptop. I
0:44:02.320,0:44:07.859
just type a dot slash FARGO3D and that[br]should run, right? And then I just wait
0:44:07.859,0:44:16.450
for ten years to finish the simulations of[br]500 timesteps or outputs. Well, that's not the best
0:44:16.450,0:44:27.660
idea. So, we need more power. And both of[br]us, for example, are using a cluster for
0:44:27.660,0:44:36.880
Baden-Württemberg and that takes down our[br]computation time by a lot. Usually, like a
0:44:36.880,0:44:45.050
factor of maybe 20, which is a lot. So, I[br]would need on my computer maybe a year and
0:44:45.050,0:44:53.040
then I just need maybe 5 hours, a few days[br]or a week on this cluster, which is
0:44:53.040,0:44:56.380
usually the simulation time about a week[br]for me.
0:44:56.380,0:45:04.440
caro: So, what you see here is that we use[br]GPUs, yes. But we do not or mostly not use
0:45:04.440,0:45:09.630
them for gaming. We use them for actually[br]actual science. Yeah, would be nice to
0:45:09.630,0:45:20.614
play on that, right? That just said![br]miosta: So, back to our Earth, actually.
0:45:20.614,0:45:27.670
So, can we now? We wanted to grow our own[br]planet. We can do that, yes of course. Can
0:45:27.670,0:45:31.600
we grow Earth? Well, Earth is a very[br]special planet. We have a very nice
0:45:31.600,0:45:37.720
temperature here, right? And we have not a[br]crushing atmosphere like Jupiter, like a
0:45:37.720,0:45:43.440
huge planet that we could not live under.[br]We have a magnetic field that shields us
0:45:43.440,0:45:53.760
from the radiation from space and we have[br]water. But just enough water so that we
0:45:53.760,0:46:00.170
still have land on this planet where we[br]can live on. So, even if we fine tune
0:46:00.170,0:46:05.230
simulations, the probability that we[br]actually hit Earth and have all the
0:46:05.230,0:46:12.800
parameters right is actually tiny. It's[br]not that easy to simulate an Earth. And
0:46:12.800,0:46:17.320
there are a lot of open questions, too.[br]How did we actually manage to get just
0:46:17.320,0:46:24.240
this sip of water on our surface? How did[br]we manage to collide enough mass or
0:46:24.240,0:46:30.060
aggregate enough mass to form this[br]terrestrial planet without Jupiter is
0:46:30.060,0:46:35.740
sweeping up all the mass in our system?[br]How could we be stable in this orbit when
0:46:35.740,0:46:42.660
there are seven other planets swirling[br]around and interacting with us? All of
0:46:42.660,0:46:48.660
this is open in our field of research[br]actually, and not completely understood.
0:46:48.660,0:46:54.620
This is the reason why we still need to [br]do astrophysics and even in all our
0:46:54.620,0:47:01.010
simulations there is no planet B. And the[br]earth is quite unique and perfect for
0:47:01.010,0:47:06.570
human life. So, please take care of the[br]Earth and take care of yourself and of all
0:47:06.570,0:47:12.270
the others people on the Congress. And[br]thank you for listening and thank you to
0:47:12.270,0:47:20.380
everyone who helped us make this possible.[br]And to the people who actually coded our
0:47:20.380,0:47:24.210
programs with which we simulate. [br]Thank you!
0:47:24.210,0:47:37.370
applause
0:47:37.370,0:47:42.320
Herald: Thank you for the beautiful talk[br]and for the message at the end, the paper
0:47:42.320,0:47:47.970
is open for discussion, so if you guys[br]have any questions, please come to the
0:47:47.970,0:47:57.160
microphones. I'm asking my Signal Angel?[br]No questions right now. But microphone two
0:47:57.160,0:48:00.160
please![br]Mic2: Oh, yeah. Thank you very much.
0:48:00.160,0:48:05.690
Really beautiful talk. I can agree. I have[br]two questions. The first is short. You are
0:48:05.690,0:48:10.980
using Navier-Stokes equation, but you have[br]on the one hand, you have the dust disc
0:48:10.980,0:48:14.940
and on the other hand, you have solid[br]planets in it. And so are you using the
0:48:14.940,0:48:18.620
same description for both [br]or is it a hybrid?
0:48:18.620,0:48:23.550
miosta: It very much depends. This is one[br]of the things I showed you that for PLUTO,
0:48:23.550,0:48:31.300
we write this C file that specifies some[br]things and about every physicist has
0:48:31.300,0:48:39.090
somewhat his or her own version of things.[br]So, some usually the planets, if they are
0:48:39.090,0:48:47.030
large, they will be put in as a gravity[br]source. And possibly one that can accrete
0:48:47.030,0:48:54.090
and pebbles are usually then put in a[br]different way. However, also pebbles are
0:48:54.090,0:48:57.540
at the moment a bit complicated. There are[br]special groups specializing in
0:48:57.540,0:49:04.080
understanding pebbles because as we said[br]in the beginning, when they collide,
0:49:04.080,0:49:10.450
usually they should be destroyed. If you[br]hit two rocks very together, they don't
0:49:10.450,0:49:14.870
stick. If you hit them hard together, they[br]splatter around and we don't end up with an bigger object
0:49:14.870,0:49:23.390
caro: Just to explain pebbles are small[br]rocks or like big sand stones or something
0:49:23.390,0:49:28.710
like that. Yeah. So bigger rocks, [br]but not very big, yet.
0:49:28.710,0:49:33.190
miosta: Yes![br]caro: It depends on which code you use.
0:49:33.190,0:49:38.370
Mic2: Thank you. Very short, maybe one.[br]Do you also need to include relativistic
0:49:38.370,0:49:46.520
effects. Or is that completely out?[br]miosta: It's a good question. Mostly if
0:49:46.520,0:49:54.680
you have a solar type system, you're in[br]the arrange where this is not necessary.
0:49:54.680,0:50:00.010
For example, with the binaries, if they[br]got very close together, then at the inner
0:50:00.010,0:50:05.200
part of the disc, that is something we[br]could consider. And actually, I know for
0:50:05.200,0:50:10.500
PLUTO, it has modules to include[br]relativistic physics, too, yes!
0:50:10.500,0:50:14.000
Mic2: Thank you![br]Herald: OK, we have quite some questions,
0:50:14.000,0:50:19.700
so keep them short. Number one, please![br]Mic1: Thank you. Yeah. Thank you very
0:50:19.700,0:50:24.490
much for your interesting talk. And I[br]think you had it on your very first slides
0:50:24.490,0:50:31.780
that about 70 percent of the universe[br]consists of dark matter and energy. Is that
0:50:31.780,0:50:37.000
somehow considered in your [br]simulations or how do you handle this?
0:50:37.000,0:50:43.020
caro: Well in the simulations we make, we[br]are doing planets and discs around stars.
0:50:43.020,0:50:47.440
It's not considered there. In the[br]simulation we showed you about the
0:50:47.440,0:50:52.670
universe at the beginning, the blueish[br]things were all dark matter. So, that was
0:50:52.670,0:50:56.260
included in there.[br]Mic1: OK, thank you.
0:50:56.260,0:51:00.510
Herald: OK. Microphone 3.[br]Mic3: Hi, thanks. Sorry, I think you
0:51:00.510,0:51:05.740
talked about three different programs. I[br]think PLUTO, FARGO3D and a third one. So,
0:51:05.740,0:51:09.620
for a complete beginner: which program[br]would you suggest is like you more use
0:51:09.620,0:51:12.570
like if you want to learn more? [br]Which one is user friendly or good?
0:51:12.570,0:51:18.590
miosta: I would suggest FARGO3D first. It's[br]kind of user friendly, has a somewhat good
0:51:18.590,0:51:26.240
support and they are always also very[br]thankful for actual comments and additions
0:51:26.240,0:51:32.030
if people actually are engaged in trying[br]to improve on that. Because we are
0:51:32.030,0:51:37.328
physicists. We're not perfect programmers[br]and we're also happy to learn more. So
0:51:37.328,0:51:42.720
yeah, FARGO3D I would suggest, it has some[br]easy ways of testing some systems and
0:51:42.720,0:51:45.440
getting something done.[br]caro: And it also has a very good
0:51:45.440,0:51:53.567
documentation and also a manual "How to[br]make the first steps on the Internet". So,
0:51:53.567,0:51:56.980
you can look that up.[br]Mic3: Awesome. Thank you.
0:51:56.980,0:52:00.150
Herald: Let's get one question from[br]outside, from my Signal Angel.
0:52:00.150,0:52:05.600
Signal Angel: Thank you for your talk.[br]There's one question from IRC: How do you
0:52:05.600,0:52:09.510
know your model is good when you can only[br]observe snapshots?
0:52:09.510,0:52:17.770
caro: Oh, that's a good question. As we[br]said, we're in theoretical astrophysics.
0:52:17.770,0:52:25.170
So, there are theoretical models and these[br]models cannot include everything. So,
0:52:25.170,0:52:32.610
every single process, it's not possible[br]because then we would calculate for years.
0:52:32.610,0:52:37.480
Yeah, to know if a model is [br]good you have to…
0:52:37.480,0:52:46.430
miosta: Usually, you have a hypothesis or[br]an observation that you somehow want to
0:52:46.430,0:52:54.064
understand. With most of the necessary[br]physics at this stage to reproduce this
0:52:54.064,0:53:01.660
image. So, also from the observation we[br]have to take into the account what our
0:53:01.660,0:53:07.650
parameters kind of should be, how dense[br]this end of the simulation should be and
0:53:07.650,0:53:13.150
things like this. So, by comparing two[br]observations, that's the best measure we
0:53:13.150,0:53:21.790
can get. If we kind of agree. Of course,[br]if we do something completely wrong, then
0:53:21.790,0:53:26.600
it will just blow up or we will get a[br]horribly high density. So, this is how we
0:53:26.600,0:53:34.270
know. Physics will just go crazy if we do[br]too large mistakes. Otherwise, we would
0:53:34.270,0:53:39.330
try to compare two observations that it[br]actually is sensible what we did.
0:53:39.330,0:53:44.440
caro: Yeah, that's one of the most[br]complicated tasks to include just enough
0:53:44.440,0:53:52.400
physics that the system is represented in[br]a good enough way. But not too much that
0:53:52.400,0:53:57.400
our simulation would blow up in time.[br]Herald: Number two, please.
0:53:57.400,0:54:03.210
Mic2: I've got a question about the[br]adaptive grids. How does the computer
0:54:03.210,0:54:10.800
decide how to adapt the grid? Because the[br]data where's the high density comes after
0:54:10.800,0:54:17.660
making the grid...[br]miosta: Yes, this is actually quite an
0:54:17.660,0:54:25.470
interesting and also not quite easy to[br]answer question. Let me try to give a
0:54:25.470,0:54:34.300
breakdown nutshell answer here. [br]The thing is, you measure and evaluate the
0:54:34.300,0:54:39.380
velocities. Or in the flux, you also[br]evaluate the velocity. And if the velocity
0:54:39.380,0:54:44.840
goes high, you know there's a lot[br]happening. So, we need a smaller grid then
0:54:44.840,0:54:50.420
there. So, we try to create more grid[br]cells where we have a higher velocity. In
0:54:50.420,0:54:55.050
a nutshell, this is of course in an[br]algorithm a bit harder to actually
0:54:55.050,0:55:00.000
achieve. But this is the idea. We measured[br]the velocities at each point. And then if
0:55:00.000,0:55:03.510
we measure a high velocity, [br]we change to a smaller grid.
0:55:03.510,0:55:08.640
Mic2: So, you can predict where the mass[br]will go and whether densities are getting high.
0:55:08.640,0:55:12.600
miosta: Exactly. Step by step so to say.
0:55:12.600,0:55:15.890
Mic2: Thanks[br]Herald: We stay with Microphone 2.
0:55:15.890,0:55:20.640
Mic2: Okay. I've got a bit of a classical[br]question. So, I guess a lot relies on your
0:55:20.640,0:55:25.201
initial conditions and I have two[br]questions related to that. So first, I
0:55:25.201,0:55:30.670
guess they are inspired by observations.[br]What are the uncertainties that you have?
0:55:30.670,0:55:33.850
And B, then what is the impact if you[br]change your initial conditions like the
0:55:33.850,0:55:41.170
density in the disc?[br]miosta: Yeah, right now my main research
0:55:41.170,0:55:46.110
is actually figuring out a sensible[br]initial conditions or parameters for a
0:55:46.110,0:55:53.220
disc. If you just let it have an initial[br]set of conditions and a sensible set of
0:55:53.220,0:56:00.420
parameters and let it run very long, you[br]expect a system hopefully to convert to
0:56:00.420,0:56:05.130
the state that it should be in. But your[br]parameters are of course very important.
0:56:05.130,0:56:12.240
And here we go back to what we can[br]actually understand from observations. And
0:56:12.240,0:56:17.880
what we need for example is the density,[br]for example. And that is something we try
0:56:17.880,0:56:24.900
to estimate from the light we see in these[br]discs that you saw in this nice grid with
0:56:24.900,0:56:31.110
all these discs we estimate OK, what's the[br]average light there? What should then be
0:56:31.110,0:56:37.790
the average densities of dust [br]and gas in comparable disks.
0:56:37.790,0:56:42.890
Mic2: Okay, thanks.[br]Herald: Okay, one more at number two.
0:56:42.890,0:56:50.150
Mic2: Yes. Thank you for the talk. When[br]you increase the detail on the grid and
0:56:50.150,0:56:59.550
you learn more. When you want to compute[br]the gravitational force in one cell, you
0:56:59.550,0:57:05.090
have to somehow hold masses from the all[br]the other cells. So, the complexity of the
0:57:05.090,0:57:07.090
calculus grows.[br]miosta: Yes
0:57:07.090,0:57:13.820
Mic2: Quadratically, at the square of the... [br]how do you solve that? With more CPUs?
0:57:13.820,0:57:20.930
caro: Well, that would be one way to do[br]that. But there are ways to simplify if
0:57:20.930,0:57:26.290
you have a lot of particles in one[br]direction and they are far away from the
0:57:26.290,0:57:34.400
object you're looking at. So, yeah. So, if[br]you have several balls here and one ball
0:57:34.400,0:57:41.710
here, then you can include all these balls[br]or you can think of them as one ball. So,
0:57:41.710,0:57:48.770
it depends on how you look at it. So, how[br]you define how many particles you can take
0:57:48.770,0:58:02.230
together is when you look at the angle of[br]this... many particles we'll have from the
0:58:02.230,0:58:08.040
seen from the object you're looking at.[br]And you can define a critical angle. And
0:58:08.040,0:58:14.230
if an object gets smaller or if lot of[br]objects get smaller than this angle, you
0:58:14.230,0:58:20.200
can just say, OK, that's one object. So,[br]that's a way to simplify this method. And
0:58:20.200,0:58:23.390
there are some, yeah, [br]I think that's the main idea.
0:58:23.390,0:58:30.920
Herald: Okay, we have another one.[br]Mic2: Do you have a strategy to check if
0:58:30.920,0:58:35.890
the simulation will give a valuable[br]solution or does it happen a lot that you
0:58:35.890,0:58:42.060
wait one week for the calculation and find[br]out OK it's total trash or it crashed in
0:58:42.060,0:58:45.400
the time.[br]caro: So, that also depends on the program
0:58:45.400,0:58:53.240
you're using. So, in FARGO3D, it gives [br]these outputs after a certain amount of
0:58:53.240,0:58:58.980
calculation steps and you can already look[br]at those outputs before the simulation is
0:58:58.980,0:59:05.210
finished. So, that would be a way to[br]control if it's really working. Yeah, but
0:59:05.210,0:59:11.530
I think...[br]miosta: It's the same for PLUTO. So, there
0:59:11.530,0:59:18.040
is a difference between timesteps and[br]actually output steps. So and you could
0:59:18.040,0:59:23.490
define your output steps not and as the[br]whole simulation, but you can look at each
0:59:23.490,0:59:31.150
output step as soon as it's produced. So,[br]I usually get like 500 outputs, but I
0:59:31.150,0:59:36.630
already can look at the first and second after [br]maybe half an hour or something like that.
0:59:36.630,0:59:39.950
caro: Yeah, but it also happens that you
0:59:39.950,0:59:44.060
start a simulation and wait, and wait, and[br]wait and then see you put something wrong
0:59:44.060,0:59:48.760
in there and well then you have to do it[br]again. So, this happens as well.
0:59:48.760,0:59:53.070
Mic2: Thanks.[br]Herald: Okay. One final question.
0:59:53.070,1:00:02.240
Mic2: Yeah, OK. Is there a program in[br]which you can calculate it backwards? So
1:00:02.240,1:00:07.180
that you don't have the starting[br]conditions but the ending conditions
1:00:07.180,1:00:15.220
and you can calculate how it had started?[br]miosta: Not for hydrodynamic. If you go to
1:00:15.220,1:00:22.240
n-body, there is a way to go backwards in[br]time. But for hydrodynamics, the thing is
1:00:22.240,1:00:31.580
that you have turbulent and almost like[br]chaotic conditions. So, you cannot really
1:00:31.580,1:00:38.810
turn them back in time. With n-body you [br]can it because actually it's kind of... Well,
1:00:38.810,1:00:44.890
it's not analytically solved, but it's[br]much closer than like turbulences,
1:00:44.890,1:00:50.470
streams, spirals and all the [br]things you saw in the simulations.
1:00:50.470,1:00:57.561
Herald: OK, I guess that brings us to the[br]end of the talk and of the session. Thank
1:00:57.561,1:01:03.266
you for the discussion and of course,[br]thank you guys for the presentation.
1:01:03.266,1:01:16.730
36c3 postroll music
1:01:16.730,1:01:30.000
Subtitles created by c3subtitles.de[br]in the year 2021. Join, and help us!