1
00:00:00,090 --> 00:00:05,220
And thank you for the chance to be here for the invitation and the opportunity to speak of the colloquium today.
2
00:00:06,750 --> 00:00:14,820
The work I'm going to present is in some part very new, but I'm not going to present the latest,
3
00:00:15,780 --> 00:00:24,599
most complicated and sophisticated thing that was built on the eight or ten years of graduate students and post-docs and scientists,
4
00:00:24,600 --> 00:00:28,620
because it becomes rather impenetrable when you get too far down into the weeds.
5
00:00:29,280 --> 00:00:38,340
I am going to try to tell you about a phenomenon in Plasmas, which is ubiquitous and known since the sixties,
6
00:00:38,580 --> 00:00:44,399
but unappreciated in the field of turbulence until recently.
7
00:00:44,400 --> 00:00:54,330
And. Alex has kind of led the way on this particular project in terms of identifying these echo phenomena that I'm going to talk about.
8
00:00:55,350 --> 00:01:07,080
And then the people on the list, they include graduate students and other other colleagues from around the planet plasma physics community.
9
00:01:07,290 --> 00:01:14,490
There are many other people who've worked on these projects and been very much involved, including Paul Deller, who's here.
10
00:01:15,300 --> 00:01:26,310
But the way I decided which names to stick up or the slides I used is that if I actually were using something specifically from someone,
11
00:01:26,640 --> 00:01:29,400
a slide they gave me, then I did that.
12
00:01:29,400 --> 00:01:39,810
Otherwise this list would be very long because I actually John Maynard Keynes has a quote as saying he said something along the lines of.
13
00:01:42,160 --> 00:01:50,980
There is no limit to the number of something like the number of mistakes a person makes when they work alone.
14
00:01:51,580 --> 00:01:54,820
And I have enjoyed working very much with colleagues.
15
00:01:54,850 --> 00:01:59,020
I'm a collaborative kind of scientist, so this list will be much longer.
16
00:01:59,620 --> 00:02:06,120
Okay, so I'm going to presume that there are people in the room who already know this work, and I know who you are.
17
00:02:06,130 --> 00:02:10,200
And then there are also people in the room who would like to know what is going on.
18
00:02:10,210 --> 00:02:16,720
And so I'm going to try to present at least part of the time, material that should be accessible to everyone.
19
00:02:18,640 --> 00:02:25,120
So it may seem slow, but I'm doing it on purpose. Let me set the stage with the General Physics Colloquium.
20
00:02:25,120 --> 00:02:31,299
I'm going to work on the classical physics problem. There's no quantum mechanics, but it is a mini body problem.
21
00:02:31,300 --> 00:02:38,380
So we're going to consider systems of charged particles, very large number ten to the 23, ten to the 25, those kinds of numbers.
22
00:02:39,310 --> 00:02:43,299
And you know, since it's a non relativistic classical system,
23
00:02:43,300 --> 00:02:49,690
the particles are newly created or destroyed and they just move around under the influence of electric and magnetic fields,
24
00:02:50,230 --> 00:02:55,000
some of which are sometimes in problems applied from the outside world.
25
00:02:55,000 --> 00:02:58,239
And sometimes or in all cases, there's also self-consistent,
26
00:02:58,240 --> 00:03:03,820
electric and magnetic fields that arise from the charges and currents that arise in the plasma.
27
00:03:05,510 --> 00:03:11,600
You calculate those electric and magnetic fields from Maxwell's equations, and then under the influence of these fields,
28
00:03:11,780 --> 00:03:20,260
particles move around in classical trajectories with position and velocity changing in time, none disappear.
29
00:03:20,270 --> 00:03:27,860
And so the basic equations to describe what's going on has to describe where all those particles are.
30
00:03:28,010 --> 00:03:37,130
The probability of finding a particle x one, the one at the first particle at x and B the second particle x2 and B two and so forth.
31
00:03:37,670 --> 00:03:44,480
And it's a very simple equation. It's just that the total probability of finding all these particles in the system doesn't change.
32
00:03:45,560 --> 00:03:49,340
And starting from there, it's just a legal equation.
33
00:03:49,340 --> 00:03:53,899
It's reversible, and one could imagine working directly with that.
34
00:03:53,900 --> 00:03:57,139
But of course, in the capital it is a very large number.
35
00:03:57,140 --> 00:04:03,200
So we want to we want to simplify things.
36
00:04:03,200 --> 00:04:10,399
And the. There are two limits where people get very interested, the strongly coupled limit and the weekly couple of limits.
37
00:04:10,400 --> 00:04:14,510
So I'm going to be showing you weekly, couple of limit and I'll tell you what that means.
38
00:04:15,740 --> 00:04:23,810
Weekly couple plasmas are systems where in in a sphere called the device sphere,
39
00:04:25,550 --> 00:04:32,780
which is lambda by radius within a sphere, the number of particles within the by sphere is extremely large.
40
00:04:33,290 --> 00:04:44,149
And if you have that system, then you will have screening on this dividing scale and there will be no long range fields other than such as Arise.
41
00:04:44,150 --> 00:04:44,900
And I'll show you in a moment.
42
00:04:44,900 --> 00:04:52,250
So this number of particles in a in the by sphere is called the plasma parameter capital lambda and the collective effects
43
00:04:52,310 --> 00:04:59,390
dominate the behaviour of your at large scales and then you have within the device sphere something else is going on.
44
00:05:00,050 --> 00:05:03,290
I'm not interested in what happens inside the device sphere.
45
00:05:03,290 --> 00:05:09,560
I'm interested in phenomena which are astronomically, literally large compared to these little device spheres.
46
00:05:09,980 --> 00:05:20,530
So these collective effects are mediated by the electric and magnetic fields that arise from the charges and currents and going to
47
00:05:20,580 --> 00:05:29,809
the central assumption that makes everything work or the central observation is that the frequency of two body collisions is much,
48
00:05:29,810 --> 00:05:34,970
much lower than the plasma frequency, which is a typical frequency of oscillation.
49
00:05:35,330 --> 00:05:38,810
If you just took a plasma and kicked it, there would be a normal mode.
50
00:05:38,990 --> 00:05:46,280
It'll make a P, it's a very high frequency and the collisions happen much less frequently than that.
51
00:05:46,280 --> 00:05:48,980
So there's is rapid plasma frequency vibration.
52
00:05:50,030 --> 00:06:00,200
And so you take this little equation down to something that can be solved analytically or on a computer by using the PBGC y hierarchy.
53
00:06:00,230 --> 00:06:04,190
I'm not going to go through that. It's something that many of you have done.
54
00:06:04,550 --> 00:06:10,580
And it reduces the dimensionality of the problem radically to simply a Boltzmann equation where you now have
55
00:06:10,700 --> 00:06:17,089
this this little F is the probability of finding a single particle at a position X one and b one at a time,
56
00:06:17,090 --> 00:06:21,410
T rather than all six in plus one.
57
00:06:21,890 --> 00:06:30,860
And this left hand side of this equation, which I'm not at this point, I'm just going to just work on the left hand side.
58
00:06:31,370 --> 00:06:35,510
It's just the total derivative along the trajectories of the particles, if you look at that.
59
00:06:35,990 --> 00:06:40,760
So that looks a lot like if that were the only thing, it would still be DFT t equals zero.
60
00:06:41,060 --> 00:06:47,030
But on the right hand side you take into account to body interactions or collisions.
61
00:06:47,390 --> 00:06:51,080
And so you go through this process and you you end up with this equation.
62
00:06:51,470 --> 00:06:55,070
The velocities are just the velocities which arise from.
63
00:06:57,410 --> 00:07:02,750
Pushing. You know, solving Newton's equations and the accelerations come from the.
64
00:07:05,110 --> 00:07:08,440
E plus, b cross b force. So that's pretty much it.
65
00:07:10,270 --> 00:07:15,580
Okay. If the right hand side were really large and dominate the equation,
66
00:07:15,970 --> 00:07:23,290
then you could solve this equation order by order and say the collision operator is the dominant thing in the problem.
67
00:07:23,410 --> 00:07:28,720
And so we look for solutions to zero order. The collision operator equals zero.
68
00:07:29,170 --> 00:07:37,090
And when you do that, you find that the solution in the equation is the usual maximal Boltzmann kinetic distribution of particles.
69
00:07:37,510 --> 00:07:50,470
So this distribution, instead of having n dimensions and velocity space, right three and dimensions V, Z, Y and z now only depends on the density,
70
00:07:51,160 --> 00:08:01,450
the temperature and the mean flow, and then the functional form of the rest of it with the velocity here is comes from the collision operator.
71
00:08:01,750 --> 00:08:06,510
So this is a fantastic simplification. Now we have a three dimensional system.
72
00:08:07,760 --> 00:08:14,410
We have a few three dimensional circles, since we need equations for the density, the flow of the temperature in space as a function of time.
73
00:08:14,680 --> 00:08:18,130
So when the collisions are large, you can go through that process.
74
00:08:18,340 --> 00:08:27,360
You only have to keep track of those things. And when the collisions are large, the mean free path is short compared to whatever you're looking at.
75
00:08:27,370 --> 00:08:30,189
In fact, I'm going to say that's what I mean about the collisions for large.
76
00:08:30,190 --> 00:08:39,129
I'm going to look at systems where the particles are colliding on scales that are short compared to whatever it is.
77
00:08:39,130 --> 00:08:44,020
I'm interested in the wavelengths of the turbulence or the size of the device and what whatnot.
78
00:08:44,020 --> 00:08:47,530
So in that limit we get a fluid description.
79
00:08:48,490 --> 00:08:55,930
In fact, if you carry this through for each species and attend together in the appropriate ways, this is exactly how you get magneto hydrodynamics,
80
00:08:56,410 --> 00:09:01,690
which is an appropriate theory for describing large scale plasma motions in a when they're collisional.
81
00:09:03,490 --> 00:09:07,030
So I did that slowly because now that's not the system I'm going to solve.
82
00:09:07,090 --> 00:09:13,030
We don't do in our group. We're not looking at magneto hydrodynamics, but this is what you would do if you were.
83
00:09:13,390 --> 00:09:17,850
Instead, what we do is the opposite limit.
84
00:09:17,860 --> 00:09:21,610
We're interested. We still the divide length is tiny.
85
00:09:21,910 --> 00:09:26,020
Let's let's say for some typical system, it's less than a millimetre.
86
00:09:27,200 --> 00:09:33,200
And the system, the scales that we're interested in are our hundred metres kilometres, something like that.
87
00:09:33,210 --> 00:09:43,580
So the Dubai sphere is this tiny thing reached in 100 metres size objects and the main free path is in between.
88
00:09:44,090 --> 00:09:50,300
Or even longer. If you look if you calculate the mean free path for a typical magnetic confinement fusion experiment,
89
00:09:50,690 --> 00:09:54,590
it can be hundreds of kilometres in a device that's three metres in radius.
90
00:09:54,680 --> 00:10:02,600
So essentially the particles are not colliding. They're pursuing some other orbits mostly, and then slowly diffusing because of other things.
91
00:10:02,900 --> 00:10:07,240
Okay, so this is the equation we're going to solve.
92
00:10:07,250 --> 00:10:14,330
And in fact we're interested in the limit where the collision operator now is small and the the left hand side is dominant.
93
00:10:16,140 --> 00:10:19,770
This is a mess. We're back to the six dimensional nonlinear problem.
94
00:10:19,770 --> 00:10:24,060
We have light waves moving across the system, high frequency plasma oscillations.
95
00:10:24,270 --> 00:10:28,770
This equation still includes things down towards the by scale.
96
00:10:29,070 --> 00:10:34,830
So that's too much. We're not interested in all that. And there's another further single simplification that we want to take.
97
00:10:35,550 --> 00:10:44,040
And what we do is we try to identify in the equation here another fast timescale that we can average over so that we can look at the slower,
98
00:10:44,040 --> 00:10:52,680
bigger things. And that fast timescale is the is going to be the gyration of particles around the background magnetic field.
99
00:10:52,690 --> 00:10:57,510
So instead of looking at non magnetised plasma, I'm going to look at Magnetised plasma,
100
00:10:57,600 --> 00:11:04,530
which is just defined by the fact that the radius of gyration of the particles in the
101
00:11:04,530 --> 00:11:11,160
plasma around the magnetic field lines is small compared to the to the plasma itself.
102
00:11:11,490 --> 00:11:15,690
Okay. I may have dynamics smaller than the radius of gyration.
103
00:11:15,690 --> 00:11:21,989
I'm not throwing away those scales. I'm just saying I'm going to make an assumption that the plasma is magnetised,
104
00:11:21,990 --> 00:11:25,650
that there's enough magnetic field to give you a radio that are inside the device.
105
00:11:26,220 --> 00:11:31,290
And as soon as you do that, there's an idiomatic invariant that you can take advantage of the magnetic moment
106
00:11:31,830 --> 00:11:37,230
and you can really hone this equation down to something that can be solved.
107
00:11:37,680 --> 00:11:43,139
So this radius of gyration is simply the, you know, all these things.
108
00:11:43,140 --> 00:11:52,470
It's the V over the frequency of the cyclotron motion, which is QB over in C or or cube over M if you don't like my units.
109
00:11:53,710 --> 00:12:00,550
And we're going to look at things which are slow compared to this gyration frequency and long mean free path.
110
00:12:00,880 --> 00:12:04,120
And that is going to be the limit that we're interested in.
111
00:12:04,720 --> 00:12:13,900
So a typical plasma looks like this. And this in this system, the particles are rapidly gyrating and kind of drifting.
112
00:12:14,020 --> 00:12:21,520
They're almost staying exactly on the field lines. But if there's there are effects which can cause them to drift off of the field lines slowly.
113
00:12:21,910 --> 00:12:23,920
But mostly they're going round and round and round.
114
00:12:24,070 --> 00:12:32,530
And the theory, which was started in the late sixties and the formalism of this transformation I'm talking about,
115
00:12:32,530 --> 00:12:39,790
was completed in the eighties, is replacing this helical motion with rings here.
116
00:12:40,660 --> 00:12:46,180
These rings then move around and you get an integral differential nonlocal equation to describe the dynamics.
117
00:12:46,570 --> 00:12:52,630
But that sounds awful, but the reduction of all the time and space skills is fabulous.
118
00:12:52,960 --> 00:12:58,060
And so this is now attractive, tractable theory. Get some bullets here.
119
00:12:59,080 --> 00:13:02,559
The plasma ends up being this kind of plasma.
120
00:13:02,560 --> 00:13:03,850
It's highly anisotropic.
121
00:13:04,180 --> 00:13:11,860
If you look if you had a magic microscope and you looked in the perturbations of the density or something like that along the magnetic field,
122
00:13:11,860 --> 00:13:15,550
they would have very long correlation lengths, would all look like spaghetti.
123
00:13:15,970 --> 00:13:21,280
And across the magnetic field, a very short correlation length and short kind of variations of things.
124
00:13:21,790 --> 00:13:28,060
And this theory kind of naturally gives you the tools to work in that asymptotic limit.
125
00:13:30,880 --> 00:13:43,210
Let's just keep going. So on this fundamentally, when you say that we're going to order out these rapid gyrations, we get more things from the theory.
126
00:13:43,840 --> 00:13:50,770
You find that there are several asymptotic multi-scale time and space scales that drop out.
127
00:13:51,260 --> 00:13:58,510
And so the first is that on large scales, very large scales compared to this radius of gyration.
128
00:13:59,500 --> 00:14:00,879
Things are essentially static.
129
00:14:00,880 --> 00:14:10,810
It's a minefield theory and the distribution function can be split into this very large scale part, plus perturbations that can be arbitrary.
130
00:14:11,710 --> 00:14:18,790
They're going to vary and do things. It's you take this large scale part,
131
00:14:18,790 --> 00:14:25,989
it has a Boltzmann factor you to the minus Q five over T you can absorb that into the you can
132
00:14:25,990 --> 00:14:30,700
separate that out from the distribution function and h becomes a thing that we like to follow around.
133
00:14:31,870 --> 00:14:39,160
As I said, that's just now we're going to add and replace these distribution functions with rings instead of particles.
134
00:14:39,760 --> 00:14:44,440
I used to joke and say, this is ring theory, but I don't know, I did it.
135
00:14:44,590 --> 00:14:52,390
Okay, so I still do it. You get an equation for H, which is going to be the equation that we try to solve.
136
00:14:52,690 --> 00:15:05,080
H is a five dimensional equation. So H is a function of three position, three position coordinates, energy and magnetic moment of the particles.
137
00:15:05,260 --> 00:15:08,450
And we've thrown away the information of where the particle is around its star.
138
00:15:08,450 --> 00:15:11,620
Well, that. There are Poisson brackets.
139
00:15:11,650 --> 00:15:15,760
Chi here is just the general. It's fine. A and all that kind of stuff.
140
00:15:15,760 --> 00:15:20,170
It's the fields. So there's brackets of the fields with the distribution function.
141
00:15:20,500 --> 00:15:26,079
There's this free streaming of the particles along the field. Line B denotes the direction of the magnetic field.
142
00:15:26,080 --> 00:15:33,400
So this is particles moving along the field line and other kinds of turns, including the collision operator.
143
00:15:34,270 --> 00:15:37,329
So you have to add to this the Maxwell equations.
144
00:15:37,330 --> 00:15:38,320
I won't take you through that.
145
00:15:38,590 --> 00:15:45,760
You solve that system of equations on a computer and you can go off and predict the behaviour of plasmas and a lot of interesting systems.
146
00:15:46,840 --> 00:15:49,840
So the slow evolution is what I promised you thought.
147
00:15:49,840 --> 00:15:54,640
You go through all the same topics and you see that the number of particles doesn't change.
148
00:15:54,670 --> 00:15:56,890
So that's good. The temperature can change.
149
00:15:57,220 --> 00:16:05,170
If there's something about this plasma, if I stir it, I can change the energy of the plasma and or if I cool it, I could.
150
00:16:05,890 --> 00:16:16,209
And those things happen slowly. It turns out we can relate this heating in the third line there to the change in entropy of the
151
00:16:16,210 --> 00:16:25,540
plasma on the microscale so that we can calculate the heating from from a proper positive,
152
00:16:25,540 --> 00:16:33,940
definite quantity here in the in in the integral. So everything is nice and copasetic and you can go ahead and do do calculations.
153
00:16:34,480 --> 00:16:37,809
This thing is conserved, but it's not relevant to this talk.
154
00:16:37,810 --> 00:16:40,350
I just had it on this slide. Okay.
155
00:16:41,580 --> 00:16:49,170
Now, look, the point of this talk is to encourage it is to show you that Landau damping can be overwhelmed by echoes.
156
00:16:49,170 --> 00:16:52,379
But to make any sense out of that, I have to show you what Landau damping is.
157
00:16:52,380 --> 00:16:55,680
Talk about it in case some of you don't know and then show you what an echo is.
158
00:16:56,010 --> 00:16:57,170
So what is Landau damping?
159
00:16:57,180 --> 00:17:04,530
Well, that's a problem that everyone in plasma, they're either the last year of undergraduate or the first or second year of graduate school.
160
00:17:05,340 --> 00:17:09,209
It's it's kind of boring. So I'm going to do it a different way.
161
00:17:09,210 --> 00:17:20,220
I'm just going to look at the linear dispersion relation. So if I went back to this equation here and threw away all the brackets, which is to say,
162
00:17:20,220 --> 00:17:23,280
throw away this term, and then I'm actually going to throw away collisions.
163
00:17:23,490 --> 00:17:31,440
Let's look at a forum only collision with plasma and just see what is the dispersion relation, what kind of waves live in that plasma.
164
00:17:32,250 --> 00:17:41,940
So that means for each species we just have diversity free streaming and this this term, which is related to energy change.
165
00:17:43,280 --> 00:17:47,200
I'm going to make even simpler. Just look at a hydrogen plot loops, hydrogen plasma.
166
00:17:48,290 --> 00:17:51,619
The electrons have a tiny jar radius, the ions have a little bit bigger.
167
00:17:51,620 --> 00:18:00,080
We'll just ignore the electron things and you get a dispersion relation which to most of you is still ugly and the red isn't showing up.
168
00:18:00,950 --> 00:18:05,480
But that's not a fix that the yellow on the left.
169
00:18:05,570 --> 00:18:10,040
Those are our fame waves. Most people have heard that there's alpha waves swimming around in plasmas.
170
00:18:10,490 --> 00:18:18,830
The red are sound waves and slow magneto sonic waves, which are going to be the focus of the talk and the right is coupling.
171
00:18:19,310 --> 00:18:25,940
The right isn't quite clear, but when you look at scales large compared to the ion gyration,
172
00:18:26,600 --> 00:18:31,700
you can ignore the green and you just get alfvén waves and slow modes decoupled.
173
00:18:32,270 --> 00:18:37,490
So you have compressible things and you have non congressional things. Alpha doesn't compress the plasma.
174
00:18:37,760 --> 00:18:45,170
It's like if you had to feel lines, they move together like this in waves and the plasma isn't pushed between them.
175
00:18:45,470 --> 00:18:48,200
The other mode of oscillation, the red ones, the planet,
176
00:18:48,200 --> 00:18:54,470
the felines are doing this and you're squeezing and squeezing the plasma two modes of oscillation roughly.
177
00:18:54,680 --> 00:19:00,940
Okay. So the red one, if you go if you look at those compression,
178
00:19:00,950 --> 00:19:10,700
all things which have density fluctuations associated with them and you can centre a plasma where the magnetic field is in an appropriate sense weak.
179
00:19:11,630 --> 00:19:20,450
And what you find is in the middle you get that the frequency compared to K parallel V thermal,
180
00:19:20,450 --> 00:19:26,959
which is a typical wave number, frequency and wave number in the system is just that.
181
00:19:26,960 --> 00:19:28,790
This is a pure imaginary object here.
182
00:19:29,210 --> 00:19:39,020
So that these compress, if I, if I take a plasma and I look at scales large compared to the ions gyration and I stir up density perturbations,
183
00:19:39,590 --> 00:19:44,090
they're supposed to be damped on the timescale of K parallel B thermal so far.
184
00:19:44,600 --> 00:19:56,060
So now later I want to remember that fact. And when I go out and look at plasmas in the world, they should not have density fluctuations with short,
185
00:19:56,900 --> 00:20:01,270
parallel, mean free path short wave numbers because those should be strongly damped.
186
00:20:01,520 --> 00:20:04,610
Okay. So that's the thing to remember.
187
00:20:04,820 --> 00:20:11,850
And now that's in the high beta limit and the low beta limit. There's another wave. It's also strongly damped and it's called the ion acoustic wave.
188
00:20:11,870 --> 00:20:20,330
You can make it weakly damped by making the ions. You know, I wrote one thing.
189
00:20:21,780 --> 00:20:25,140
You can make it weakly damped according to this formula. Let's look at that wave.
190
00:20:25,410 --> 00:20:30,540
Okay, let's look at one of the waves that arises and see why it's strongly damped.
191
00:20:30,690 --> 00:20:32,490
This is Landau damping we're talking about.
192
00:20:33,760 --> 00:20:41,200
So we're going to do the collisional stamping of the ion acoustic wave with what's called a case van Kampen approach,
193
00:20:41,200 --> 00:20:44,889
which is to say it's kind of a continuous eigen function thing.
194
00:20:44,890 --> 00:20:48,850
So we have the original equation dft t. It was age before.
195
00:20:48,850 --> 00:20:57,040
Sorry, it's f for a moment. DFT t v. Parallel feedback grid f and this is e parallel the.
196
00:21:01,650 --> 00:21:09,000
Electric field, it arises because particles maybe get bunched up. So we can look at the long wavelength limit, just make things a little bit simple.
197
00:21:09,420 --> 00:21:13,379
And what we need to solve is an integral equation.
198
00:21:13,380 --> 00:21:24,360
And this is the solution. Jcvi has a principle part and the belt function part and these functions it arrives here
199
00:21:24,360 --> 00:21:29,939
f of v capital F of V is just the Maxwell in and delta n is the density perturbation,
200
00:21:29,940 --> 00:21:33,000
which is the integral of this velocity perturbation.
201
00:21:33,810 --> 00:21:39,030
So it turns out you can, with these continuous eigen functions,
202
00:21:39,480 --> 00:21:46,170
solve this problem and all you do is make the appropriate substitutions, match up the initial conditions and let it go.
203
00:21:46,530 --> 00:21:53,339
And what you see is a really beautiful picture of how Landau damping works.
204
00:21:53,340 --> 00:22:03,000
And so this picture that I'm going to show you is on the x axis, on the x axis, on the horizontal axis, I should say, is V parallel.
205
00:22:03,780 --> 00:22:10,260
And on the vertical axis is the perturbation of the of the distribution function.
206
00:22:10,380 --> 00:22:14,970
Okay. So this thing, what I want you to see here is it's highly oscillatory.
207
00:22:15,840 --> 00:22:25,469
And the lesser importance is that the analytical solution is the blue curve and the dots are solutions from this code.
208
00:22:25,470 --> 00:22:34,220
So we're capable of calculating highly oscillatory functions to find the density perturbation.
209
00:22:34,230 --> 00:22:41,100
I integrate over velocity, and this is a lot like problems that you've seen in steepest descent and other kinds of things.
210
00:22:41,100 --> 00:22:46,640
The function is getting highly oscillatory, so there's negative, positive, negative, positive, negative, positive,
211
00:22:46,650 --> 00:22:55,200
lots of contributions to the in a role that almost cancel and overall you don't get much of a density perturbation from this still complicated.
212
00:22:55,200 --> 00:22:59,520
So let me show you the movie. So this is the distribution function evolving in time.
213
00:23:01,990 --> 00:23:05,140
And you see that it gets more and more oscillatory as time goes on.
214
00:23:06,170 --> 00:23:10,190
In the beginning of this this before the movie starts.
215
00:23:10,700 --> 00:23:18,020
This the integral of the is is the density and the integral is whatever is under that curve.
216
00:23:18,440 --> 00:23:23,960
And so you have a finite density of perturbation. That density gives rise to an electric field.
217
00:23:24,290 --> 00:23:30,710
So this thing is something you could measure in a plasma. So you start this wave in the plasma and you let it go,
218
00:23:31,250 --> 00:23:37,640
and the distribution function gets more and more oscillatory so that the integral of it is going to zero very rapidly.
219
00:23:38,060 --> 00:23:41,420
And that's what landau damping is. There's no irreversibility here.
220
00:23:41,420 --> 00:23:46,760
You could turn time backwards and it would come back, but we don't know how to turn it on backwards.
221
00:23:46,760 --> 00:23:51,030
So. Basically it just keeps getting more and more oscillatory.
222
00:23:51,810 --> 00:23:55,800
Now, to get this solution, we turned off the collisions.
223
00:23:56,490 --> 00:24:08,970
And if we stopped this. I guess we're not able to if we if you look at this thing, the collision, we just respond to sharp curvature in this function.
224
00:24:09,390 --> 00:24:18,330
It's a low placing in velocity space. So collisions would tend to kill this thing, even if the frequency of collisions is very, very, very low.
225
00:24:18,750 --> 00:24:23,579
The variation of F of V is getting very, very strong.
226
00:24:23,580 --> 00:24:30,540
And so even a tiny amount of collisions will finally kill this thing off and give you the irreversibility.
227
00:24:31,200 --> 00:24:33,870
So let's go through it one more time. Very slowly.
228
00:24:34,950 --> 00:24:42,330
We take the equation, DFT plus V, DFT Z, and we put everything else on the right hand side and just call it a source and forget about it.
229
00:24:43,340 --> 00:24:49,190
Okay. If I have a partial differential equation of that form and the source isn't problematic,
230
00:24:49,610 --> 00:24:55,610
then the solutions are any function of z minus v t and that means,
231
00:24:56,990 --> 00:25:06,350
you know, we've solved to solve this partial differential equation, we can put in a plane wave E to the AKC, a z and evolve on time.
232
00:25:06,350 --> 00:25:10,010
And that actually courts horns are very closely to the picture I just showed you.
233
00:25:10,340 --> 00:25:18,170
The heat of the AKC factors out and the e to the ak v t is just something getting more and more oscillatory.
234
00:25:18,590 --> 00:25:24,020
Okay. If we took a source on the right hand side to be a kick a delta function in time,
235
00:25:24,560 --> 00:25:31,100
and we gave it a maxwell in distribution and then we let it and one harmonic cosine.
236
00:25:31,100 --> 00:25:36,470
KC In the z direction, you would in fact get this solution in the bottom right.
237
00:25:37,100 --> 00:25:39,919
And this is where phase mixing as terminology comes from.
238
00:25:39,920 --> 00:25:48,470
The phase is the of the i k v t the second term in this parentheses, and that phase is getting advancing in time.
239
00:25:48,800 --> 00:25:53,150
It's a complex phase. So it gets the function gets more and more wiggly.
240
00:25:53,150 --> 00:25:57,740
The mixing comes from the Maxwell part E to the minus B square.
241
00:25:58,130 --> 00:26:02,780
And so the convolution of those two things gives rise to the overall damping.
242
00:26:04,460 --> 00:26:14,810
Here's a picture of it. Now I'm looking at velocities in the vertical direction and X or Z is in the horizontal direction.
243
00:26:15,050 --> 00:26:21,770
This is just a picture of a perturbation of the distribution function in phase space.
244
00:26:22,070 --> 00:26:26,120
So in this equation, everything moves trivially. So in the upper equation,
245
00:26:26,360 --> 00:26:30,530
everything below the dash line is going to move to the left because it has negative velocity
246
00:26:30,530 --> 00:26:33,770
and everything above the dash line is going to move to the right has positive velocity.
247
00:26:34,400 --> 00:26:41,060
So in the beginning, if I integrate, say it zero vertically, I get a density bump because everything is the same sign red.
248
00:26:41,540 --> 00:26:43,340
Okay, so there's a density bump.
249
00:26:44,560 --> 00:26:52,180
As time goes on, those contours turn over and I start to pick up a little bit of blue and later it tips over even further.
250
00:26:52,180 --> 00:26:56,320
And it's red, blue, red, blue, red, blue. This is a different picture of the same thing I showed you in the movie.
251
00:26:56,810 --> 00:27:03,790
Okay. And the density dies off very quickly as these as these things get twisted over.
252
00:27:04,150 --> 00:27:08,350
That's phase mixing. That's essentially the process behind Landau damping.
253
00:27:08,770 --> 00:27:14,440
And now if you go and solve that original equation in all its glory and try to find roots of the dispersion relation,
254
00:27:14,980 --> 00:27:19,870
you know, you do you find on the left is the potential versus time from some way.
255
00:27:20,200 --> 00:27:24,099
It's a solitary it's got some high frequency, some slow damping.
256
00:27:24,100 --> 00:27:27,250
The ratio of the damping to the frequency is ten to the minus three.
257
00:27:27,670 --> 00:27:31,000
And the codes happily find these things and that's landau damping.
258
00:27:31,780 --> 00:27:37,980
And the right hand side is just you can drive the system and look at the resonance of the plasma.
259
00:27:37,990 --> 00:27:44,410
So these are normal mode. Q Curves, right? So the narrow ones are weakly damped and the fat one is so strongly damped.
260
00:27:47,190 --> 00:27:55,980
And then you can do that for zillions of parameters and understand how all the normal modes of the plasma evolve in the linear system.
261
00:27:57,560 --> 00:28:05,450
But we're not in a linear system. We're in turbulence. So the amplitudes, I assume, to get all of that, that the nonlinear terms were irrelevant.
262
00:28:05,450 --> 00:28:13,100
And we can only aim for the linear terms when we now include the fact that turbulence is going to mix things around in nonlinear fashion.
263
00:28:13,220 --> 00:28:17,510
With order one fluctuations, we get a different picture.
264
00:28:17,870 --> 00:28:24,500
And the typical story of turbulence is there's something happening at the large scales, sort of cartoonish at the top,
265
00:28:25,340 --> 00:28:29,120
and those large scale eddies interact with one another,
266
00:28:29,120 --> 00:28:33,860
break up into smaller scale and smaller scale and smaller scale until finally you get to tiny scales.
267
00:28:34,250 --> 00:28:38,780
Now we're in space instead of velocity space. And this is a kind of a turbulent picture.
268
00:28:41,630 --> 00:28:45,050
Here's how that would look in a in the literature.
269
00:28:45,050 --> 00:28:49,940
If you look in the literature and you say, I'm interested in these collision loss plasmas,
270
00:28:50,780 --> 00:28:55,940
the vertical axis, would we be on a log scale, the energy and the fluctuations?
271
00:28:56,450 --> 00:29:00,730
So you would stir at some large scale and those eddies would break up and break up.
272
00:29:01,550 --> 00:29:08,960
And after you wait and get into steady state, there's some scale where there's viscosity or something which damps out.
273
00:29:09,350 --> 00:29:13,580
And in the middle you have in a the kamagra or spectrum of K to the minus five thirds.
274
00:29:14,210 --> 00:29:21,320
So in a turbulent plasma where you ignore Landau damping, this is the kind of thing you're supposed to get.
275
00:29:21,470 --> 00:29:24,650
You're supposed to get energy going like the flux.
276
00:29:24,770 --> 00:29:30,620
You can predict the fluctuation amplitude of the turbulence where it's universal,
277
00:29:30,620 --> 00:29:35,960
a way from the special way you're stirring it and the dissipation and you get to the minus five thirds.
278
00:29:36,650 --> 00:29:42,260
But we just went through a bunch of work to say these slow modes, these compression perturbations,
279
00:29:42,260 --> 00:29:47,960
any density fluctuations that we see should be damped by this phase mixing process.
280
00:29:48,290 --> 00:29:54,829
And so, in fact, the blue curve, the slow modes or the compression of waves shouldn't be the same.
281
00:29:54,830 --> 00:29:56,239
They shouldn't be minus five thirds.
282
00:29:56,240 --> 00:30:05,030
We think in the presence of this landau damping, maybe they will be steeper, maybe it's non universal and not even a power law.
283
00:30:06,320 --> 00:30:16,910
It's a complicated system. So, you know, if you look at the literature before Alex started this kind of.
284
00:30:18,660 --> 00:30:26,700
Looking back through everything and straightening things out, what you would find is people thought that the slow modes would be completely absent.
285
00:30:26,910 --> 00:30:31,110
And all we would do if we looked at one space or in an experiment, we would see our fan waves.
286
00:30:31,410 --> 00:30:34,080
We would see no density. Perturbations, nothing. Compression all.
287
00:30:36,140 --> 00:30:44,060
But when we look out in space or in experiments, there are tons of density perturbations over a large range of scales, particularly on the right.
288
00:30:44,090 --> 00:30:51,590
This is data from the solar wind. This is normalised density fluctuations versus wave number.
289
00:30:52,010 --> 00:30:58,430
And over two or three decades, there are larger, you know, K to the minus five, the density fluctuations.
290
00:30:59,030 --> 00:31:04,700
So we have a serious kind of glitch. The literature literature says there should be no density perturbations.
291
00:31:05,360 --> 00:31:09,050
Straightforward theory says there should be no no density perturbations.
292
00:31:09,500 --> 00:31:16,820
But when you look out in space for a typical large plasma, there are enormous density fluctuations.
293
00:31:16,970 --> 00:31:20,150
So what gives? What is the solution to this paradox?
294
00:31:21,080 --> 00:31:25,910
That's the point of of this talk. Why why are there these fluctuations?
295
00:31:27,840 --> 00:31:35,010
So we're going to redo that little calculation, including one more term, and that's some nonlinear, nonlinear, turbulent advection.
296
00:31:35,670 --> 00:31:42,690
So we have the same thing DFT, free streaming V, parallel DFT Z, and now we're going to have some flows,
297
00:31:42,870 --> 00:31:46,260
some turbulent flows, mixing things in the perpendicular direction.
298
00:31:46,830 --> 00:31:53,760
This is advection by the e cross velocity of here in the no, but basically perpendicular flows added at unit amplitude.
299
00:31:54,790 --> 00:32:01,060
So what do you get? Well, you don't have to do a lot of algebra to see something interesting if you put in plain waves.
300
00:32:02,230 --> 00:32:15,910
You'll see that the nonlinear term, the grad f will couple these perturbations together and you'll get the wave numbers coupling such that,
301
00:32:15,910 --> 00:32:19,899
you know, you can have a wave wave number P in a wave with the wave number.
302
00:32:19,900 --> 00:32:26,430
Q They will come together to give you the one. K And this coupling is going to be the thing that produces echoes.
303
00:32:26,890 --> 00:32:28,630
Echoes are going to be really interesting.
304
00:32:28,720 --> 00:32:36,910
Okay, so all we're going to do is look at the importance of a quadratic nonlinear d in this land l damping problem.
305
00:32:37,060 --> 00:32:47,580
Okay. Pretty straightforward. So what what's going to happen in the next few slides instead of these waves damping away, as I showed you,
306
00:32:47,580 --> 00:32:53,280
that we can calculate analytically and numerically to to to nothingness,
307
00:32:54,240 --> 00:32:59,790
the density perturbations are going to be continuously reconstituted by echoes from this wave coupling.
308
00:32:59,940 --> 00:33:04,320
Okay, so how would that work? Well to skip ahead.
309
00:33:04,830 --> 00:33:11,280
What's going to happen is we're going to create perturbations in this x v space which are tilted to the left.
310
00:33:11,610 --> 00:33:14,340
Let's see, I should do it this way, which are tilted to the left.
311
00:33:14,760 --> 00:33:22,350
So now we start with no density perturbation at some moment in time and free streaming reconstitute some density perturbation.
312
00:33:22,950 --> 00:33:29,640
And later you get a density when the natural, you know, just free streaming of particles with their velocities.
313
00:33:31,800 --> 00:33:36,630
Uh, leads to this kind of thing. So this is what's going to happen.
314
00:33:36,660 --> 00:33:43,410
But the question is, why would you ever expect to get something coherent up there on the left that would unfazed makes to give you this echo.
315
00:33:44,900 --> 00:33:49,960
Well, the answer to that is pretty straightforward. The skeleton of the idea goes like this.
316
00:33:50,920 --> 00:33:55,300
We could solve the equation and we have to do this here.
317
00:33:55,750 --> 00:34:01,389
And we would get that some plane wave with wave in a group we go to like E to the IP,
318
00:34:01,390 --> 00:34:07,930
Z minus T, because remember the linear waves just have it's any function of Z minus v t.
319
00:34:08,170 --> 00:34:12,800
Oops. Sorry. Oops. Wrong way.
320
00:34:14,270 --> 00:34:17,599
So let's let that run for a time.
321
00:34:17,600 --> 00:34:22,100
Capital T, we're just going to do a fake problem. So we imagine we have a perfect control of our plasma.
322
00:34:22,610 --> 00:34:26,239
We put in a little perturbation. It starts to mix over right.
323
00:34:26,240 --> 00:34:31,300
And it and we go until it has accumulated phase PBT.
324
00:34:32,660 --> 00:34:40,790
Then we put another mode in each of the I choose the different wave number and we let that one go.
325
00:34:40,820 --> 00:34:43,970
Now we know that we're going to get just broke the pointer.
326
00:34:45,260 --> 00:34:49,910
We know that because of the coupling P plus Q has to be K.
327
00:34:50,330 --> 00:34:56,479
And so this one is now going to evolve. Thanks. It's the instantaneously looks like this.
328
00:34:56,480 --> 00:35:02,600
And now we let it run forward in time. And everywhere the Z appears, you get z minus v, t.
329
00:35:02,930 --> 00:35:10,410
And so F of k now looks like. Even the ak z minus v, t minus I APB capital t.
330
00:35:11,320 --> 00:35:20,800
So we have the opportunity to have something interesting happen when the phase is such that the amplitude of the phase is one.
331
00:35:21,190 --> 00:35:28,570
And that happens when you've got a time T which is P over K times, the original time that we let things go.
332
00:35:29,440 --> 00:35:36,550
If you do that, you get an echo. Well, the echo has to happen after the two perturbations, not before.
333
00:35:36,760 --> 00:35:42,069
And that gives you a constraint. It says that P over K has to be bigger than one.
334
00:35:42,070 --> 00:35:47,920
P has to be bigger than K. So certain kinds of wave numbers are going to interact to give you this wave number.
335
00:35:48,310 --> 00:35:53,530
And in fact, when you're done, what you find out is the sign of K or P times.
336
00:35:53,530 --> 00:35:55,420
The sign of Q has to be minus one.
337
00:35:55,810 --> 00:36:03,010
So it's a certain kind of of one way of going this way and a way of going that way can interact to give you an echo.
338
00:36:03,550 --> 00:36:06,910
Okay. So this is sort of abstract.
339
00:36:07,330 --> 00:36:13,660
You can get an echo. The question will be, do we get echoes and are they numerous and interesting?
340
00:36:14,500 --> 00:36:19,960
So again, this is the kind of perturbations you get. All we're doing is unwrapping phase space and then letting it go forward.
341
00:36:21,480 --> 00:36:26,880
If we do that, we we're trying to find the fate of these elfin in slow modes.
342
00:36:27,330 --> 00:36:31,889
So let me just remind you very quickly, the iPhone waves on the left, no density perturbations.
343
00:36:31,890 --> 00:36:36,180
They're incompressible. All they are velocity and magnetic field fluctuations.
344
00:36:36,660 --> 00:36:40,710
And they're described by a theory which you can get from what I've shown you.
345
00:36:41,340 --> 00:36:48,720
The slow loads are compressive. They give density fluctuations. Also delta field strength, magnetic field strength fluctuations.
346
00:36:48,990 --> 00:36:52,860
And they're basically the slow modes. Follow the alpha waves around.
347
00:36:54,030 --> 00:36:59,850
They're slaved and evicted. So they are fame waves.
348
00:37:00,330 --> 00:37:04,500
You know, you can work out you can work out the description of the waves.
349
00:37:05,670 --> 00:37:10,860
It's these variables, Zeta plus and Zeta minus. I don't want to belabour the point.
350
00:37:10,860 --> 00:37:14,879
Those are left going and right going alpha and waves in the box. So there's some going this way.
351
00:37:14,880 --> 00:37:19,680
Some going that way. And they're independently conserved.
352
00:37:20,130 --> 00:37:24,860
It's not too important right now. The slow modes are.
353
00:37:25,770 --> 00:37:29,540
Let's see. I have to point with this. DVT.
354
00:37:29,730 --> 00:37:33,770
It's the convective derivative with the velocity and the beat out grid is long.
355
00:37:33,780 --> 00:37:40,620
The perturbed feline is what that means, but essentially developed as a non, as a linear and a nonlinear part.
356
00:37:40,620 --> 00:37:44,640
And beta grad is a linear and a nonlinear part. And you get this equation down here at the bottom.
357
00:37:45,180 --> 00:37:53,940
This is the key equation for the moment. It says that this is this is this is the slow motion part of all that very kinetics.
358
00:37:54,390 --> 00:38:01,260
And we just have a it's it's a it's straightforward equation to solve.
359
00:38:02,010 --> 00:38:07,640
And now it looks kind of ugly. I think the next slide, I show you the approach that we're going to do.
360
00:38:08,450 --> 00:38:17,690
Well, yeah, that's to wake you up. So this equation describes the evolution of G, which is the perturbation of the distribution function.
361
00:38:17,780 --> 00:38:22,190
So it's a function of velocity. Okay. And position.
362
00:38:23,000 --> 00:38:29,719
The alpha waves here are described by the Zeta pluses and minuses.
363
00:38:29,720 --> 00:38:36,680
So my grad student wrote a code he needed to solve for G and the Alpha Waves, and he called it Gandalf or Gandalf.
364
00:38:38,070 --> 00:38:44,639
That's where he got the name. So he wrote this code. It runs on a GPU, so it runs at teraflop speeds.
365
00:38:44,640 --> 00:38:47,220
It has lots of exciting numerical properties.
366
00:38:47,580 --> 00:38:54,210
It's also all the equations I've shown you, and it does it for you in real space and hermie polynomials in velocity space.
367
00:38:54,690 --> 00:38:59,940
And you know, we're going to investigate just because echoes could happen or can happen.
368
00:38:59,940 --> 00:39:03,600
Do they happen and are they interesting to the evolution of the system?
369
00:39:04,530 --> 00:39:10,830
And so before I do that, I want to show you the kind of what do we do?
370
00:39:11,220 --> 00:39:16,620
Well, here's the equation without the nonlinear term.
371
00:39:17,010 --> 00:39:19,860
And on the right hand side, this kind of t is the source.
372
00:39:19,870 --> 00:39:27,090
So we're just going to take the linear equation instead of solving an initial value problem like I showed you before, where the wave decays away.
373
00:39:27,330 --> 00:39:33,750
I'm going to do it a little differently. I'm just going to stir constantly and with a lines of an equation.
374
00:39:34,020 --> 00:39:40,860
So you can think of it as as a driven problem instead of an initial value problem.
375
00:39:41,130 --> 00:39:47,130
The properties of this source are just delta correlated white noise with some amplitude epsilon.
376
00:39:47,970 --> 00:39:54,900
And when you have a problem like this is a different problem on the bottom.
377
00:39:54,900 --> 00:40:05,400
When you have a problem where you have just a scalar with the launch of an equation, with some driving chi at amplitude alpha and some damping gamma,
378
00:40:05,700 --> 00:40:10,859
then you can predict with the fluctuation dissipation theorem what the amplitude of five square will look like,
379
00:40:10,860 --> 00:40:17,010
the energy in the in the electric field and it looks like this epsilon alpha squared over the damping.
380
00:40:18,040 --> 00:40:24,790
Well, we're doing a different problem. This is for a scalar. We're doing a a continuous variable in velocity space.
381
00:40:24,790 --> 00:40:28,330
And so the first thing to do is to see what that answer is.
382
00:40:28,960 --> 00:40:36,970
And what you get is by square, the energy looks like some general function of alpha, the driving and instead of gamma,
383
00:40:36,970 --> 00:40:42,100
it's absolute value k which is a lot like the damping gamma when you get down to it, but it's different.
384
00:40:42,670 --> 00:40:48,530
And this function F of Alpha is a transcendental function that's a mass to calculate.
385
00:40:48,530 --> 00:40:51,940
But the grad student did a good job. And so his code,
386
00:40:52,120 --> 00:40:57,520
the analytical response is suppose this is the f of alpha function looks like this
387
00:40:57,550 --> 00:41:03,879
as a function of alpha and his his Gandalf code is able to calculate that properly,
388
00:41:03,880 --> 00:41:11,020
all those little red dots so he knows he can do a driven system and it's behaving right.
389
00:41:11,020 --> 00:41:14,169
That's the linear equations. How did he do it?
390
00:41:14,170 --> 00:41:21,520
He did it with hermy polynomials. And so you took that D that equation for G and expand instead of GSV.
391
00:41:21,520 --> 00:41:31,870
Now we're going to have general hermit moments g sub them so that you get a hierarchy of o.d well one one
392
00:41:31,870 --> 00:41:40,090
level simpler pdes and the bottom one is the is the case for all hermit mode numbers bigger than two.
393
00:41:40,540 --> 00:41:49,839
It's kind of all the rest of the ends of the same after you get past the sort of junk at the beginning and the coupling of
394
00:41:49,840 --> 00:41:59,020
the M plus one hermit moment to the and the M minus one to the end is the expression of phase mixing and landau damping.
395
00:41:59,020 --> 00:42:04,360
So we got to watch those two guys carefully. And in fact, if we just go to the large now,
396
00:42:04,360 --> 00:42:10,270
we're getting to a little more sophisticated view of the same thing I've been talking about for the last hour.
397
00:42:10,660 --> 00:42:20,350
We do a transformation and we end up with these functions G plus tilde and G minus tilde, which I'm not going to justify.
398
00:42:20,350 --> 00:42:24,460
It's appeared in Paul's papers and in Alex's papers. It's a good transformation.
399
00:42:25,270 --> 00:42:28,740
And you look at a large m so we're looking at the vote.
400
00:42:28,960 --> 00:42:32,440
We're looking at that distribution function when it's very wiggly. Right?
401
00:42:32,440 --> 00:42:36,370
Very. The hermit mode number is large and we're seeing what happens.
402
00:42:37,120 --> 00:42:43,480
It turns out you can write the equation in the form, in the box. And if you stop and look at that, it's it's fantastic.
403
00:42:43,930 --> 00:42:45,400
The G plus solution,
404
00:42:45,730 --> 00:42:56,170
you get a plus sign here and you have a flux in M in hermit mode number and some damping from the collisions on the right hand side.
405
00:42:56,530 --> 00:43:05,049
So this plus solution manifestly represents a flux of energy from low hermit numbers to high hermit numbers.
406
00:43:05,050 --> 00:43:10,330
And that's the picture I showed you. That's the distribution function getting starting off not wiggly and getting
407
00:43:10,330 --> 00:43:14,080
more and more wiggly as the energy goes to higher and higher hermit numbers.
408
00:43:14,860 --> 00:43:20,649
The minus solution is the opposite. The minus solution is energy coming from very wiggly modes.
409
00:43:20,650 --> 00:43:24,310
The flux is negative to the low ends. That's the echoes.
410
00:43:24,820 --> 00:43:29,830
So this formalism gives you the echoes right away.
411
00:43:30,460 --> 00:43:41,800
And so you just form the appropriate look in the middle, you form the appropriate G plus squared as a function of hermit number and g minus square.
412
00:43:42,130 --> 00:43:48,670
And this tells you the flux plus is the flux of energy too high numbers to be
413
00:43:48,670 --> 00:43:53,710
Landauer damped away and the minus is the flux of energy to the low hermit numbers.
414
00:43:54,720 --> 00:44:02,250
They have a bunch of constants and then this one is one over squared compared to this where S squared is just the hermit number.
415
00:44:03,000 --> 00:44:11,129
These have the form of parasitic solutions that Paul showed me correspond to fermion doubling in quantum mechanical systems.
416
00:44:11,130 --> 00:44:15,690
But those are fake and come from the democratisation scheme.
417
00:44:15,690 --> 00:44:21,390
These are physical waves that are, you know, plus and minus and in hermit space.
418
00:44:22,260 --> 00:44:26,040
So what do they look like? I just want to show you a picture.
419
00:44:26,610 --> 00:44:32,820
If I take that linear system that I told you about in this space, the driven system, you get a spectrum.
420
00:44:33,330 --> 00:44:40,080
Now, we the point of all those transformations is to separate these two things and give one of them no variation.
421
00:44:40,080 --> 00:44:47,670
So the F plus solution is just constant as a function of square root of M, and the other solution goes like one over A squared.
422
00:44:47,820 --> 00:44:53,290
So the top one is landau damping. And the bottom one is the echoes in the linear limit.
423
00:44:54,100 --> 00:45:04,030
So if I stir at at low mode number, low put large scale perturbations in velocity space, then I'll fill up phase space.
424
00:45:04,240 --> 00:45:08,230
It'll it'll fill up the black curve until I get to the collisions.
425
00:45:08,500 --> 00:45:15,880
And in fact, if you let the collisions go to zero, you get an infinite amount of energy stuffed into the system.
426
00:45:16,450 --> 00:45:21,170
It goes like one over new to the one third. The Echo Park.
427
00:45:21,810 --> 00:45:25,799
You would have to store at large M and you would find energy going back.
428
00:45:25,800 --> 00:45:30,300
And then the linear problem where we just put electric fields, you don't get perturbations that large.
429
00:45:30,300 --> 00:45:34,910
M So we're not accustomed to seeing the echoes. The echoes have to come from something else.
430
00:45:34,910 --> 00:45:41,639
So they're not electric field perturbations. And so if you wanted to see them, you could mark it up.
431
00:45:41,640 --> 00:45:48,990
This is the same equation now it has the second term up here is the advection.
432
00:45:49,890 --> 00:45:55,379
And if we so I put this slide in hesitantly,
433
00:45:55,380 --> 00:46:02,850
but I wanted to show you remember we're still driving G with this chi function which is del function correlated in time coming from.
434
00:46:05,340 --> 00:46:11,730
And then we're going to also stellar random perpendicular velocity across the field.
435
00:46:11,790 --> 00:46:18,390
Okay. So we have some large scale electric field wiggles and we have some convection wiggles.
436
00:46:20,860 --> 00:46:26,860
And this is the beautiful thing you go through and you do that transformation on the nonlinear term.
437
00:46:27,190 --> 00:46:32,560
And what you find, what Alex found was the nonlinear term breaks into two parts.
438
00:46:34,340 --> 00:46:39,400
G plus is coupled to g plus.
439
00:46:39,410 --> 00:46:42,770
So that this this one represents.
440
00:46:47,040 --> 00:46:54,900
Just advection. If Delta Plus if the sign of K and Q are the same, if they're opposite,
441
00:46:55,710 --> 00:47:02,820
then you excite this term and you get G minus or looking at the other way for g minus a g plus x is a source.
442
00:47:03,210 --> 00:47:08,430
So this means that when the wave numbers are opposite, just as I showed you in the toi problem,
443
00:47:08,910 --> 00:47:14,340
the nonlinear t gives you a source and you will find that you're generating echoes.
444
00:47:15,550 --> 00:47:20,650
So it's very natural. If you have wave numbers of all kinds, you're going to have some kind of echo.
445
00:47:21,040 --> 00:47:27,880
The remaining problem is a numerical problem to say you get a lot of those echoes or do you get not much.
446
00:47:28,450 --> 00:47:38,750
And so. My student worked out the slow modes and for the problem I just showed you, the blue is where there's.
447
00:47:39,300 --> 00:47:42,860
I'll just tell you if you don't have to think about it. These are the different wave numbers.
448
00:47:43,190 --> 00:47:50,210
The blue is where there's a lot of echo and the red is where there's a lot of landau damping or phase mixing.
449
00:47:50,480 --> 00:47:57,139
And there's a line you can work out between them where and this all sort of works for the little fake problem that I just showed you,
450
00:47:57,140 --> 00:48:01,580
where I'm fake, stirring everything. What we like to do is the real problem.
451
00:48:03,660 --> 00:48:08,360
And. And we'd like to do this slow mode.
452
00:48:08,390 --> 00:48:11,930
So now I'm almost done. Now we do the real problem.
453
00:48:11,930 --> 00:48:16,150
We do stick the slow modes, the alpha waves. We put it all into the code.
454
00:48:16,160 --> 00:48:26,180
We have a self-consistent motion of the field lines, self-consistent flows and everything is calculated.
455
00:48:26,630 --> 00:48:27,410
Self-consistent.
456
00:48:27,590 --> 00:48:33,470
The only thing we're doing is it's kind of weird as we're stirring the alpha waves instead of letting them arise from some natural thing.
457
00:48:33,480 --> 00:48:39,140
So we stir the alpha waves. We've got some slow modes and we want know what happens.
458
00:48:39,920 --> 00:48:43,760
And just so that you can see that we're not cheating,
459
00:48:43,760 --> 00:48:52,549
this is the energy in the velocity perturbations and the perpendicular field perturbations that should be roughly K to the minus five thirds.
460
00:48:52,550 --> 00:48:59,660
If our simulations are right, this is the alpha waves. This is an important point, but I'm not going to belabour it.
461
00:48:59,690 --> 00:49:06,770
What we find is that the slow modes develop a parallel spectrum and it goes like K parallel to the minus to so the density.
462
00:49:07,060 --> 00:49:13,280
So you see these modes have density fluctuations that have some k parallel spectrum.
463
00:49:16,200 --> 00:49:21,660
What does that mean? Well, how how how do we have these? How do we have these modes when they're supposed to be damned?
464
00:49:22,530 --> 00:49:26,880
Well, you know, we can look individually at the field compression and the density fluctuations.
465
00:49:27,390 --> 00:49:31,770
At the end of the day, I can put those there for if you want to come back.
466
00:49:32,190 --> 00:49:41,550
This is the key picture. Blue means that there were echoes dominating and any other colour is where it's got landau damping.
467
00:49:42,030 --> 00:49:48,420
And so before we did these, before the theory, before the simulations, people assumed that this picture would be red.
468
00:49:49,290 --> 00:49:55,860
And in fact, when you go through and you calculate, you find that there are as many echoes as there are forward.
469
00:49:56,220 --> 00:50:00,060
Landauer damping kinds of processes going and and what does that mean?
470
00:50:00,150 --> 00:50:05,490
That's it's it's quite surprising. It means you should get density fluctuations when you look out in space.
471
00:50:07,740 --> 00:50:12,870
If I wanted to ask the question, is the echo happening?
472
00:50:12,870 --> 00:50:21,970
Where things? Is it, you know, just because there's a lot of echo, maybe that little yellow bit Lando dampens down to everything away.
473
00:50:22,020 --> 00:50:30,690
Okay, so this is the picture of F, the capital F that I showed you was flat when it's Landauer damping in the turbulent problem.
474
00:50:30,690 --> 00:50:37,050
It's got a very definite slope. And that slope is, you know, we'll work.
475
00:50:37,320 --> 00:50:40,860
Alex will do the theory and figure out if he can predict that slope.
476
00:50:41,700 --> 00:50:49,379
I'm sure he will. Anyone else that wants to do it should that slope arises from all the self-consistent dynamics.
477
00:50:49,380 --> 00:50:52,650
And remember, this is the linear result, the top CR.
478
00:50:53,430 --> 00:50:59,190
If we had Landau damping dominating, which is what everybody said would happen, you'd have a flat curve in this picture.
479
00:50:59,190 --> 00:51:03,810
And instead it's a steep slope of something like my minus five halves.
480
00:51:04,260 --> 00:51:06,450
So that's really quite remarkable.
481
00:51:08,130 --> 00:51:17,340
In fact, you can you can look and this is the curves of the phase mixing in the energy and modes that is beta phase mixing and unfazed,
482
00:51:17,340 --> 00:51:24,120
mixing in the fact that they're almost on top of each other tells you that the echoes are balancing the rest of the problem.
483
00:51:24,870 --> 00:51:32,930
So what does that mean? What it means is. These are all the different moments in the plasma and it gets very small.
484
00:51:32,940 --> 00:51:39,810
The amplitude drops off quickly instead of being flat. It means this collision, this plasma is acting like a fluid.
485
00:51:40,470 --> 00:51:50,160
The turbulence is somehow through this echo mechanism, causing the plasma act like a fluid instead of a collision the system.
486
00:51:51,710 --> 00:51:59,690
So that's the end. These numerical simulations support what Alex has in a recent paper that several of us here are co-authors on.
487
00:51:59,930 --> 00:52:04,040
The Echoes, in fact, do overwhelm the landau damping for the slo mo problem.
488
00:52:04,340 --> 00:52:10,570
And you should expect to see density fluctuations in the interstellar medium and the solar, wind and things like that.
489
00:52:12,080 --> 00:52:23,659
This low collision ality plasma in a turbulent state can be really well-described by only a few low hermit numbers of room moments,
490
00:52:23,660 --> 00:52:30,110
which is to say it's behaving like a fluid, not like a a collision with complicated plasma.
491
00:52:30,380 --> 00:52:36,740
Many of us over the years model this turbulent dissipation as a nonlinear cascade to give you the amplitudes and landau damping,
492
00:52:36,740 --> 00:52:39,940
to give you the absorption and this is wrong.
493
00:52:39,950 --> 00:52:43,339
There is no the plasma. It's not absorbing.
494
00:52:43,340 --> 00:52:51,600
It's it's echoing. So this may changed the predictions of iron an energy on an electron heating in various plasmas.
495
00:52:51,920 --> 00:52:55,580
And to really say what's going on, we have to better understand the physics.
496
00:52:56,090 --> 00:53:03,050
Once you get down to the radii, which the theory, you know, that's the next step and we'll use the same machinery.
497
00:53:03,170 --> 00:53:04,760
So that's my time. Thank you.