Latest / Daniel and Kelly’s Extraordinary Universe / Can we simulate the Universe?

Transcript

Can we simulate the Universe? · Daniel and Kelly’s Extraordinary Universe

• 0:07
  Speaker 1: How did the universe get this way? It's got all
• 0:10
  Speaker 1: these beautiful stars and galaxies and quasars and flamingos and bicycles.
• 0:15
  Speaker 1: Why is it this way and not some other way?
• 0:19
  Speaker 1: Do we understand that? How could we possibly explore that
• 0:22
  Speaker 1: question without seeing it happen in real time or making
• 0:26
  Speaker 1: universe-size experiments? That's where the simulation comes in. Computers these
• 0:32
  Speaker 1: days can do your homework and read your emails, but
• 0:35
  Speaker 1: they can also... answer tough science questions like, why is
• 0:39
  Speaker 1: the universe this clumpy? Why doesn't it have more stars
• 0:43
  Speaker 1: or more flamingos? Today, we'll dig into how to study
• 0:47
  Speaker 1: the universe on a computer. And yes, of course, we'll
• 0:51
  Speaker 1: talk about whether our universe is someone else's homework problem.
• 0:56
  Speaker 1: Welcome to Daniel and Kelly's Extraordinary Maybe Simulated Universe.
• 1:01
  Speaker 2: Intro Hello, I'm Kelly Wienersmith.
• 1:16
  Speaker 3: I study parasites and space, and I sometimes wonder if
• 1:21
  Speaker 3: we live in a simulation. Only because of the kinds
• 1:24
  Speaker 3: of nerds I hang out with.
• 1:27
  Speaker 1: You ever wonder if the flamingos are simulated?
• 1:29
  Speaker 3: Whoa, I hope they are. Actually, I hope they're robots,
• 1:32
  Speaker 3: you know, collecting data on us. Nobody would suspect it.
• 1:36
  Speaker 1: Hi, I'm Daniel. I'm a particle physicist, and I write
• 1:38
  Speaker 1: books about aliens, and I make heavy, heavy use of
• 1:42
  Speaker 1: simulations in my science.
• 1:44
  Speaker 3: Oh, I have done some simulations for my work as well.
• 1:47
  Speaker 3: But my question for you, Daniel, is if tomorrow you
• 1:50
  Speaker 3: received incontrovertible, oh, good word, incontrovertible.
• 1:56
  Speaker 1: Inspellable.
• 1:58
  Speaker 3: If you found out you definitely live in a simulation, 100%,
• 2:04
  Speaker 3: what would you change about your life?
• 2:06
  Speaker 1: Wow.
• 2:06
  Speaker 4: Wow.
• 2:07
  Speaker 1: Am I given access to the source code so I
• 2:09
  Speaker 1: can change? Or is it just the kind of changes
• 2:11
  Speaker 1: one could normally make, like eat a sundae every day?
• 2:14
  Speaker 3: Yes, that kind of thing.
• 2:15
  Speaker 1: All right. So I don't get godlike powers in this question, Kelly?
• 2:19
  Speaker 4: Nope.
• 2:19
  Speaker 1: Nope. Sorry.
• 2:20
  Speaker 3: You'd kill us all. You'd bring the aliens and you'd
• 2:24
  Speaker 3: trade us all for parts.
• 2:25
  Speaker 2: That's true.
• 2:26
  Speaker 1: No, if I had access to the source code, I
• 2:27
  Speaker 1: wouldn't need the aliens. I'd have the source code.
• 2:29
  Speaker 4: Oh.
• 2:31
  Speaker 1: But it's a good question. You know, if I'm limited
• 2:34
  Speaker 1: to living in this universe anyway with its rules, then
• 2:38
  Speaker 1: I guess I would probably make a lot of the
• 2:39
  Speaker 1: same decisions I make. But really, your question is like, yeah,
• 2:43
  Speaker 1: but you're always saying if you understood the context of
• 2:46
  Speaker 1: our lives, it could change how you live your life.
• 2:48
  Speaker 1: I'm always saying that, which is true. So if I
• 2:52
  Speaker 1: got a big update on the context of our lives,
• 2:54
  Speaker 1: would that make me change how I lived it? I
• 2:57
  Speaker 1: don't think so. I think I live anyway with the
• 3:01
  Speaker 1: uncertainty of are the people around me actually real? Do
• 3:04
  Speaker 1: they have a first person experience or are they philosophical zombies?
• 3:08
  Speaker 1: And I don't think the answer matters to me because
• 3:10
  Speaker 1: I feel like I love them and I feel their
• 3:12
  Speaker 1: pain and their joy. And that matters to me, even
• 3:15
  Speaker 1: if they're philosophical zombies or real flesh and blood creatures
• 3:20
  Speaker 1: or simulation artifacts.
• 3:22
  Speaker 2: All right.
• 3:23
  Speaker 1: How's that for an answer?
• 3:24
  Speaker 3: Yeah, no, that's a good answer. It also revealed that
• 3:28
  Speaker 3: you think about stuff like this often, it sounds like.
• 3:32
  Speaker 2: I don't.
• 3:33
  Speaker 3: I'm just like, I'm doing what I need to do.
• 3:35
  Speaker 3: Here's my to-do list. I got to get through it.
• 3:37
  Speaker 3: Simulation or not, I want to get this stuff done.
• 3:40
  Speaker 3: I don't know.
• 3:40
  Speaker 1: I think you are. You just don't realize it. Like,
• 3:42
  Speaker 1: aren't you always thinking about, like, what are my goats feeling?
• 3:45
  Speaker 1: Do they want to be crushed or not want to
• 3:47
  Speaker 1: be crushed?
• 3:49
  Speaker 4: Oh, no.
• 3:49
  Speaker 3: Back to that. But does that... Would that matter? I
• 3:55
  Speaker 3: guess I'm not thinking, is their pain real or are
• 3:58
  Speaker 3: we all just source code? I'm thinking this is a
• 4:02
  Speaker 3: real universe that I live in and there are decisions
• 4:05
  Speaker 3: I need to make that will determine whether or not
• 4:07
  Speaker 3: I look back at the end of my life and think,
• 4:09
  Speaker 3: did I live a good life or not? But I
• 4:11
  Speaker 3: never think the source code is making these decisions or something.
• 4:16
  Speaker 1: All right. Well, if you found out that your goats
• 4:18
  Speaker 1: were just source code, would you treat them differently?
• 4:22
  Speaker 4: No.
• 4:22
  Speaker 3: No, it's a really good simulation. I think I might
• 4:27
  Speaker 3: enjoy life less if I knew that. So I think
• 4:29
  Speaker 3: I'd be like, all right, take that information away from
• 4:32
  Speaker 3: me if you can. I'm just going to keep going
• 4:35
  Speaker 3: on and enjoying doing what I'm doing because I can't
• 4:37
  Speaker 3: get out of this simulation. And so I'm just moving forward.
• 4:41
  Speaker 1: All right. Well, today in the episode, we're not just
• 4:43
  Speaker 1: talking about simulating Kelly's goats. And we're not talking about
• 4:46
  Speaker 1: simulating particles of the Large Hadron Collider. We're talking about
• 4:49
  Speaker 1: simulating the whole universe. Can we, in fact, describe the
• 4:55
  Speaker 1: whole universe? Why would anybody ever want to? What can
• 4:58
  Speaker 1: we learn from it? How close are we to being
• 5:01
  Speaker 1: able to do this? And this question is not just
• 5:04
  Speaker 1: something Daniel sits around and thinks about while not smoking
• 5:07
  Speaker 1: banana peels. while Kelly is blissfully free of these philosophical questions.
• 5:12
  Speaker 1: I think.
• 5:15
  Speaker 3: We've discovered that I'm not as philosophically curious as you are.
• 5:18
  Speaker 3: This isn't a good thing. But the good thing is
• 5:20
  Speaker 3: that if we do live in a simulation, we live
• 5:23
  Speaker 3: in a simulation filled with curious listeners who share their
• 5:27
  Speaker 3: thoughts and hopes and dreams and questions with us.
• 5:29
  Speaker 1: That's true. And in real or simulated Ann Arbor is Joe,
• 5:34
  Speaker 1: who sent me this question about a recent universe simulation advance.
• 5:38
  Speaker 1: Here's the question from Joe.
• 5:40
  Speaker 4: Hi, Daniel and Kelly, but unfortunately, mostly Daniel. This is
• 5:45
  Speaker 4: Joe from Ann Arbor, and I wanted to know more
• 5:48
  Speaker 4: about the Flamingo project done by the Virgo Consortium. They
• 5:53
  Speaker 4: apparently used some wild amount of computing power to simulate
• 5:56
  Speaker 4: the evolution of the universe, and it includes both gravitation
• 6:00
  Speaker 4: and hydrodynamics. Apparently, there are some results about clumpiness that
• 6:05
  Speaker 4: I can help answer, but nothing went on to enough
• 6:08
  Speaker 4: detail to satisfy my curiosity.
• 6:10
  Speaker 3: Could you help?
• 6:11
  Speaker 1: So Joe was talking about a really exciting new result,
• 6:14
  Speaker 1: hilariously called FLAMINGO, in another example of tortured science acronyms.
• 6:20
  Speaker 3: Oh no, it's an acronym? I didn't realize that from
• 6:23
  Speaker 3: the outline. You're going to have to tell me what
• 6:24
  Speaker 3: the acronym stands for when we get to it, but congratulations, physicists.
• 6:28
  Speaker 1: Ha ha ha!
• 6:30
  Speaker 3: I know.
• 6:31
  Speaker 1: They called it flamingo. It has nothing to do with flamingos.
• 6:34
  Speaker 3: Oh, I thought surely the universe is pink. There was
• 6:37
  Speaker 3: going to be something in there.
• 6:39
  Speaker 1: And, you know, I reached out to our listeners to
• 6:41
  Speaker 1: ask them about simulating the universe. And I thought about
• 6:43
  Speaker 1: including flamingo in the question, but I was pretty sure
• 6:46
  Speaker 1: it was just going to lead them all astray because
• 6:49
  Speaker 1: the science actually has nothing to do with pink birds
• 6:52
  Speaker 1: that eat shrimp.
• 6:53
  Speaker 3: And they trust us, so we shouldn't lead them astray.
• 6:56
  Speaker 1: That's right.
• 6:56
  Speaker 4: Yes.
• 6:57
  Speaker 1: Instead, I asked folks the question behind the question, which is,
• 7:01
  Speaker 1: can we simulate the whole universe? Think about it for
• 7:06
  Speaker 1: a minute. And if you would like to add your
• 7:08
  Speaker 1: voice to this chorus of curious contributors, we would love
• 7:12
  Speaker 1: to have you please write to us to questions at danielandkelly.org.
• 7:16
  Speaker 1: And you'll get a bunch of weird questions in your
• 7:18
  Speaker 1: inbox that you can either respond to or not. Up
• 7:21
  Speaker 1: to you. In the meantime, think about it for yourself
• 7:24
  Speaker 1: for a minute. Do you think it's possible to simulate
• 7:26
  Speaker 1: the whole universe? Here's what our clever listeners had to say.
• 7:30
  Speaker 1: I guess the question is, how good do you want
• 7:35
  Speaker 1: the simulation to be? I strongly doubt that we can
• 7:38
  Speaker 1: actually simulate the entire universe with any medium smaller than
• 7:43
  Speaker 1: a universe.
• 7:45
  Speaker 3: Yes, I think it would be possible with a supercomputer.
• 7:48
  Speaker 1: Could any advanced civilization simulate the universe? Then maybe yes,
• 7:52
  Speaker 1: maybe we do live in a simulated universe. One, we
• 7:55
  Speaker 1: don't even know what the whole universe is. Two, starting conditions,
• 7:59
  Speaker 1: we've got to jump in at some point to get
• 8:02
  Speaker 1: the simulation right. And three, it's just too much. Too much.
• 8:06
  Speaker 1: We would need to simulate the simulation that we are running,
• 8:10
  Speaker 1: so it will be an endless recursion.
• 8:13
  Speaker 3: Things have to be less.
• 8:14
  Speaker 4: complicated in the simulation than they are in reality?
• 8:17
  Speaker 1: It would be something like Space Engine, the game.
• 8:20
  Speaker 4: No, we can't simulate the whole universe because we don't
• 8:23
  Speaker 4: know anything about most of the universe.
• 8:26
  Speaker 3: Probably not with the technology humans have right now. Well,
• 8:30
  Speaker 3: the listener's gut reaction pretty much agrees. I was going
• 8:34
  Speaker 3: to say pretty much agrees with mine, but then someone
• 8:36
  Speaker 3: said yes on a supercomputer. And I am not convinced
• 8:40
  Speaker 3: that we have the computing power to simulate the entire universe.
• 8:43
  Speaker 3: But maybe, maybe the question is like, on what level? Like,
• 8:48
  Speaker 3: are you just simulating the movement of some planets? Is
• 8:51
  Speaker 3: that what you mean? Or are we talking about like
• 8:53
  Speaker 3: down to the level of atoms? Because that sounds nuts.
• 8:55
  Speaker 2: Yeah.
• 8:57
  Speaker 1: Even on a supercomputer? What about a super-duper computer? Whoa, whoa.
• 9:02
  Speaker 3: I don't know. Do we have super-duper computers yet? How
• 9:05
  Speaker 3: far out is that technology, Daniel?
• 9:08
  Speaker 1: I don't know, but now I want to write a
• 9:09
  Speaker 1: grant proposal and use that phrase, super-duper computers.
• 9:15
  Speaker 3: I would fund that grant, no doubt.
• 9:17
  Speaker 1: Or maybe, you know, hyper-computers. What other prefixes could we add?
• 9:21
  Speaker 3: Hyper-super-duper computers. Hyper, super, duper, fragilistic.
• 9:25
  Speaker 1: Computers. Queen of the Latin names. I thought you were
• 9:30
  Speaker 1: going to go with something more scientific sounding, but I
• 9:33
  Speaker 1: like that.
• 9:34
  Speaker 3: No, no, I'm having fun this morning. So let's start
• 9:37
  Speaker 3: by talking about what is a simulation. We've talked about
• 9:40
  Speaker 3: this before, but I think let's do like a brief
• 9:42
  Speaker 3: explanation of what a simulation is and then dig into
• 9:44
  Speaker 3: like what level of detail is really being asked for
• 9:48
  Speaker 3: in this question.
• 9:49
  Speaker 1: Yeah, I think that's a good idea because you already
• 9:51
  Speaker 1: raised a lot of the issues that we need to tackle.
• 9:54
  Speaker 1: How much do you need to simulate? Why do you
• 9:56
  Speaker 1: do it? Et cetera, et cetera. So what is a simulation?
• 10:00
  Speaker 1: It's some kind of program run on a computer that
• 10:03
  Speaker 1: uses usually step-by-step methods to describe the approximate, very keyword,
• 10:09
  Speaker 1: approximate behavior of a mathematical model which resembles a physical system. Okay.
• 10:15
  Speaker 1: So let's say, for example, you want to describe a
• 10:18
  Speaker 1: ball getting thrown across your backyard, and you want to
• 10:21
  Speaker 1: know where it's going to land. And so you can
• 10:23
  Speaker 1: change the velocity or change the angle, and you want
• 10:25
  Speaker 1: to know where is this ball going to go. Is
• 10:27
  Speaker 1: it going to go into my friend's hand, or am
• 10:29
  Speaker 1: I going to miss them or something? So the first
• 10:31
  Speaker 1: thing you can do is say, well, I'm going to
• 10:32
  Speaker 1: make it very, very simple. There's just gravity, and there's
• 10:35
  Speaker 1: just the ball, and there's no air resistance or anything,
• 10:39
  Speaker 1: in which case you can just calculate it. You can
• 10:41
  Speaker 1: use Newton's laws. You can say F equals MA. It's
• 10:44
  Speaker 1: a very simple problem. Boom, you can do it in
• 10:46
  Speaker 1: one step. You can do it on a computer. You
• 10:48
  Speaker 1: can also just do it on pencil and paper. Either way,
• 10:50
  Speaker 1: it's still technically a simulation because you've taken your physical system,
• 10:55
  Speaker 1: the ball and the backyard, and you described it with
• 10:58
  Speaker 1: a mathematical model, right? You said there's mass, there's gravity,
• 11:02
  Speaker 1: there's a few components. Here's how they interact. There are
• 11:05
  Speaker 1: rules for how things happen. And then you've used your
• 11:08
  Speaker 1: mathematical model to make a prediction about the actual physical system,
• 11:11
  Speaker 1: the thing in your backyard. So the mathematical model is
• 11:14
  Speaker 1: the simulation of the thing in your backyard. And if
• 11:18
  Speaker 1: everything is very, very simple, the model can be very,
• 11:20
  Speaker 1: very simple. Good so far?
• 11:22
  Speaker 3: We are.
• 11:22
  Speaker 1: All right. So maybe you say, well, but it gets
• 11:25
  Speaker 1: it wrong. It doesn't exactly drop where I predict. You think, well,
• 11:28
  Speaker 1: that's probably because my simulation is too simple. I'm assuming
• 11:32
  Speaker 1: there's no air resistance, that there's no wind, that this
• 11:35
  Speaker 1: is a very simple problem. And so if I want
• 11:37
  Speaker 1: a more accurate answer, then I add more to my simulation.
• 11:41
  Speaker 1: And some people might not. They say, I'm fine with
• 11:43
  Speaker 1: this answer. This is all I need. And already we've
• 11:46
  Speaker 1: learned something crucial, which is how much detail you add,
• 11:50
  Speaker 1: how much accuracy you have depends on the question you're asking, right?
• 11:55
  Speaker 1: And if you are asking a question that doesn't need
• 11:57
  Speaker 1: to be very, very precise, then you're done. You don't
• 11:59
  Speaker 1: need to add bells and whistles and simulate everything in
• 12:02
  Speaker 1: great detail. But if somebody really wants a super precise
• 12:05
  Speaker 1: description for whatever reason, then they got to add more.
• 12:08
  Speaker 1: And in physics, we're always playing this game. We're saying,
• 12:11
  Speaker 1: What's the simplest, useful model? Because adding more costs resources
• 12:15
  Speaker 1: and brain power and everything. So let's say you needed more, right?
• 12:18
  Speaker 1: You wanted to also include air resistance. Well, you can
• 12:22
  Speaker 1: add that to Newton's laws and you can make that
• 12:25
  Speaker 1: work and you have a more complicated equation to solve,
• 12:28
  Speaker 1: but you can still get it to work. But what
• 12:30
  Speaker 1: if you want to also add like the pattern of
• 12:32
  Speaker 1: wind across your backyard and you set up a bunch
• 12:34
  Speaker 1: of sensors to measure like exactly what is the flow
• 12:37
  Speaker 1: of wind across my backyard and you measure this in
• 12:40
  Speaker 1: great detail. You want to incorporate that into your model. Well,
• 12:43
  Speaker 1: now you don't have a simple equation anymore, but you
• 12:46
  Speaker 1: can still solve something. You can say, well, I'm going
• 12:49
  Speaker 1: to move the ball one centimeter at a time, and
• 12:52
  Speaker 1: I'm going to count for the wind at that location.
• 12:54
  Speaker 1: I'm going to let it push. And so I don't
• 12:56
  Speaker 1: have a single equation that's going to describe what happens
• 12:59
  Speaker 1: to the ball across the whole backyard, but I can
• 13:01
  Speaker 1: slice it in time and solve an approximate version of
• 13:04
  Speaker 1: it each step and I can measure where the ball
• 13:07
  Speaker 1: goes and I can correct it. And now I have
• 13:08
  Speaker 1: a very elaborate, very complicated simulation of something pretty simple,
• 13:13
  Speaker 1: just throwing a ball across your backyard. And so all
• 13:16
  Speaker 1: of those are simulations, but they're at varying levels of fidelity, right?
• 13:20
  Speaker 1: The model becomes more and more accurate, becomes closer and
• 13:23
  Speaker 1: closer to the physical system. And so its answers are
• 13:26
  Speaker 1: more precise.
• 13:27
  Speaker 4: Got it.
• 13:27
  Speaker 3: And is every mathematical model a simulation?
• 13:32
  Speaker 1: I think a simulation is using the mathematical model to
• 13:35
  Speaker 1: make a prediction and that can be on a computer,
• 13:38
  Speaker 1: but it can also be a physical system itself, right?
• 13:42
  Speaker 1: Like you can build a little model, like a physical
• 13:46
  Speaker 1: model of like how water flows in a river. You
• 13:49
  Speaker 1: can make a miniature version of it and you can
• 13:51
  Speaker 1: pour water into it and see what happens, right? That's
• 13:55
  Speaker 1: a simulation. Like you can simulate the Hoover Dam, right?
• 13:58
  Speaker 1: You know, with a small version of it before you
• 14:00
  Speaker 1: build it to see like, hmm, have I misunderstood anything?
• 14:03
  Speaker 1: Is there anything I'm missing? Right. And so you can
• 14:05
  Speaker 1: use the universe to simulate other parts of the universe.
• 14:08
  Speaker 1: Or you can do it on a computer or you
• 14:10
  Speaker 1: can do it on pencil and paper, which is how
• 14:11
  Speaker 1: they used to do it. Like in the beginning, when
• 14:13
  Speaker 1: they were trying to predict the weather, it was like
• 14:16
  Speaker 1: a whole bunch of calculations, pencil and paper. And to
• 14:18
  Speaker 1: predict the weather six hours in advance took like six weeks.
• 14:22
  Speaker 3: Which is not helpful anymore. And folks should check out
• 14:24
  Speaker 3: the amazing weather episode that you did to see how
• 14:27
  Speaker 3: we ended up getting a lot better at that.
• 14:29
  Speaker 1: Yeah. And there's other examples, like sometimes we want to
• 14:32
  Speaker 1: understand what happens in black holes. There's certain relationships between
• 14:36
  Speaker 1: what happens around a black hole and what happens in
• 14:39
  Speaker 1: certain fluids that are spinning really, really fast. And so...
• 14:43
  Speaker 1: For example, people have made like sonic black holes as
• 14:46
  Speaker 1: an analog of black holes in order to try to
• 14:48
  Speaker 1: understand black holes. They're not actual black holes you've built
• 14:51
  Speaker 1: in the laboratory. Not yet, unfortunately.
• 14:54
  Speaker 3: Unfortunately. Oh my gosh, Daniel. You are looking for ways
• 14:58
  Speaker 3: to kill us.
• 14:59
  Speaker 1: No, I just want to learn about the universe.
• 15:02
  Speaker 3: Humans be darned.
• 15:03
  Speaker 1: I will learn about the universe. Consequences irrelevant. But you
• 15:07
  Speaker 1: can build something which has a mathematical relationship like If
• 15:10
  Speaker 1: the mathematics of your fluids and sound waves are similar
• 15:14
  Speaker 1: to the mathematics of general relativity, then by building this system,
• 15:17
  Speaker 1: you can learn something about black holes as described by
• 15:20
  Speaker 1: general relativity. So you're always making this analogy. You're saying,
• 15:23
  Speaker 1: I want to learn about X. I'm going to build
• 15:25
  Speaker 1: system Y, which is similar to it in the ways
• 15:28
  Speaker 1: that's important to me. But you're never going to get
• 15:30
  Speaker 1: all the details right. Even if you describe the air
• 15:33
  Speaker 1: molecules in your backyard, there's still something you don't know.
• 15:36
  Speaker 1: There's uncertainty in the measurements of those things. You're making assumptions.
• 15:40
  Speaker 1: about how you move your simulation forward in time. There's
• 15:43
  Speaker 1: always going to be some imprecision. You can never get
• 15:45
  Speaker 1: everything exactly right. But often you don't have to because
• 15:49
  Speaker 1: you're asking a science question. You're not just like, I
• 15:51
  Speaker 1: must describe the universe at the most granular level. People
• 15:55
  Speaker 1: don't simulate just to simulate. People simulate to answer questions like,
• 15:58
  Speaker 1: how much dark matter do we need in the universe?
• 16:00
  Speaker 1: Or how important is air resistance to throwing a ball?
• 16:04
  Speaker 1: Or can I build this dam or will it collapse?
• 16:07
  Speaker 1: Or what happens if I throw a goat into a
• 16:09
  Speaker 1: black hole, right? People have a science question they want
• 16:11
  Speaker 1: an answer to and they use simulation to answer that question.
• 16:15
  Speaker 3: No, no spaghettifying my goats.
• 16:19
  Speaker 1: Goats like spaghetti, Kelly. That's what I hear.
• 16:21
  Speaker 2: Do they?
• 16:22
  Speaker 3: Who are you talking to? Because my goats are very
• 16:25
  Speaker 3: picky eaters.
• 16:26
  Speaker 1: What? I thought goats would eat anything. Don't they eat
• 16:28
  Speaker 1: like cans and carb bumpers and stuff?
• 16:30
  Speaker 3: This is why stereotypes are harmful, Daniel.
• 16:32
  Speaker 1: Oh my gosh.
• 16:33
  Speaker 3: You can't just drop them in a junkyard. And watch
• 16:36
  Speaker 3: them thrive and flourish. No, they don't eat cans.
• 16:39
  Speaker 1: I've just been slandering goats. Oh my gosh.
• 16:42
  Speaker 3: I mean, that's what I've been saying. So, okay. So
• 16:46
  Speaker 3: what you're saying is we can specify narrow questions. We can...
• 16:51
  Speaker 3: answer them accurately. We decide what we want to have
• 16:55
  Speaker 3: in the model and what we leave out. And sometimes
• 16:57
  Speaker 3: we even leave out important stuff just because we're like, well,
• 17:00
  Speaker 3: you know what? We don't have enough computing power.
• 17:02
  Speaker 4: Yes.
• 17:03
  Speaker 3: What would you say is like the biggest simulation or
• 17:06
  Speaker 3: biggest thing we've ever tried to simulate?
• 17:09
  Speaker 1: Bigger than the universe? I don't know.
• 17:11
  Speaker 3: Or most detailed.
• 17:12
  Speaker 1: You know, in particle physics, we simulate all sorts of
• 17:15
  Speaker 1: stuff down to the particle level. You know, what happens
• 17:19
  Speaker 1: when a muon slams into our detector and it's made
• 17:22
  Speaker 1: of copper and there's layers of uranium there. We model
• 17:26
  Speaker 1: all of those details down to the atom. The muon
• 17:29
  Speaker 1: emits a photon, which then slams into a nucleus, which
• 17:32
  Speaker 1: then emits this particle, which then emits that particle. And
• 17:35
  Speaker 1: that's why our simulations are so computationally expensive and so slow.
• 17:40
  Speaker 1: It can take like tens of minutes to simulate an
• 17:43
  Speaker 1: individual collision at the Large Hadron Collider. Because we got
• 17:47
  Speaker 1: to track all those details because we're interested in like, well,
• 17:50
  Speaker 1: how many photons ended up on this one part of
• 17:54
  Speaker 1: this one detector, because that's what we measure. And we
• 17:57
  Speaker 1: want to compare the simulations to reality and we can't
• 18:00
  Speaker 1: skip any details. So from my point of view, those
• 18:02
  Speaker 1: are some of the most involved simulations. But I'm also
• 18:05
  Speaker 1: really in love with these physical simulation models. Like there's
• 18:09
  Speaker 1: this working hydraulic scale model of the San Francisco Bay
• 18:14
  Speaker 1: and the San Joaquin River Delta system that the Army
• 18:16
  Speaker 1: Corps of Engineers built in the 50s to study engineering
• 18:20
  Speaker 1: interventions in the bay, like what to build and how
• 18:23
  Speaker 1: to change the currents and stuff like that. It's super awesome.
• 18:26
  Speaker 3: Oh, I worked on the Sacramento-San Joaquin River Delta systems
• 18:30
  Speaker 3: pod problem.
• 18:31
  Speaker 1: Oh my gosh. What's the problem?
• 18:32
  Speaker 3: The pod problem is pelagic organismal decline.
• 18:36
  Speaker 1: There are all of these- Oh, it's nothing to do
• 18:37
  Speaker 1: with podcasts? No, I mean- Like we don't have enough
• 18:40
  Speaker 1: podcasts about the river system?
• 18:42
  Speaker 3: I mean, probably we could use a podcast about the
• 18:44
  Speaker 3: river system, but no, the fish that lived in like
• 18:47
  Speaker 3: the big open water area, they used to dominate this ecosystem.
• 18:51
  Speaker 3: But over time they started disappearing and people were trying
• 18:54
  Speaker 3: to figure out what was the cause of the loss
• 18:56
  Speaker 3: of all of these native fish. And I was trying
• 18:58
  Speaker 3: to figure out if the problem was that A bunch
• 19:01
  Speaker 3: of invasive species had been added to the system. And
• 19:03
  Speaker 3: so now you've got these invasive plants and largemouth bass
• 19:06
  Speaker 3: and all these other fish in the system that didn't
• 19:09
  Speaker 3: used to belong there. And so I spent a significant
• 19:12
  Speaker 3: amount of my PhD on boats riding around the river
• 19:14
  Speaker 3: Delta system. And I kind of on some of those
• 19:18
  Speaker 3: days would have preferred if I could have just been
• 19:20
  Speaker 3: moving a boat around a little hydraulic model and being like,
• 19:23
  Speaker 3: this is the same thing.
• 19:24
  Speaker 1: Yeah. Yeah, and a simulation could really help you in
• 19:29
  Speaker 1: that kind of situation. You have a hypothesis like, I
• 19:31
  Speaker 1: think this is because of this endangered system. How would
• 19:34
  Speaker 1: you test that hypothesis? Well, if you had access to
• 19:36
  Speaker 1: a simulation of the system and you could, in simulation,
• 19:40
  Speaker 1: inject your invasive species and see what happens and see
• 19:44
  Speaker 1: if it lines up with reality, then you could confirm
• 19:48
  Speaker 1: or reject that hypothesis immediately. without doing any actual physical experiments.
• 19:52
  Speaker 1: These are like virtual experiments that help us understand what
• 19:56
  Speaker 1: might be happening in the real world.
• 19:58
  Speaker 3: And ecologists do this kind of thing all the time.
• 20:00
  Speaker 3: And just like you were saying, you know, the goal
• 20:02
  Speaker 3: is to try to figure out what are the important
• 20:04
  Speaker 3: features of the system that we need to understand better?
• 20:06
  Speaker 3: What can we ignore? How can we sort of tinker
• 20:09
  Speaker 3: with the system to get it in a better state? Yeah,
• 20:12
  Speaker 3: these are all things ecologists do as well.
• 20:14
  Speaker 1: And those are great examples of the kind of questions
• 20:17
  Speaker 1: you can ask of a simulation. The reason a simulation
• 20:20
  Speaker 1: is useful is is that you have higher level questions,
• 20:22
  Speaker 1: emergent questions, right? Like, why did this species die off?
• 20:26
  Speaker 1: Or where did this ball go? And you don't know
• 20:30
  Speaker 1: how to model that directly. You can't write down equations
• 20:32
  Speaker 1: to tell you exactly what's going to happen. It's too complicated. Instead,
• 20:36
  Speaker 1: what you can do is model the lower level stuff.
• 20:39
  Speaker 1: This fish will eat that fish. This air molecule will
• 20:42
  Speaker 1: bounce off the ball. That lower level stuff we can model.
• 20:45
  Speaker 1: And then if we can describe enough of it, we hope,
• 20:47
  Speaker 1: that the higher level questions emerge from that system, right?
• 20:51
  Speaker 1: So simulation is very useful when you have a handle
• 20:55
  Speaker 1: on the lower level details and you have questions about
• 20:57
  Speaker 1: the bigger picture, but you don't have equations that describe it.
• 21:01
  Speaker 1: In the same way that like on the weather episode,
• 21:02
  Speaker 1: we talked about we're interested in, is that hurricane going
• 21:05
  Speaker 1: to hit Alabama? Yes or no. And we don't have
• 21:08
  Speaker 1: an equation that can tell us that. Instead, we have
• 21:10
  Speaker 1: equations that tell us what water droplets do in certain
• 21:13
  Speaker 1: velocities of wind. So we can model a bunch of
• 21:16
  Speaker 1: water droplets in simulation and get answers to the bigger
• 21:19
  Speaker 1: questions that we have. But in order to do that,
• 21:22
  Speaker 1: simulation only works when you have two things. One, you
• 21:25
  Speaker 1: understand the lower level details, like you can describe the
• 21:28
  Speaker 1: physics of water droplets or air molecules or fish eating
• 21:33
  Speaker 1: fish or whatever. And this is probably what Kelly was
• 21:36
  Speaker 1: involved in, the data, right? This only works, you can
• 21:40
  Speaker 1: only describe what's happening in your backyard if you know
• 21:42
  Speaker 1: the wind patterns. You can only Describe what's happening in
• 21:45
  Speaker 1: your river delta. If you know the crucial information, the
• 21:48
  Speaker 1: shapes of it, the current flows, where the fish are,
• 21:50
  Speaker 1: for example. So you need the initial conditions and you
• 21:53
  Speaker 1: need to know how things evolve. Those are two crucial
• 21:56
  Speaker 1: ingredients in making a simulation.
• 21:58
  Speaker 3: Well, I managed to get fish into this conversation. So
• 22:00
  Speaker 3: I know that no one is simulating being interested in
• 22:03
  Speaker 3: what we're talking about. And so let's take a break.
• 22:05
  Speaker 3: And when we get back, we will talk more about simulations.
• 22:10
  Speaker 1: Thank you.
• 22:29
  Speaker 3: And we're back. OK, so we are talking about simulations.
• 22:32
  Speaker 3: And Daniel, I have heard that quantum computers are going
• 22:37
  Speaker 3: to help us simulate things in better ways. Quantum computers,
• 22:41
  Speaker 3: I think when you listen to pop articles about quantum computers,
• 22:44
  Speaker 3: they're able to do all kinds of amazing things. And
• 22:46
  Speaker 3: I'm not actually sure that they're able to do all
• 22:48
  Speaker 3: of the things that they are said to be able
• 22:50
  Speaker 3: to do.
• 22:51
  Speaker 1: And so you don't think they can quantum balance your checkbook?
• 22:55
  Speaker 3: I do a fine job at– you know what? Actually,
• 22:57
  Speaker 3: I don't have a checkbook. I just every once in
• 22:59
  Speaker 3: a while open up my bank account, and I'm like,
• 23:00
  Speaker 3: that's what's happening there.
• 23:02
  Speaker 1: You're like, oh, I hope the number's not negative. Yay,
• 23:04
  Speaker 1: it's not negative.
• 23:05
  Speaker 3: Still not negative. We don't buy a lot of stuff.
• 23:08
  Speaker 3: But anyway, so, okay, so would quantum computing make it
• 23:12
  Speaker 3: easier for us to simulate the universe, or is it
• 23:15
  Speaker 3: essentially no different than a classical computer in this regard?
• 23:18
  Speaker 1: I mean, yes and no. The no is because anything
• 23:22
  Speaker 1: a quantum computer can do, a classical computer can also do, right?
• 23:25
  Speaker 4: Okay.
• 23:26
  Speaker 1: So there's no problems that quantum computers can solve that
• 23:28
  Speaker 1: classical computers can't. Can they do it faster? And so
• 23:33
  Speaker 1: not in general. It's not like quantum computers are faster
• 23:36
  Speaker 1: at everything than classical computers. You know, there are a
• 23:39
  Speaker 1: few cases where we think quantum computers could be faster
• 23:43
  Speaker 1: than classical computers. But so far, those are like really,
• 23:47
  Speaker 1: really specialized cases, you know, like one factoring algorithm here
• 23:51
  Speaker 1: or another mathematical algorithm there. So, so far, it's still
• 23:54
  Speaker 1: like at the, ooh, this would be cool. And maybe
• 23:58
  Speaker 1: it's the harbinger of things to come. And eventually, everybody
• 24:00
  Speaker 1: will have a quantum computer on their wrist. That would
• 24:03
  Speaker 1: be cool. And it's exciting because it's like another way
• 24:06
  Speaker 1: to do computing, which is fascinating from a philosophical point
• 24:09
  Speaker 1: of view. But practically, it's not going to speed up
• 24:12
  Speaker 1: our simulations of the Large Hadron Collider. It's not going
• 24:14
  Speaker 1: to help us predict the weather. It's not going to
• 24:15
  Speaker 1: balance your checkbook. There is one way in which it's cool.
• 24:19
  Speaker 1: And this is one of the initial, I think, impetuses
• 24:22
  Speaker 1: for quantum computing, which is, look, a lot of the
• 24:25
  Speaker 1: physics we do is quantum, right? And so doesn't it
• 24:28
  Speaker 1: make more sense to have a quantum computer to model
• 24:31
  Speaker 1: quantum processes rather than a classical computer? In some sense,
• 24:35
  Speaker 1: the universe is quantum, so why are we building classical
• 24:38
  Speaker 1: computers onto it to then model quantum processes? There are
• 24:44
  Speaker 1: ideas about building mini-quantum systems to describe those quantum systems,
• 24:48
  Speaker 1: the same way you build a mini-river delta with real
• 24:52
  Speaker 1: water to describe the real water in the actual river delta.
• 24:56
  Speaker 1: You could build a little mini-quantum system that has quantum
• 24:59
  Speaker 1: properties and help you model a bigger or more complicated
• 25:02
  Speaker 1: quantum system you're interested in. But it's not a general
• 25:05
  Speaker 1: overall simulation. It's not going to be instantaneous or anything
• 25:08
  Speaker 1: like that. Okay.
• 25:10
  Speaker 3: And just to sort of bottom line the problem that
• 25:12
  Speaker 3: we're trying to solve here, when someone says we're going
• 25:14
  Speaker 3: to simulate the universe, do they mean like at the
• 25:18
  Speaker 3: quantum level, like everything at the quantum level and up?
• 25:22
  Speaker 3: Is that what you would need to create a simulated universe?
• 25:24
  Speaker 3: Or could you... Create a simulation at like a higher level,
• 25:28
  Speaker 3: but just let the quantum stuff happen to produce the
• 25:32
  Speaker 3: higher level stuff without modeling it. Does that make sense?
• 25:35
  Speaker 1: Yeah. Well, the answer is always you're doing a simulation
• 25:38
  Speaker 1: to answer a specific question. Nobody's out there being like,
• 25:42
  Speaker 1: let's simulate the whole universe, man, just because. I mean,
• 25:46
  Speaker 1: it would be massively complicated and inexpensive.
• 25:48
  Speaker 4: Right.
• 25:48
  Speaker 3: But isn't that what we're talking about today? We're not
• 25:50
  Speaker 3: talking about trying to simulate the whole universe?
• 25:52
  Speaker 2: I mean it.
• 25:53
  Speaker 1: No, well, we are because we ask questions about the universe.
• 25:56
  Speaker 1: Like we ask, how old is the universe? Or how
• 25:59
  Speaker 1: did galaxies form? And will they continue to form? And
• 26:03
  Speaker 1: why are galaxies distributed in this way and that way?
• 26:05
  Speaker 1: Those are questions about the universe. And so- To answer
• 26:08
  Speaker 1: those questions, we need a simulation of the universe. That
• 26:11
  Speaker 1: simulation doesn't have to include everything in the universe, because
• 26:15
  Speaker 1: some details about the universe don't change the answers to that. Like,
• 26:19
  Speaker 1: if you bought your kid a red bicycle or a
• 26:21
  Speaker 1: blue bicycle for their ninth birthday, it doesn't change the
• 26:24
  Speaker 1: answer to, is Andromeda going to smash into the Milky Way? Or,
• 26:28
  Speaker 1: why is this galaxy bigger than the other galaxy? Your
• 26:31
  Speaker 1: kid's bike is awesome, but really it's kind of irrelevant
• 26:34
  Speaker 1: on the cosmic scale. And in the same way, lots
• 26:37
  Speaker 1: of stuff is irrelevant. And so we don't include it
• 26:39
  Speaker 1: in the simulation because we don't need it to answer
• 26:42
  Speaker 1: that question. So there's no general purpose simulation. So just
• 26:45
  Speaker 1: like simulate everything just because we're always building simulations to
• 26:49
  Speaker 1: answer science questions. And often it's the only way we
• 26:53
  Speaker 1: can answer a question. Like you want to know what's
• 26:56
  Speaker 1: going on inside the sun or at the heart of
• 26:58
  Speaker 1: neutron stars? We can't yet observe that stuff. And so
• 27:02
  Speaker 1: the only way to think about it is using simulations.
• 27:05
  Speaker 1: Also because we don't have equations that describe it. There's
• 27:07
  Speaker 1: no like F equals MA emergent physics that describes the
• 27:11
  Speaker 1: turbulence inside the sun. And so we can simulate it
• 27:15
  Speaker 1: and build models of it, but we don't have any
• 27:17
  Speaker 1: exact solutions.
• 27:18
  Speaker 3: Right, but that's the human limitation. If we were living
• 27:22
  Speaker 3: in a simulation created by aliens much smarter than us,
• 27:25
  Speaker 3: would their simulation have to include everything from the quantum
• 27:29
  Speaker 3: scale to your son's bike?
• 27:32
  Speaker 1: If we're living in a simulation run by super intelligent
• 27:36
  Speaker 1: or just hilarious aliens, then whatever their simulation includes is
• 27:43
  Speaker 1: our reality. And so really then you're asking me, are
• 27:47
  Speaker 1: they modeling everything down to the quantum level? Well, if
• 27:51
  Speaker 1: that exists, then they're modeling it, assuming our universe is
• 27:54
  Speaker 1: a simulation, right?
• 27:55
  Speaker 3: Okay. Yeah, I think so.
• 27:56
  Speaker 1: So the answer to that is yes. But we can
• 27:58
  Speaker 1: dig into that question at the very end after we
• 28:01
  Speaker 1: talk about the Flamingo Project. What?
• 28:03
  Speaker 3: All right, we're back to animals.
• 28:04
  Speaker 2: No, I know.
• 28:05
  Speaker 3: You told me it's a tortured acronym. What is the
• 28:08
  Speaker 3: tortured acronym here?
• 28:10
  Speaker 1: So the Flamingo Project stands for Full Hydro Large Scale
• 28:15
  Speaker 1: Structure Simulations with All-sky mapping for the interpretation of next-generation observations.
• 28:22
  Speaker 3: Holy cow. Wait, but I thought you said full hydros.
• 28:24
  Speaker 3: Was it the Flamingo Project? I lost track of the
• 28:28
  Speaker 3: rest of it after that.
• 28:29
  Speaker 1: No, no, you can just pick letters from your name
• 28:33
  Speaker 1: to make an acronym in physics, apparently. F from full hydro,
• 28:37
  Speaker 1: L from large-scale structure simulations with... A from all sky.
• 28:41
  Speaker 1: M from mapping. For the I from interpretation of G
• 28:45
  Speaker 1: from generation. O from observations. Flamingo. I told you it
• 28:49
  Speaker 1: has nothing to do with birds. Wow.
• 28:51
  Speaker 4: Okay.
• 28:52
  Speaker 3: Nothing to do with birds. And I think that actually
• 28:55
  Speaker 3: wins the most tortured acronym that I have come across award.
• 28:59
  Speaker 3: But anyway, I'm glad that they had animals on the brain.
• 29:01
  Speaker 1: I love an acronym that's tortured, but then also manages
• 29:05
  Speaker 1: to actually be related to to the topic at hand
• 29:08
  Speaker 1: you know like if it's just totally irrelevant then i
• 29:11
  Speaker 1: don't think it's very.
• 29:12
  Speaker 3: Impressive but to be flamingo doesn't have anything to do
• 29:15
  Speaker 3: with the topic at hand.
• 29:16
  Speaker 1: Right okay absolutely nothing other than there are flamingos in
• 29:19
  Speaker 1: the universe and they are simulating the universe but that's a.
• 29:21
  Speaker 3: Stretch okay fair enough and they're probably one of the
• 29:24
  Speaker 3: better parts of this universe because who doesn't love a
• 29:26
  Speaker 3: flamingo all right so what is this project working on.
• 29:30
  Speaker 1: This project wants to understand how did the universe get
• 29:33
  Speaker 1: the way that it is Like we look out in
• 29:35
  Speaker 1: the universe and we see there's structure there, right? There
• 29:38
  Speaker 1: are stars, there are galaxies, There are clusters of galaxies.
• 29:42
  Speaker 1: And there's dark matter laying behind all of it, right?
• 29:45
  Speaker 1: Everywhere there's a galaxy, almost there's a big halo of
• 29:49
  Speaker 1: dark matter surrounding it. There's dark matter between the galaxies.
• 29:53
  Speaker 1: There's streams of matter and dark matter between clusters of galaxies.
• 29:57
  Speaker 1: How did everything get into this configuration? Well, we have
• 30:01
  Speaker 1: an idea about that. We have these theories that 14
• 30:04
  Speaker 1: billion years ago-ish, the universe was filled with a hot,
• 30:08
  Speaker 1: dense plasma with small fluctuations in density. And we say, well, gravity.
• 30:13
  Speaker 1: Gravity clumped those things together. A little bit of density here,
• 30:16
  Speaker 1: a little bit of density there. Had more gravity, which
• 30:18
  Speaker 1: pulled on more things. Dot, dot, dot, you get galaxies. Well,
• 30:22
  Speaker 1: what we really want to do is flesh out the dot, dot,
• 30:24
  Speaker 1: dot part and understand, does our physics predict that when
• 30:29
  Speaker 1: you go from a universe filled with plasma with slight
• 30:32
  Speaker 1: over-densities and under-densities, that you get the large-scale structure that
• 30:36
  Speaker 1: we see. And if we run the simulation and it disagrees,
• 30:40
  Speaker 1: that tells you that either something is wrong with our
• 30:43
  Speaker 1: picture of the initial conditions of the universe, or something
• 30:46
  Speaker 1: is wrong with our understanding of how things evolve in time.
• 30:49
  Speaker 1: Super-duper useful, right? Because it gives you a handle on
• 30:53
  Speaker 1: something that's going on in the universe.
• 30:55
  Speaker 3: Yeah, but just to wrap, what does agreement look like? Like,
• 30:59
  Speaker 3: would agreement mean, like, every star, for example, needs to
• 31:02
  Speaker 3: be in exactly a predictable place? Or just, like, sort of,
• 31:05
  Speaker 3: it generally has the same shape? Like, what would we
• 31:08
  Speaker 3: consider to be a model that had simulated the universe?
• 31:11
  Speaker 1: Great question. And I think a lot of people imagine
• 31:13
  Speaker 1: that when you simulate the universe, you're predicting the outcome
• 31:16
  Speaker 1: for our exact universe. That you should, like, be able
• 31:19
  Speaker 1: to predict exactly the stars in the sky and this
• 31:21
  Speaker 1: galaxy and that galaxy.
• 31:22
  Speaker 4: Right.
• 31:23
  Speaker 1: And in principle, you might be able to do that
• 31:25
  Speaker 1: if you knew the exact initial conditions of our exact universe,
• 31:29
  Speaker 1: like where every particle was 14 billion years ago and
• 31:32
  Speaker 1: where all the densities were, et cetera. You might be
• 31:35
  Speaker 1: able to do that. We don't know that specifically. We
• 31:38
  Speaker 1: only know it statistically. Like we know on average, what
• 31:42
  Speaker 1: was the density of stuff? How big were the fluctuations, right?
• 31:46
  Speaker 1: We can't see the whole early universe. We can look
• 31:49
  Speaker 1: out in the sky and we can see light from
• 31:51
  Speaker 1: the very early universe. But we're just getting one slice
• 31:54
  Speaker 1: of that light. You know, the universe is 14 billion
• 31:57
  Speaker 1: years old. And so we're getting the light that has
• 31:59
  Speaker 1: just now arrived to us from that plasma in every direction.
• 32:03
  Speaker 1: Imagine like a huge but very thin shell of plasma
• 32:08
  Speaker 1: really far away 14 billion years ago, all emitting light
• 32:12
  Speaker 1: towards where we are now. We're here. It arrives here.
• 32:15
  Speaker 1: So we see that little slice of the universe. And
• 32:17
  Speaker 1: as time goes on, we see a different slice. That
• 32:19
  Speaker 1: shell grows as the time goes on. But we never
• 32:22
  Speaker 1: see the whole initial slice of the universe. So we
• 32:24
  Speaker 1: don't know enough to predict our specific universe. Instead, what
• 32:29
  Speaker 1: agreement looks like is, well, here's what we think the
• 32:31
  Speaker 1: distribution of stuff looked like statistically. So do we get
• 32:35
  Speaker 1: galaxies of the same size? Do we get galaxies in
• 32:39
  Speaker 1: the same kind of arrangements? Not specifically this galaxy and
• 32:42
  Speaker 1: that galaxy, but like, What's the distribution of galaxy sizes?
• 32:45
  Speaker 1: What's the typical distance between galaxies? This kind of stuff.
• 32:48
  Speaker 1: So it's more a statistical comparison than a specific one,
• 32:52
  Speaker 1: which is kind of unfortunate. It would be awesome to
• 32:54
  Speaker 1: simulate our exact universe and be like, look, that's where
• 32:57
  Speaker 1: Kelly appears.
• 32:59
  Speaker 3: Yeah, but you know, it would not make it more
• 33:02
  Speaker 3: useful to know where Kelly appears, even if it would
• 33:04
  Speaker 3: make it more awesome to know where Kelly appears.
• 33:07
  Speaker 1: But it does allow you to ask really cool questions like,
• 33:11
  Speaker 1: What would happen if you had a universe without dark matter?
• 33:13
  Speaker 1: Like take the universe as we know it, fill it
• 33:16
  Speaker 1: with the photons and the protons and the neutrons that
• 33:19
  Speaker 1: we think were there, but don't put in the dark matter.
• 33:22
  Speaker 1: What happens? Well, we can answer that question. You do
• 33:25
  Speaker 1: that simulation and you don't get galaxies. Like 14 billion
• 33:29
  Speaker 1: years is not enough time for gravity to pull protons
• 33:33
  Speaker 1: and neutrons together to make stars and galaxies without the
• 33:38
  Speaker 1: help of the gravity from dark matter. So that's already
• 33:42
  Speaker 1: a really interesting result. It says that the structure we see,
• 33:46
  Speaker 1: the fact that we have galaxies and stars, is itself
• 33:50
  Speaker 1: evidence for dark matter or evidence for something out there
• 33:54
  Speaker 1: helping pull stuff together because the visible matter doesn't have
• 33:57
  Speaker 1: enough gravity to do it on its own. Okay. And
• 34:01
  Speaker 1: so I think that's really cool because it shows you
• 34:03
  Speaker 1: the power of simulations. It says... If your initial conditions
• 34:07
  Speaker 1: and the rules to evolve it are inconsistent with what
• 34:10
  Speaker 1: we see, right, then something is wrong with one of
• 34:13
  Speaker 1: those two. And so if you put in no dark
• 34:15
  Speaker 1: matter and you run the rules, you get something which
• 34:17
  Speaker 1: looks very much not like our universe. So that's probably wrong.
• 34:21
  Speaker 1: But if you put dark matter in, then you do
• 34:24
  Speaker 1: see formation of galaxies. Dark matter itself clumps. Remember, there's
• 34:28
  Speaker 1: a lot more dark matter than anything else. And it
• 34:30
  Speaker 1: pulls those stars together and it pulls those galaxies together.
• 34:33
  Speaker 1: And so if you put in the dark matter we
• 34:35
  Speaker 1: think was there and it's measured independently in like nine
• 34:38
  Speaker 1: other ways, then you get the formation of structure just
• 34:41
  Speaker 1: like the structure that we see.
• 34:43
  Speaker 3: And I assume this is something we've simulated before the
• 34:46
  Speaker 3: Flamingo Project. Is that right?
• 34:48
  Speaker 1: Yes. Okay. That's right. And people have played with it.
• 34:51
  Speaker 1: They've been like, well, what if we put in more
• 34:52
  Speaker 1: dark matter? Let's double the amount of dark matter. What happens? Well,
• 34:55
  Speaker 1: the universe collapses. There's too much gravity and things get
• 34:58
  Speaker 1: squished and it doesn't look like our universe anymore. So
• 35:00
  Speaker 1: you can use this to measure the amount of dark
• 35:03
  Speaker 1: matter in our universe by tuning the simulation, the initial conditions,
• 35:08
  Speaker 1: until you get an outcome that matches what we're seeing today.
• 35:12
  Speaker 1: It's like if you didn't know how much salt Zach
• 35:16
  Speaker 1: put in dinner, but you ate the dinner and you
• 35:18
  Speaker 1: were like, okay, let me try it. I'm going to
• 35:20
  Speaker 1: try it without salt. No, it comes out bland. I'm
• 35:22
  Speaker 1: going to try it with a pound of salt. No,
• 35:24
  Speaker 1: that comes out too salty. And you refine it, comparing
• 35:27
  Speaker 1: your experiments to your memory of, dinner that Zach made
• 35:30
  Speaker 1: until you figure out how much salt Zach put in
• 35:33
  Speaker 1: his dinner. It's just like that with dark matter in
• 35:35
  Speaker 1: the universe.
• 35:36
  Speaker 3: Got it. Okay, let's take a break. And when we
• 35:39
  Speaker 3: get back, we will find out what the Flamingo Project
• 35:42
  Speaker 3: did in particular to create a newsworthy simulation.
• 36:05
  Speaker 1: All right.
• 36:05
  Speaker 3: And we're back. And Daniel, in closing, I said that
• 36:07
  Speaker 3: this was a newsworthy simulation, but maybe it only appears
• 36:11
  Speaker 3: newsworthy to me because it's got the word flamingo in it.
• 36:13
  Speaker 3: And so I feel like it should be in the
• 36:15
  Speaker 3: news because that's awesome. Did this simulation tend to make
• 36:18
  Speaker 3: the news?
• 36:19
  Speaker 1: Well, I mean, the news that I read, of course,
• 36:23
  Speaker 1: and the emails I get from listeners, I definitely heard
• 36:25
  Speaker 1: a lot about it. It was in Sky and Telescope
• 36:27
  Speaker 1: and a bunch of other PopSci places.
• 36:28
  Speaker 2: Okay, cool.
• 36:29
  Speaker 1: It was a step forward. I don't always understand why
• 36:31
  Speaker 1: a particular piece of science gets a bunch of attention
• 36:34
  Speaker 1: and another piece of science I think is really cool doesn't. But,
• 36:37
  Speaker 1: you know, that's just the way the PopSci cookie crumbles.
• 36:41
  Speaker 1: But this is legitimately a really cool piece of science. One,
• 36:44
  Speaker 1: because it's huge. Like, they took on a really big task.
• 36:48
  Speaker 1: They wanted to simulate a big chunk of the universe. So,
• 36:52
  Speaker 1: not the whole universe. But they described a cube of
• 36:56
  Speaker 1: universe with sides 10 billion light years on a side.
• 37:00
  Speaker 4: Wow.
• 37:00
  Speaker 1: So, like, that's a lot of stuff. And in that
• 37:03
  Speaker 1: are a lot of particles. They couldn't model all of them.
• 37:05
  Speaker 1: With their computing budget, what they could do is describe...
• 37:08
  Speaker 1: particles with masses, like 130 million times the mass of
• 37:13
  Speaker 1: our sun. So basically like they could zoom into like
• 37:16
  Speaker 1: mini galaxies. They couldn't describe individual stars or individual planets
• 37:20
  Speaker 1: or flamingos or podcasters. They can only have granularity down
• 37:24
  Speaker 1: to like mini galaxies. But in a cube of universe
• 37:28
  Speaker 1: that big, there's like 300 billion of those. So that's
• 37:33
  Speaker 1: what they described, like 300 billion clumps of matter each
• 37:37
  Speaker 1: about as massive as a mini galaxy.
• 37:39
  Speaker 3: And for reference, do you have a sense for how
• 37:43
  Speaker 3: much of an advance in specificity this is relative to
• 37:47
  Speaker 3: prior models?
• 37:48
  Speaker 1: This is a big chunk of compute, but the real
• 37:51
  Speaker 1: step forward is not that they zoomed in more or
• 37:54
  Speaker 1: they had a bigger slice of the universe. The specificity
• 37:56
  Speaker 1: is that they made the rules much more realistic. So
• 38:00
  Speaker 1: previous simulations were like, let's model the dark matter and
• 38:03
  Speaker 1: we'll also put the protons in But we won't let
• 38:06
  Speaker 1: the protons interact with each other because, ooh, that's really complicated.
• 38:09
  Speaker 1: Like dark matter is very simple. It doesn't really interact
• 38:11
  Speaker 1: with itself. It's just gravity, and it flows around, and
• 38:15
  Speaker 1: it clumps, and we can handle that. Protons are hard
• 38:18
  Speaker 1: because they have the strong interaction, and they have the
• 38:21
  Speaker 1: weak interaction, and they generate other kinds of particles, which
• 38:24
  Speaker 1: are really complicated. And we know because we eat them
• 38:27
  Speaker 1: every day, and that's why they're delicious, because of all
• 38:29
  Speaker 1: their complexity. And so most simulations... Have the atoms, the
• 38:33
  Speaker 1: baryons in them, but don't let them interact because that's
• 38:36
  Speaker 1: too complicated. And so what the Flamingo Project did was
• 38:40
  Speaker 1: they included the baryons and their physics. So they have
• 38:43
  Speaker 1: things like gas flows and hydrodynamics and how do winds
• 38:48
  Speaker 1: get generated. All these things are really important for the
• 38:51
  Speaker 1: structure of galaxies. Like when there's massive radiation from the
• 38:55
  Speaker 1: central black hole that blows out stars in the centers
• 38:58
  Speaker 1: of galaxies. And it changes how stars form in galaxies.
• 39:03
  Speaker 1: And you have like active galactic nuclei that can shape
• 39:06
  Speaker 1: the structure of nearby clusters. So they included all of
• 39:10
  Speaker 1: those details in the simulation. And they included neutrinos. So
• 39:15
  Speaker 1: previous simulations at this scale hadn't included this baryonic physics, like,
• 39:19
  Speaker 1: you know, all the interactions between protons or the neutrinos. Okay, awesome.
• 39:23
  Speaker 3: That seems like quite a big step forward. Did they
• 39:26
  Speaker 3: have access to a hyper super duper computer that we
• 39:29
  Speaker 3: didn't have? have access to before?
• 39:32
  Speaker 1: They got time on a very fragilistic computer at Durham University.
• 39:36
  Speaker 1: They ran it on 30,000 CPUs using 50 million compute hours.
• 39:42
  Speaker 1: So this is a computer that existed. It wasn't built
• 39:43
  Speaker 1: specifically for them, but they got a bunch of time
• 39:46
  Speaker 1: on it. So, you know, people always want time in
• 39:49
  Speaker 1: these computers. It's very valuable. And you're competing with folks
• 39:51
  Speaker 1: who like want to simulate hurricanes or want to simulate
• 39:55
  Speaker 1: nuclear weapons or want to simulate other stuff. So it
• 39:58
  Speaker 1: was awesome that they got all this compute time to
• 40:01
  Speaker 1: answer this question because there are really interesting open puzzles.
• 40:05
  Speaker 1: Like we talked about modeling how dark matter changes the
• 40:08
  Speaker 1: structure of the universe and we described it as if
• 40:10
  Speaker 1: it sort of works and it does mostly work and
• 40:13
  Speaker 1: it's cool. But the more we study it, the more
• 40:15
  Speaker 1: we see discrepancies. Just like when we were talking about
• 40:19
  Speaker 1: throwing a ball and you use your first model and
• 40:21
  Speaker 1: it mostly works, but then you notice, hmm, it's not
• 40:23
  Speaker 1: exactly right. What are we missing? As we get more
• 40:26
  Speaker 1: and more specific, we can look more and more carefully
• 40:28
  Speaker 1: at the discrepancies and wonder, what are we missing? And
• 40:31
  Speaker 1: you've probably heard of the Hubble tension, you know, this
• 40:34
  Speaker 1: discrepancy between how rapidly the universe was expanding at early times,
• 40:39
  Speaker 1: like 14 billion years ago, as measured in the Cosmic
• 40:41
  Speaker 1: Microwave Background Radiation, and how rapidly it seems to be
• 40:44
  Speaker 1: expanding today, as measured by supernova and other things. There
• 40:48
  Speaker 1: are other tensions in cosmology, places where our models of
• 40:52
  Speaker 1: the universe don't quite agree with our observations, and one
• 40:55
  Speaker 1: of them is called the S8 tension. It's not as
• 40:57
  Speaker 1: sexy or doesn't get as much press as the Hubble tension,
• 41:00
  Speaker 1: but it's really important. This S8 is a measurement of
• 41:04
  Speaker 1: how clumpy is the universe. Has stuff pulled together into
• 41:08
  Speaker 1: big clumps, or is it still mostly spread out? And
• 41:11
  Speaker 1: just like with the Hubble tension, we have two measurements
• 41:13
  Speaker 1: from the CMB and from late times, and those disagree.
• 41:17
  Speaker 1: So the CMB says the universe was clumpier, and the
• 41:20
  Speaker 1: late time says, no, the universe was less clumpy. And
• 41:23
  Speaker 1: we want to understand, like, which of these measurements is wrong?
• 41:27
  Speaker 1: Can you build a model of the universe that's compatible
• 41:29
  Speaker 1: with both of them? So far, we haven't been able to.
• 41:32
  Speaker 1: And so this simulation thought, well, let's study this question.
• 41:36
  Speaker 1: Let's include as all the physics that we can, because
• 41:38
  Speaker 1: we know that protons and neutrinos affect clumpiness and they
• 41:42
  Speaker 1: ran 28 slightly different simulations tweaking cosmic parameters like the
• 41:47
  Speaker 1: dark matter fraction or how massive is the neutrino or
• 41:50
  Speaker 1: how important are these active galactic nuclei this hydrodynamics of
• 41:55
  Speaker 1: the baryons we were talking about how big are stars
• 41:58
  Speaker 1: when they're formed typically um you know which percentage of
• 42:02
  Speaker 1: stars fall in which mass ranges they ran 28 different
• 42:04
  Speaker 1: versions to see if they could understand this problem.
• 42:07
  Speaker 3: I bet they agonized over the 28 different simulations. I
• 42:12
  Speaker 3: remember when I used to do simulations, the hardest part
• 42:14
  Speaker 3: was figuring out what part of the parameter space am
• 42:17
  Speaker 3: I going to explore? And yeah, trying to figure out
• 42:20
  Speaker 3: the values of the variables for those different simulations was
• 42:23
  Speaker 3: probably painstakingly done.
• 42:25
  Speaker 1: Oh, you know, they had meetings, they had disagreements. More meetings.
• 42:29
  Speaker 3: More meetings.
• 42:30
  Speaker 1: You know, people suggested stuff, other people asked their suggestions,
• 42:33
  Speaker 1: you know, there was grumbling and teeth gnashing. And, you know,
• 42:36
  Speaker 1: these are important decisions because later when you get the results,
• 42:39
  Speaker 1: you're like, oh, that's interesting. I wish we had another
• 42:41
  Speaker 1: point with this other tweaked value. But, you know, you
• 42:44
  Speaker 1: used up all your time. Sorry. Or one answer comes
• 42:48
  Speaker 1: out really boring. You're like, well, that was a big
• 42:50
  Speaker 1: waste of a zillion hours of computing. Thanks, Joe. Great idea, Joe. So, yeah, exactly.
• 42:58
  Speaker 1: It's definitely very painful. when that happens. But they ran
• 43:02
  Speaker 1: these simulations, and what they found is that the baryonic effects,
• 43:07
  Speaker 1: including all these protons, does change the clumpiness, but not
• 43:12
  Speaker 1: enough to explain this S8 tension. And so the question
• 43:16
  Speaker 1: they went in with was, does adding the protons, making
• 43:19
  Speaker 1: the simulation more realistic, does this solve the problem? Is
• 43:22
  Speaker 1: the problem just an artifact of our simulation being too
• 43:26
  Speaker 1: simplistic and not including all the physics we love and know?
• 43:28
  Speaker 1: And the answer to the question is no, it's not
• 43:31
  Speaker 1: an artifact of the simulation being too simplistic in this
• 43:34
  Speaker 1: one way. So that would have been very cool if
• 43:38
  Speaker 1: they had solved that problem. Like, oh, look, the universe
• 43:40
  Speaker 1: now makes sense. But now it's very cool because it means, oh,
• 43:43
  Speaker 1: the universe doesn't fully make sense, which means there's something
• 43:46
  Speaker 1: to learn, right? Either maybe there's some kind of dark
• 43:49
  Speaker 1: matter that has a self-interaction that's not being included in
• 43:52
  Speaker 1: these models, right? We describe usually dark matter just as
• 43:55
  Speaker 1: having gravity and no other kind of interaction, but Maybe
• 43:58
  Speaker 1: dark matter is complicated, and there's lots of different kinds
• 44:01
  Speaker 1: of it, and some of it has interesting interactions with
• 44:04
  Speaker 1: other kinds, and that's not being included in the model,
• 44:07
  Speaker 1: and that's why the models don't agree with reality. That's
• 44:10
  Speaker 1: just like one hypothesis. So there's a lot still to
• 44:13
  Speaker 1: learn here.
• 44:14
  Speaker 3: It's such a roller coaster doing science. Cause like, you know,
• 44:16
  Speaker 3: obviously you set out to do an experiment. You, you
• 44:19
  Speaker 3: kind of think, you know what the result's going to be,
• 44:21
  Speaker 3: or at least that's what you say in your grant.
• 44:23
  Speaker 3: And if you get the result, you're like, what?
• 44:25
  Speaker 4: Yay.
• 44:26
  Speaker 3: But then if you don't get the result, you're like,
• 44:28
  Speaker 3: but this could be equally as interesting or even maybe
• 44:32
  Speaker 3: even more interesting, especially if it sets you on like,
• 44:35
  Speaker 3: gives you a little bit of a hint for what
• 44:36
  Speaker 3: you were maybe missing that you should think more about.
• 44:38
  Speaker 3: But like, yeah, this, so this sounds like something where
• 44:41
  Speaker 3: the result would have been fascinating no matter what.
• 44:44
  Speaker 1: Yeah, exactly. And I think it made news because it's
• 44:47
  Speaker 1: a big, awesome piece of computing. Like, wow, what a
• 44:50
  Speaker 1: cool project. But unfortunately, it didn't get the answer they
• 44:53
  Speaker 1: were hoping for. But, you know, we learned something. We
• 44:56
  Speaker 1: crossed something off of our list of hypotheses, and we're
• 44:58
  Speaker 1: still on the hunt to understand the universe and figure
• 45:01
  Speaker 1: out why it is exactly as clumpy as it is today.
• 45:05
  Speaker 1: I think it's wonderful. I don't think it's too clumpy.
• 45:08
  Speaker 1: I don't think it's overly clumped. I don't think it's
• 45:11
  Speaker 1: under clumped. I think it's just right. What do you think, Kelly?
• 45:14
  Speaker 3: I think you are a very accepting universe partner. And,
• 45:18
  Speaker 3: you know, I haven't thought too much about the universe's clumpiness.
• 45:21
  Speaker 3: I like it for its personality. And so I'm totally
• 45:24
  Speaker 3: okay with the universe as its clumpy self, you know?
• 45:27
  Speaker 1: You're making me sound superficial here. I'm only into the
• 45:30
  Speaker 1: universe because of its clumpiness.
• 45:31
  Speaker 3: I mean, you were the one who was focusing on
• 45:33
  Speaker 3: the clumpiness. I'm just saying.
• 45:36
  Speaker 1: I didn't say it's the only thing I like about
• 45:38
  Speaker 1: the universe, Kelly.
• 45:39
  Speaker 3: The universe is multifaceted.
• 45:41
  Speaker 1: Yes, it is. So I think that answers Joe's question,
• 45:45
  Speaker 1: and we'll send this to him and hear if he
• 45:47
  Speaker 1: has any follow-ups. But, you know, the episode, I suppose,
• 45:50
  Speaker 1: sort of asks a bigger question. Like, all right, this
• 45:53
  Speaker 1: simulates a bunch of stuff in the universe, but it
• 45:56
  Speaker 1: glosses over a lot of details. If the smallest thing
• 45:59
  Speaker 1: in your simulation... is a particle with 130 million solar masses,
• 46:04
  Speaker 1: then you're not describing little Timmy's bicycle. Is it red
• 46:08
  Speaker 1: or is it blue? And I think it's an interesting
• 46:10
  Speaker 1: question that the listeners were responding to and that you
• 46:12
  Speaker 1: were asking about earlier, like, is it possible to describe
• 46:16
  Speaker 1: the whole universe? And I think that the answer to
• 46:18
  Speaker 1: that question has to be no, because a simulation containing
• 46:23
  Speaker 1: the complete quantum state of the universe is would require
• 46:26
  Speaker 1: at least as much information as the universe itself contains, right?
• 46:30
  Speaker 1: And so like a smaller computer cannot contain a lossless
• 46:35
  Speaker 1: description of a larger physical system. And if that computer
• 46:37
  Speaker 1: is in the universe, then it has to be smaller
• 46:41
  Speaker 1: than the universe. You can't use the whole universe to
• 46:44
  Speaker 1: simulate the universe. And even if you did, right, there
• 46:46
  Speaker 1: are parts of your computer which are not being described
• 46:49
  Speaker 1: in the simulation. And so I think there's this cool
• 46:52
  Speaker 1: recursive issue Your simulation would have to include a description
• 46:56
  Speaker 1: of the simulation. And so I don't think it's possible
• 46:59
  Speaker 1: to describe the whole universe within the universe. The simulation
• 47:04
  Speaker 1: hypothesis is another question. It's like, is our universe a simulation?
• 47:09
  Speaker 1: That implies that our universe is running on a computer
• 47:12
  Speaker 1: in some larger universe. The same way that we describe
• 47:17
  Speaker 1: Super Mario Brothers in a simulation. We create a universe
• 47:20
  Speaker 1: and it has its own rules. And that exists within
• 47:23
  Speaker 1: our universe. And in our universe is a little piece
• 47:26
  Speaker 1: of the universe we call a Nintendo system, which is
• 47:28
  Speaker 1: complicated enough to describe the Super Mario universe. And so
• 47:33
  Speaker 1: if our universe is a simulation, then that implies that
• 47:37
  Speaker 1: there's a computer out there capable of describing all of
• 47:40
  Speaker 1: these details that must be within another universe that's at
• 47:44
  Speaker 1: least as big.
• 47:45
  Speaker 3: Yeah, so I was going to ask why would anybody
• 47:47
  Speaker 3: want to simulate our universe? But now that you've used
• 47:49
  Speaker 3: Super Mario Brothers as an example, our universe is clearly
• 47:53
  Speaker 3: better with it in it. And so now maybe we
• 47:56
  Speaker 3: are the fun part of some other creature's universe. I dig.
• 48:00
  Speaker 1: Yeah, but I want to point out some fun loopholes here. Like,
• 48:03
  Speaker 1: you know, people talk about looking at the way our
• 48:07
  Speaker 1: universe operates and looking for evidence that it is a simulation.
• 48:11
  Speaker 1: I think that's really dangerous and misleading information. Because what
• 48:15
  Speaker 1: you're looking for is evidence that the universe operates the
• 48:18
  Speaker 1: way our simulations operate. But if our universe is a
• 48:23
  Speaker 1: simulation in some other universe, we have no idea what
• 48:26
  Speaker 1: the physics is of that universe. The same way that
• 48:28
  Speaker 1: Mario has no idea what the rules of physics as
• 48:32
  Speaker 1: Newton and Einstein have developed are. His universe has different physics.
• 48:36
  Speaker 1: And so if he's looking for hints in the Mario
• 48:39
  Speaker 1: universe as to how our universe operates, he's not going
• 48:42
  Speaker 1: to find anything. And so if we're looking in our
• 48:45
  Speaker 1: universe for evidence that operates like our computers, we're on
• 48:49
  Speaker 1: the wrong track. We need to look in our universe
• 48:52
  Speaker 1: for evidence that operates like the meta-universes' computers, about which
• 48:56
  Speaker 1: we have no information because they follow other laws of
• 48:59
  Speaker 1: physics we don't know anything about. And so I think
• 49:02
  Speaker 1: that's a little bit hopeless.
• 49:04
  Speaker 3: Okay, so just to be clear, Daniel is being the
• 49:06
  Speaker 3: wet blanket today and is essentially saying there's nothing in
• 49:09
  Speaker 3: this universe that can answer the question– are we living
• 49:12
  Speaker 3: in a sim or not?
• 49:14
  Speaker 1: I think that's true, yeah.
• 49:15
  Speaker 2: Oh, man.
• 49:15
  Speaker 1: But there are some other fun loopholes like, you know,
• 49:17
  Speaker 1: can you simulate a whole universe? I don't think you can,
• 49:20
  Speaker 1: but you can simulate large parts of the universe, right? Like,
• 49:24
  Speaker 1: you could, a big chunk of the universe, really high fidelity,
• 49:27
  Speaker 1: and that can answer a lot of physics questions. Often,
• 49:29
  Speaker 1: you don't need to simulate the whole universe. Like to
• 49:33
  Speaker 1: answer this question about dark matter, you don't need to
• 49:35
  Speaker 1: add Timmy's bicycle and whether it's red or blue, it
• 49:37
  Speaker 1: doesn't change the answer. And so the fact that the
• 49:40
  Speaker 1: universe breaks into these levels of emergence where you don't
• 49:44
  Speaker 1: need to know all the lower level details to make
• 49:47
  Speaker 1: chicken soup, for example, like you can make chicken soup
• 49:49
  Speaker 1: without knowing quantum gravity or string theory. That's great. Chicken
• 49:53
  Speaker 1: soup is delicious, but it also means we can do
• 49:55
  Speaker 1: science at various scales without knowing all the details. That's
• 49:58
  Speaker 1: a blessing. So it's good actually to, that our simulations
• 50:02
  Speaker 1: don't have to include everything in the universe. And then
• 50:05
  Speaker 1: the other thing is that sometimes you can get away
• 50:08
  Speaker 1: with not describing the universe in all of its gory
• 50:11
  Speaker 1: detail because there are patterns. So we have laws of physics,
• 50:14
  Speaker 1: we have symmetries that tell us, constrain what can happen.
• 50:18
  Speaker 1: So for example, say you're simulating a particle and it
• 50:21
  Speaker 1: decays into two other particles. Well, conservation of momentum tells
• 50:25
  Speaker 1: you that the first particle was at rest, The next
• 50:28
  Speaker 1: two particles have to be back to back. They have
• 50:30
  Speaker 1: to have momentum that's exactly opposite each other so that
• 50:33
  Speaker 1: the total momentum is still zero. What does that mean?
• 50:35
  Speaker 1: How does it help you? It means you only have
• 50:37
  Speaker 1: to store in your simulation one of the momenta. You
• 50:41
  Speaker 1: don't have to store both because the other one you
• 50:43
  Speaker 1: can derive instantly from your rule. So symmetries and rules
• 50:48
  Speaker 1: mean you don't have to store the full state of
• 50:51
  Speaker 1: the simulation in your computer. You only have to store
• 50:54
  Speaker 1: enough information to derive it when you need it. And
• 50:57
  Speaker 1: so these symmetries, these constraints essentially are like a way
• 51:00
  Speaker 1: to compress your knowledge of the universe more compactly and losslessly.
• 51:06
  Speaker 1: So there are some loopholes here, too. You have to
• 51:09
  Speaker 1: have a computer the size of the universe in order
• 51:11
  Speaker 1: to describe the universe.
• 51:13
  Speaker 4: Yes.
• 51:13
  Speaker 3: Like what if we are one run out of 10,000
• 51:15
  Speaker 3: of a simulation that's testing five rules and we are
• 51:20
  Speaker 3: the result of testing some of those rules? Would that work?
• 51:23
  Speaker 4: Yeah.
• 51:25
  Speaker 1: Yeah, that could work. And I wonder if our universe
• 51:29
  Speaker 1: is one that somebody fought for in inclusion. That's right.
• 51:33
  Speaker 1: Thank you, Cosmic Joe, for including our universe.
• 51:36
  Speaker 3: Our Joe totally messed it up. I hope we've pleased you. Yeah,
• 51:40
  Speaker 3: but Cosmic Joe. Thank you, Cosmic Joe.
• 51:43
  Speaker 1: All right, so we'll send this episode to Joe from
• 51:46
  Speaker 1: Ann Arbor to hear if we've answered his question about
• 51:49
  Speaker 1: the Flamingo Simulation and about simulating the universe in general.
• 51:54
  Speaker 1: Here's Joe's answer.
• 51:55
  Speaker 3: I totally forgot that the listener's name was Joe when
• 51:57
  Speaker 3: I picked Joe as the... Maybe that was in the
• 52:00
  Speaker 3: back of my brain subconsciously the whole episode.
• 52:04
  Speaker 1: Sorry, Joe.
• 52:04
  Speaker 3: We didn't mean to blame you for the bad run
• 52:06
  Speaker 3: of the simulation.
• 52:06
  Speaker 1: It just goes to Joe, you.
• 52:09
  Speaker 3: Thank you so much.
• 52:10
  Speaker 4: I really loved that episode. Daniel, I know that you
• 52:13
  Speaker 4: would especially like to contact the being running the simulation,
• 52:17
  Speaker 4: but I can neither confirm nor deny whether I am
• 52:19
  Speaker 4: Cosmic Joe. I can confirm, however... that the universe is
• 52:24
  Speaker 4: overly clumpy in the vicinity of my desk. But seriously,
• 52:28
  Speaker 4: thank you both for the thorough explanation of simulations and
• 52:31
  Speaker 4: the Flamingo Project. Emergent properties at different scales seem to
• 52:35
  Speaker 4: be a major theme on the podcast lately, and I
• 52:38
  Speaker 4: am fascinated of the idea of simulations being a way
• 52:41
  Speaker 4: to probe from one layer of abstraction to the next.
• 52:46
  Speaker 4: I don't have a follow-up question, but I do want
• 52:48
  Speaker 4: to urge everyone to ask their own questions. Listening to
• 52:52
  Speaker 4: the raw audio was a treat, especially right after listening
• 52:54
  Speaker 4: to the episode with Matt Kesselman. Danielle and Kelly are
• 52:58
  Speaker 4: somehow even more charming in real time. My only complaint
• 53:01
  Speaker 4: is that I didn't hear a single goat. Thanks for
• 53:04
  Speaker 4: answering my question. I love getting smarter with you.
• 53:07
  Speaker 1: Thanks very much, everybody, for taking this trip down the
• 53:10
  Speaker 1: simulated universe to understand how we think about the universe,
• 53:13
  Speaker 1: how we understand it, and how simulation is an absolutely
• 53:16
  Speaker 1: vital part of modern science.
• 53:18
  Speaker 3: May your simulation be a good one.
• 53:27
  Speaker 1: Thanks, everybody, for listening. Please go and do us a
• 53:29
  Speaker 1: favor and rate the show on whatever podcast app you're using.
• 53:32
  Speaker 1: It really helps people find us.
• 53:35
  Speaker 3: Daniel and Kelly's Extraordinary Universe is edited by the amazing
• 53:39
  Speaker 3: Matt Kesselman.
• 53:39
  Speaker 1: He really is a wizard. You can also find us
• 53:43
  Speaker 1: online on Blue Sky, Instagram, and XDNK Universe. Come engage
• 53:49
  Speaker 1: with us.
• 53:49
  Speaker 3: You can email us at questions at danielandkelly.org. We really
• 53:54
  Speaker 3: do want to hear from you.
• 53:55
  Speaker 1: And you can find our website, www.danielandkelly.org, where you'll also
• 54:00
  Speaker 1: find an invitation to join our Discord, where everybody comes
• 54:04
  Speaker 1: and talks about the amazing universe.
• 54:06
  Speaker 3: And we also have the most amazing moderators. This is
• 54:10
  Speaker 3: an iHeart Podcast. Thanks for joining us.