Is the Universe a Simulation? Andrew Pontzen [Ep. 466]

Please join my mailing list here 👉 https://briankeating.com/list to win a meteorite 💥 Simulations today are powerful tools for exploring the mysteries of our universe, but how close can they come to replicating reality? Can we recreate everything through simulations, or are there limits we can’t overcome? And how do today’s powerful simulations shape our understanding of everything from galaxies to climate change? Here today, to answer all of these questions and more, is none other than Andrew Pontzen. Andrew is a cosmologist and professor known for his expertise in using computer simulations to understand the universe. He’s the author of The Universe in a Box, where he unpacks the complexities of cosmic simulations and their surprising limitations. Pontzen’s work is at the cutting edge of how we model and predict phenomena ranging from galaxy formation to climate change, bridging theoretical physics with practical, big-picture questions about reality. As we head toward an era where simulations and AI are central to scientific discovery, these questions are more important than ever. Tune in to learn about the power and limitations of simulation! Key Takeaways: 00:00 Intro 00:44 Is it possible to simulate reality? 04:49 Early simulation experiments 10:30 Fluid dynamics and weather simulations 13:51 Judging a book by its cover 16:01 End of the universe 19:13 Cosmic microwave background and spinning universes 24:11 Verification and validation 27:34 Big Bang controversy and the JWST 31:35 Galaxy simulations 42:23 Multiverse and quantum computing 48:58 Ethics and energy impact of high-performance computing 55:34 Outro Additional resources: ➡️ Learn more about Andrew: 📚 Get The Universe in a Box: https://a.co/d/2V7vuFR ✖️ Follow Andrew on Twitter: https://x.com/apontzen ➡️ Follow me on your fav platforms: ✖️ Twitter: https://twitter.com/DrBrianKeating 🔔 YouTube: https://www.youtube.com/DrBrianKeating?sub_confirmation=1 📝 Join my mailing list: https://briankeating.com/list ✍️ Check out my blog: https://briankeating.com/cosmic-musings/ 🎙️ Follow my podcast: https://briankeating.com/podcast Into the Impossible with Brian Keating is a podcast dedicated to all those who want to explore the universe within and beyond the known. Make sure to follow/subscribe so you never miss an episode! Learn more about your ad choices. Visit megaphone.fm/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Shop Black Friday appliance savings and get up to an extra $1,250 off select appliance purchases like GE Profile. Plus, free delivery on appliance purchases of $398 or more. Make life easier with the fridge that doubles as your personal kitchen assistant. Or upgrade your laundry room with the Ultra Fresh Vent System Plus Front Load Washer. So your clothes stay fresh after every single wash. Shop Black Friday appliance savings with fast free delivery at The Home Depot. Offer valid November 4th through December 2nd at participating stores and at homedepot.com. U.S. only. See store online for details.

• 0:31Speaker 01

  Are you ever shopping online and suddenly a discount that helps you save pops up? It makes your day, sometimes even your week. Want that feeling month after month? Well, new customers who save by switching their home and auto insurance to Progressive save hundreds. Cue your happy dance. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers not available in all states.

• 0:57Speaker 02

  One has to be incredibly clear about what we're trying to achieve when we perform a simulation. So we know we never kind of fully correctly solve the equations that we're trying to solve. Even if we could solve the equations that we write down perfectly, we are not literally reproducing the universe. I think we always have to start from that point, that there is no perfect solution to this problem.

• 1:22Brian Keating?

  Andrew Pontzen, welcome to the Into the Impossible podcast. Great to have you.

• 1:26Speaker 02

  Oh, thanks for having me.

• 1:27Brian Keating?

  In your book, you conclude with a simultaneous discussion of two extremely hot-button issues: aliens in the universe and a hypothetical instantiation of the simulation hypothesis, popularized by many-time past guest Nick Bostrom. Is it possible, knowing as you do so much about simulations of cosmic and terrestrial phenomena, is it even possible, computationally, energetically, for any civilization to truly simulate reality, such that you and I perceive that we're having this conversation, but it's truly just a giant version of Minecraft?

• 2:06Speaker 02

  Short answer, no. I don't think it is. But of course, I mean- The long answer can can take a very long time indeed, because there's a there's a huge amount to kind of unpack in in what seems to be a kind of straightforward question. There's actually a huge amount going on there under the surface. And in a sense, that was one of the things I wanted to to write about in the book. The short answer is that the richness of our universe, and we can actually measure that in something that maybe we'll come on to, I don't know, called qubits. That that's bits in a quantum computer. The number of qubits accessible within our universe is a fixed number. And we can kind of make estimates of that number. It's some humongous number. The exact number doesn't matter too much, but it's absolutely humongous.

• 2:49Speaker 02

  And the thing is, if you wanted to simulate that with perfect fidelity, you would need a quantum computer with that many qubits. So in other words, you would need to use the resources of the entire universe to simulate the universe. And and so, you know, at at the heart of it, that is why I just do not believe that this is really a a kind of plausible explanation for kind of future path of simulations.

• 3:15Brian Keating?

  We've had our countryman and Nobel laureate Tim Palmer discussing his book on chaos and and computer simulations, which has a very famous anecdote about abysmal weather forecasting in your fine country. We'll get into that. We talk a lot about climate change skepticism on the podcast with him from two years ago now, the interview that we did. And, you know, part of the impression that he came made to me was that to simulate the climate, we, according to him, need a CERN level, you know, climate simulator, et cetera. But ideally, we'd have an earth-sized simulator. Doesn't that response that you just made, you know, about qubits, I mean, we could translate that to classical computing

• 3:58Brian Keating?

  and just ordinary bits. Don't the sort of proponents of skepticism or the skeptics about anthropogenic climate change, don't they have a point then to say, well, if you can't, you know, if you need a full-sized computer to simulate galaxies in the entire universe, why not, you know, say that there are limitations to the very much smaller than earth-sized simulations that we use for anthropogenic climate change concerns? What do you say to those critics?

• 4:24Speaker 02

  Well, the the thing to bear in mind is that when it comes to climate simulations, We are not trying, and I say we in a broad sense, I mean, I have nothing to do with climate simulations myself, but that we as a community are not trying to reproduce the earth in every detail. That's the first thing. It's just, that's just not part of the goal. What one is trying to extract from climate simulations is some impression of the patterns that our planet may go through in the future. And the kind of thing about simulations of the universe as a whole, but but also of our planet, is it does prove to be possible to draw out certain patterns, despite the fact that not every single detail

• 5:06Speaker 02

  is is represented in there. So I think, you know, again, this is one of the kind of driving motivations for me writing about this, that this is actually at the heart of what we do in simulations is we try to find ways to simplify, it has to be simplified, to fit inside a computer, and yet still get insights and and valid insights into the kind of patterns that are going on in our universe or on our planet.

• 5:32Brian Keating?

  You've pioneered many of these, I guess you can call them experiments. I've had some trouble, you know, I'm an experimental cosmologist, you can see the WMAP globe in the background, and actually, the W in WMAP was my grand advisor, David Wilkinson. So I've been doing this a long time, and I always found it distasteful. You know, we have colleagues around here that talk about they have a laboratory for brown dwarfs, you know, my colleague Andrew Berger, Adam Berger, yeah, he's not doing lab work, you know, per se. And Mike Norman, who is my colleague here at UC San Diego, does many, many great simulations, and the Flatiron, and all these groups do similar, and some of them call themselves experimentalists, or they're doing experiments. And I always kind of laughed at them, you know, silly

• 6:16Brian Keating?

  scientist, you know, when you learn how to use a wirebonder and, you know, a vacuum, a turbomolecular pump, then you can call me, I'm so arrogant, right? But in the book, you talk about a very early simulation of the galaxy, using light bulbs, photomultiplier tubes, and so forth. This was a real experiment, whoever was doing it, can you talk about that incredible? All experiment with light bulbs and how this was a true honest to goodness experiment, much to my chagrin.

• 6:45Speaker 02

  The interesting thing about simulations is in some sense, it doesn't matter whether they're running on a computer or not. The fact that we have bits and and and that we have computer code and so on is is kind of almost beside the point. You know what we're really trying to do is build systems that allow us to get some insight into other systems. And the other system happens to be this huge thing that we call the universe, or perhaps some kind of portion of it. But in the end, what we use to to make the simulation of it doesn't have to be a computer. And you're absolutely right that some of the earliest attempts to simulate the universe did not make use of computers at all. So that's true, for example,

• 7:27Speaker 02

  for simulating the the weather and the climate. There were various early attempts to do that just with pen and paper. But also, when it came to galaxies, this incredible experiment that was done, I think in in Lund University, if I remember correctly, where it involved a kind of whole laboratory full of light bulbs and actually whole whole groups of stars, because you can't have enough light bulbs to represent all the stars in a galaxy. And if you measure the intensity of light coming from a bulb, as you move away from that bulb, the intensity of that light drops off in a very particular way. It's called the inverse square law. And that is essentially how gravity works as well, that the force

• 8:09Speaker 02

  of gravity as you move away from a massive object also drops off as an inverse square law. So within a laboratory, it was possible to measure the intensity of light coming from different directions and use that as a as a proxy for what the force of gravity would be if you kind of scaled this up to to to galaxy sizes. And then, you know, move the ball, literally physically move the bulbs around as though they were stars being pulled around inside a galaxy by the force of gravity. And I just think it's a really neat example of, you know, what we're really trying to do is make one physical thing stand in for

• 8:48Brian Keating?

  The scientist there, Holmberg, that you describe in the book. Talk about the ways that the techniques that he invented, including the drift step and the kick step, that you're still using, you and your and your ilk are still using these, Andrew. I found it fascinating. Can you explain what are these things and how do we implement them in today's modern simulations?

• 9:07Speaker 02

  I've talked about the force of gravity. So the the first thing you have to do if you want to know how a galaxy or a collection of galaxies or anything much out in space is going to move, is figure out what that force of gravity is. And you can use that force to work out how things are accelerating. In other words, what the the change in their motion is going to be like. But that's just like that's just like a snapshot. It's just a moment in time. If I go round in this lab that, as you say, Holmberg had, go round in the lab and measure the light intensity in different places, I can figure out the forces. This gives me a snapshot of how the motion should be changing.

• 9:48Speaker 02

  And that change in motion that I can get from doing that is, we we call it a kick, because we imagine, okay, you can think of these forces as just every now and again, they they kick anything that's there, like a star, onto a slightly different trajectory than the one it was on previously. And then we do what's called a drift step. Now, in the laboratory, this would consist of moving the light bulbs in a straight line along their current direction of travel by a certain amount, which is proportional to their speed. In the real universe, this all happens at the same time. You know, the forces are acting and they're kind of reshaping the trajectories continually of the stars out there in the real universe.

• 10:30Speaker 02

  But in a lab, you can't really do that. That becomes too complex. So what you end up doing is you you move it along a straight line and and that's that's the drift. Then you recalculate all of these forces or you remeasure them and you put things onto new trajectories and then you and and that's the kick that that you've done. And then you drift again. And in this way, you sort of take baby steps and and watch how the thing. Changes over time through these kind of kick drift, kick drift cycles. So it's a kind of way of splitting up the different things that are going on in the real universe into into steps that we can actually concretely take either in a lab or on a computer.

• 11:13Brian Keating?

  Yeah, I found it truly amazing that these these tools are still employed in in many different ways throughout the realm, you know, from cosmological simulations to in these early simulations until to much later ones where we still can't replicate, you know, the full scale objects as we just talked about. It requires, you know, the entire universe and the entire climate is the best. And these are, you know, really belying the fact that we're talking about here are complex systems, right, that can't be replaced except faithfully replaced, except with their exact one to one correspondence. And yes, we we have to make we have to make approximations, of course. You know, the other thing that I was very

• 11:53Brian Keating?

  much interested in is is the sort of weather simulations where we, you know, I often show my students, you know, a hurricane and then a spiral galaxy, or I show the Orion Nebula and M51 I have somewhere in the background over here. You know, these have these superficial resemblances, but do they have actual, in other words, if our climate friends in the next building over and Tim Palmer, if they made some stupendous breakthrough, if they had a CERN for climate, could we immediately port that over one to one and then apply it in the simulations that you did?

• 12:26Speaker 02

  Maybe not immediately, but there's definitely a lot in common. And and I mean, the most fundamental thing they have in common is that they both deal with what we call in the physics world, we call them fluids, which would suggest it's stuff like water or oil, but, you know, it's it's more general than that. It's actually anything that can kind of be moved around and and it's not a solid, but it can be moved around. So we would include the air around us as a fluid, for instance, because that can be moved around if you've got a fan or whatever, and material that the atmosphere is made out of here on Earth. So the the air that it's made out of, but also the moisture in the atmosphere, that can be thought of as a fluid, and all the gas in the universe can be thought of

• 13:10Speaker 02

  as a fluid. And that has a technical meaning to it. It means that we can follow certain equations that tell us how we think it should behave. And of course, it's not kind of the most fundamental view. So it goes back to our earlier conversation that you are not directly talking about what we know it's actually made out of, which of course is molecules and atoms and the subatomic particles that make those up. But it's an extremely good effective view of what's going on. And it just considers material to be something that can be reshaped and it can be compressed, or it can expand, it can be pushed around, it can push on other bits of the fluid and so on. So there are these

• 13:52Speaker 02

  rules that are encoded into something called the Navier-Stokes equations, named after some 19th century physicists. And those rules or equations just encapsulate beautifully how material of this sort behaves on mass. So to a large extent, when we look at a weather forecast or a climate change prediction, it is based around these Navier-Stokes equations. And they also apply in the universe at large. So being able to solve them better in one domain does help in another domain, but the particular circumstances we're trying to solve them under also varies quite a bit.

• 14:34Brian Keating?

  We are now due for a slight commercial break to promote your wonderful new book, The Universe in a Box. And I'd like you to do what you're never supposed to do, which you're forbidden to do according to some, which is to judge a book by its cover. So you've got the British version, UK version there and paperback, and that is prompting me to do the inner, I listen to it, you narrate beautifully. If you ever need a fallback career, your mellifluous voice tones will be useful to all authors that don't want to record their own audiobooks like me. The Universe in a Box, explain the title, subtitle, and this artwork on the cover.

• 15:15Speaker 02

  All right. Yeah. So The Universe in a Box is the idea that you can take the universe and fit it inside a box, which is a computer. Now, I think our conversation already has made clear that it's a bit, you know, it's a bit tongue in cheek, right? You can't really do that. And, and, and the book, of course, unpicks why you can't do it and, and how nonetheless, you can get insights into the universe from attempting to fit some of its phenomena into the box that we call a, a computer. That's the idea behind the sort of main title. Interestingly, the subtitle is different in, in different editions around the world. So here in the UK, it's called, uh, The Universe in a Box: A New Cosmic History.

• 15:59Speaker 02

  The idea being that I think in the UK market, we were particularly highlighting that it talks about both about cosmological history, but also about some of the historical figures who've contributed to our ability to even do this, or even dream of doing this. And, and, and they're perhaps a little less well-known than some of the kind of big figures of theoretical cosmology that, that we more often talk about. So that, that's the UK subtitle. The US subtitle, Simulations and the Quest to Code the Cosmos, is, I guess, a little bit more literal and focuses in more on the actual coding side, which it's in there. But, but I, I think a lot of the book really talks about the things that lead up to the code rather than the code itself, if you like.

• 16:44Brian Keating?

  There's a, no shortage of books. There's another book by upcoming guest, Romeo Dave at DAVE. Uh, I guess I'll ask him how to pronounce his name when we finally do meet. Uh, if you're out there, Romeo, I'm planning to, to, uh, get in touch to do it. One of the striking things to me about this wonderful book, which is unique because it, it's unique in many ways, Andrew, in, it's, it's probably covers the most number of former, you know, past guests on the Into the Impossible podcast, including Richard Dawkins makes an

• 17:11Speaker 00

  Shop Black Friday appliance savings and get up to an extra $1,250 off select appliance purchases like GE Profile. Plus free delivery on appliance purchases of $398 or more. Make life easier with the fridge that doubles as your personal kitchen assistant. Or upgrade your laundry room with the Ultra Fresh Vent System Plus Front Load Washer. So your clothes stay fresh after every single wash. Shop Black Friday appliance savings with fast free delivery at The Home Depot. Offer valid November 4th through December 2nd at participating stores and at homedepot.com. US only. See store online for details.

• 17:41Brian Keating?

  In a book about astronomical simulations, which I find fascinating, as I already mentioned, the appearance by Nick Bostrom, who did me an incredible honor once. He sent me a PR pitch to come on the podcast and in it, it said, Nick Bostrom has appeared on the Joe Rogan Experience, the Lex Friedman Podcast, and Into the Impossible with Brian Keating. That was very flattering to be in that company. But I want to bring up another book, which is that by Katie Mack, who, as I understand, was your former office mate. Is that true?

• 18:13Speaker 02

  That's absolutely true. Yeah, we shared an office at Cambridge University for a couple of years, I think.

• 18:18Brian Keating?

  And I had her on many, many years ago. And towards the end of the book, she's interviewing folks, including Hirania Paris, who was a very close advisor to the Simons Observatory for many years. And Hirania says, "It's very depressing about the end of the universe. I don't know what else to say about it." It does provoke some perspective-taking, very depressing. We have a lot of utter darkness. And then she quotes Andrew Pontzen. "It momentarily makes me feel sad," agrees Andrew Pontzen. "Then I very quickly start worrying." What did you mean by that? What are you worried about? At the end of the universe, according to Katie, it won't come for many dozens of billions of years. What are you worried about?

• 18:55Speaker 02

  I remember Katie coming and talking to me about this. I can't remember exactly what I was saying to her, but I mean, I think my overall feeling about this is, yes, the end of the universe is a bit sad. You know, the idea that there is a finite lifespan to our universe and we can't just go on forever. But it's a very, very, very long time indeed, as Katie writes about. And I think, you know, my worry is nearly always over kind of much shorter term things. You know, I worry about where we're headed here on Earth in the here and now, or, you know, in decades or centuries, much more than I worry about the long-term future of cosmology. I suppose this is, this is part of the, and you must experience this.

• 19:38Speaker 02

  This is part of the weirdness of being a cosmologist, that on the one hand, one is grappling with some of these absolutely huge ideas and huge spans of time and things that have enormous implications. And yet, on the other, on the other hand, you live your life like, like anybody else. And there's a, there is a set of certain level of partition in your head that you, you can, you can comprehend intellectually what's going on in the universe, which is fantastic. And it's so exciting to be alive at a time when we can comprehend some of these things intellectually. But in a sort of instinctive human sense, I, I, I still kind of worry much more about

• 20:20Speaker 02

  what's going on around us right now than about the long-term fate of the galaxy or the sun or, or the universe as a whole.

• 20:27Brian Keating?

  Talk about the work that you did go into with the cosmic microwave background, beach ball that, you know, is also representative of how the bread gets buttered around the Keating household. Talk about the work that you did in the CMB. You, you seem to take a break from simulation research to do some actual pencil and paper theory, as I understand it. So explain what was the outcome of that research and why did it, it distract you from your, you know, putting particles in a box?

• 20:55Speaker 02

  As I talk about in the book, I had a moment of kind of crisis in my relationship to simulation.

• 21:00Brian Keating?

  Don't we? I tell my students, if you don't, you'll, you're not a good graduate.

• 21:04Speaker 02

  Yeah. I mean, I think, I think whenever you start digging into something, right? You see where the, the bodies are buried and it relates to the stuff we've already talked about, right? I mean, simulations and, and I think actually, you know, there's a serious point here that we can be guilty sometimes in the simulation community of, of portraying simulations as a literal snapshot of our universe and, and really telling us about the real universe. And of course it's not, you know, you start digging into it and you realize there are all of these simplifications, all of these approximations. And, and I had this kind of moment of crisis when I was doing my PhD where I thought, yeah, this just, it doesn't really make any sense at all. Of course, I later came out of that and I reached what I think is a kind of considered

• 21:49Speaker 02

  view of this, which is that there are some severe shortcomings, but it can also tell you something very important as well. Anyway, so I was having this crisis and I just looked around in the department, what else is there to do? And it was right around the time, of course, there was huge excitement around WMAP. The WMAP results, the first results from WMAP came out right around the time I was starting my PhD. And I was in a department where people were thinking about the cosmic microwave background a lot. There were characters like George F. Stadvieu there, who'd done a huge amount of early work on figuring out

• 22:25Brian Keating?

  Also a past guest on the podcast.

• 22:27Speaker 02

  Right, right. So he'd done a huge amount of work in figuring out what the properties of it would be. And so I just kind of went around asking people, you know, is there anything I can do on this? And it was Anthony Chana, who I'd end up working with closely, who would later play a big role in things like the Planck satellite. And he came up with this suggestion. He'd seen stuff about spinning universes. So one of the weird things about Einstein's general relativity is that it permits things like the entire universe to spin, which is quite a strange idea when you stop and think about it. Apart from anything,

• 23:07Speaker 02

  you might think, spin relative to what? And I think Einstein would have worried about this if he'd been thinking about spinning universes. I don't think he ever spent much time thinking about this.

• 23:20Brian Keating?

  I think his buddy Kurt Gödel.

• 23:23Speaker 02

  Yeah, indeed, indeed. Yeah, Gödel had written down what we call the metric for one of the first ever thought of spinning universes. And it had very strange properties, like, for example, it permits time travel. So Kurt Gödel had cited this as a kind of problem with general relativity. And I've never seen any evidence it bothered Einstein too much. Nonetheless, if our universe were to be found to be spinning, it would be profound. Because it would open up a lot of weirdnesses in general relativity. And it would also cast doubt on a lot of the prevalent theories of what was going on in the early universe.

• 24:03Speaker 02

  So these are theories like inflation doubts, despite being on somewhat dodgy territory. We do lean on quite heavily in cosmology. So it was interesting to work with these universes. And we realised we could calculate what imprint they would leave on the cosmic microwave background. And more specifically on the polarization. So Stephen Hawking, some years before, and also John Barrow, had done some work on calculating this. But we came along and we basically calculated everything there was to know about if you're in a spinning universe. What impact would that have on the cosmic microwave background? We did some quick check against the WMAP data as it was available then.

• 24:45Speaker 02

  And we found, yes, there's no evidence at all that our universe is spinning. And later on, a student of mine, many years later, came and did a really careful comparison with the Planck data, which was later a really high resolution view of the cosmic microwave background. And again, absolutely no evidence at all that the universe spins. So it's one of those things where, yeah, I mean, the answer, of course, is a bit disappointing. You know, if we'd found that the universe is spinning, we'd probably have been on our way to Stockholm to pick up a Nobel Prize or something. But on the other hand, I learnt a huge amount of physics doing this. It was a fun calculation and, you know, it's good to check these things.

• 25:24Brian Keating?

  In computational cosmology, there's often a distinction between what I call maybe verification. Are we solving the equations correctly? And validation, are we solving the correct equations? How do you approach this competing tension and the challenge that these two different competing ideas face in your research?

• 25:45Speaker 02

  I think one has to be incredibly incredibly clear about what we're trying to achieve when when we perform a simulation. So the first thing to say is that we know we never kind of fully correctly solve the equations that we're trying to solve. Even if we could solve the equations that we write down perfectly and and that is beyond us, you know, it's just not possible. As I was mentioning earlier, you know, we're not solving the equations of quantum field theory or something. So, you know, we are not literally reproducing the universe. So I I think we always have to start from that point, that there is no perfect solution to this problem. So we're always doing some approximations. So then you have to kind of approach it quite carefully to say, okay, what am I going to do then?

• 26:30Speaker 02

  I know I'm doing approximations. I know that there's some freedom within those approximations about what I choose to approximate and how. And therefore, to some extent, I can get different answers depending on what kind of approximations I make. So you can play the game of trying to match everything there is to match that's known about the universe, for example. So you you end up, this is through something called subgrid, which maybe we'll we'll come on to in detail. But I mean, you end up with the ability to kind of tweak things within your simulations and and you get different results. And you can play the game of let's try and and match everything there is to match in the universe.

• 27:10Speaker 02

  Let's go and see all the observations that have ever been taken, galaxies and galaxy clusters and and and try and match them. And some people do play this game, you know, they tweak tweak the parameters that they've got in front of them. And then they get a nice looking universe and and go, yep, we've done it. I think that that's, you know, not not the most helpful approach to this. Normally, what the the best thing to do is is to to be very open and honest about the fact that there is this freedom. But then to restrict what you're going to fit your simulation to. So you say, okay, I I want to make sure that I get certain things about the universe right.

• 27:51Speaker 02

  And I'm going to use the fact that I can tweak things within my simulation to do that. But then I'm going to draw a line and say, okay, that's that's what I calibrate on. And now here's my prediction. And in the end, you know, even in cosmology, we can make predictions. It's not like we can go out into the universe to test them, but we can make predictions for what a new telescope or new facility is going to discover. So I think that in essence, I think is the best, you know, the best simulation work does that. It kind of draws a line around what you're going to force the simulation to match and then makes a prediction for something new and then goes and tests it.

• 28:34Speaker 02

  And, and you know, sometimes those predictions do come out to be right, not always, but but sometimes they do. And, and then you feel like you're learning something, you're, you're stepping forwards in your understanding of, of how the universe sort of ticks on these large scales.

• 28:47Brian Keating?

  You may be aware of this, but about two years ago, in the summer of 2022, there was all this great hullabaloo, great controversy about whether or not the big bang even happened. Because these images from the James Webb Space Telescope appeared to show galaxies that were more mature, they were, they had disk shapes, they were rotating. And this was claimed by a cosmologist, amateur cosmologist named Eric Lerner, to be evidence that the big bang never happened, which was also the title of his book, When the Hubble Deep Field Came Out, you know, Before You Were Born. And the question I have for you is, you know, what, what kind of new advances in simulations

• 29:26Brian Keating?

  have there been to kind of dissuade you from the, even having any credibility in the, in the veracity of the hypothesis there was no big bang? How do simulations help us disprove that conjecture?

• 29:40Speaker 02

  If you'd focus in on was there a big bang specifically, then in fact, it's your field, the cosmic, the cosmic microwave background, which is the single most compelling piece of evidence. If you just want to point to one thing to say, look, this really happened, then it would be the cosmic microwave background and the fact that we can compute its detailed properties, and that was a prediction, right? And then, and then it turned out to be correct. So, so, you know, if, if that's your goal, just to show that the big bang happened loosely in the way that, that we think, then, then that's the way I would do it. In terms of the simulations, though, you know, these images from James Webb Space Telescope did cause a stir within, within the professional community. They showed galaxies which were brighter than I think most people were expecting to see.

• 30:27Speaker 02

  At very, very early times. So the point is, you know, James Webb is essentially looking back in time because it peers so deep into the universe, it's seeing light that's taken most of the age of the universe to reach us. And contrary to, I think, what most people were expecting, it saw a few really bright galaxies back in, in, at those early times. I was immediately sceptical that, that this told us anything kind of that earth-shattering about the history of the universe. Certainly, given the other evidence we have for the Big Bang, it would not, uh, kind of shift me from my belief in the fundamental correctness of, of the, the basic Big Bang picture. But I think it, it has provoked a lot of discussion

• 31:09Speaker 02

  amongst simulators about, you know, how did we go wrong here? You know, why, why is it that we weren't expecting to see such bright things? And a few, and a few simulators, by the way, point back to papers that they published before James Webb launched and said, actually, you know, we got it right. So it, it's not that everyone got it wrong. I think it's just that the sort of overall consensus with, within the community was expecting fewer really, really bright things. And honestly, we, I don't think it's a fully solved problem. A few things have happened since then. First of all, some of the really bright galaxies turned out to be misidentified. So, you know, in the really early days of James Webb, some of those things that were claimed to be really bright things, uh, very early in the history

• 31:53Speaker 02

  of the universe, turned out to be a bit of a mistake. And actually, they'd just been misidentified. They were actually closer to us and therefore not quite so ancient as, as we thought. But another thing that's probably going on is that the way that stars form in the very early universe isn't yet fully understood. That we kind of understand how stars form in today's universe, uh, because we have telescopes where we can see stellar nurseries where new stars are forming. But in the early universe, this probably happened in quite a different way. You know, the early universe was much denser, for one thing. And for another thing, it didn't have the kind of heavy elements that are in the universe today. And both of those things can make quite a big difference to how stars form, and therefore what exactly you would

• 32:38Speaker 02

  expect to see within a, within an early galaxy. But this is still being quite actively debated, and I, I think James Webb is certainly proving provocative.

• 32:48Brian Keating?

  Absolutely. No, it's fascinating. I, I talked with Chris Haywood at the Flatiron Institute not too long ago about the role of feedback. And, and you mentioned this already in the context of having a discussion about what subgrid, you know, resolution scale effects there are. But let's take a bigger step backwards and go back to somebody I'd heard about. You know, we have posters of her in the astronomy department. UC San Diego, our brand new astronomy department, you know, still unwrapping, still has that new department smell, you know, just just last year formed Allison Coyle as chair. Beatrice Tinsley. Again, I'd seen her, known some things about her, but she turns out to play, you know, sort of an equal role to, you

• 33:30Brian Keating?

  know, say, her counterpart, who actually did work here, which is Vera Rubin, who worked with Margaret and Jeffrey Burbidge. And in fact, this is where she learned how to do rotation curves. But Tinsley seems to play a role equal in many ways to that of Vera Rubin. Am I, am I wrong there? And if you'd be kind enough to kind of discuss her contributions to the field that we now

• 34:22Brian Keating?

  Study in galaxy simulations.

• 34:24Speaker 02

  I think you're absolutely right to put them on on an equal equal pedestal. You know, Vera Rubin obtained some of the most compelling evidence for dark matter in the early days as that as that theory of of what our universe is made out of started to gather steam. And Beatrice Tinsley, you know, around a similar time, maybe even slightly earlier, was thinking about how galaxies change and evolve over time. And she was particularly interested in this because at the at the time cosmology was a very, very different field. One of the big characters in cosmology at the time was Alan Sandage. And he was trying to use galaxies as what we would now call a standard candle.

• 35:07Speaker 02

  So in other words, he was assuming that if you see one galaxy and another galaxy in different parts of the universe, on average, they shine with the same brightness. I mean, not not down to every single object being identical, but just on average, they would shine with the same brightness. And he was using that supposed fact to map out the expansion of the universe by seeing where different galaxies were and how they were moving. And one of the key ingredients in this to figure out how far away is a galaxy is to look at how bright a light are you receiving from it? And if you want to use that as a way to figure out its distance and thereby map out the universe,

• 35:49Speaker 02

  you have to assume you know, at least on average, how bright a galaxy is. And so Alan Sandage was assuming that. And then along came Beatrice Tinsley and did a PhD thesis where she was pretty much operating in isolation. She she was very, very isolated. Um, she was working on a topic in in Texas that not many other people were interested in. But she got hold of access to an early computer and just went, OK, well, if we're assuming that all of these galaxies shine the same, then one of the things that we are sort of quietly assuming is that they don't change over time either. Because when we look to different distances, we're seeing.

• 36:32Speaker 02

  Being light delayed by different lengths of time. So if galaxies change over time, then Alan Sandage's assumptions are all wrong. So she programmed what we could now recognize as the kind of bare bones of a galaxy simulation into to sort of rudimentary computer that was available at the time. And she did the kind of exercise we were talking about a moment ago. She she didn't know, you know, exactly how to simulate a galaxy. Nobody did, still, nobody does. But what she was able to do was kind of scan over, well, what are the different things that might be happening inside a galaxy? And how would that affect how the galaxy shines? And in particular, as stars age, eventually they die.

• 37:16Speaker 02

  And so if stars are dying, then the brightness of a galaxy is dimming and its color is changing as well. So all of these different things she kind of programmed into to her code. And then she found that there is no way that you can make galaxies that just sit there staying the same brightness for all time. And so in this kind of calm, methodical way, using, you know, what what we can now recognize as one of the first ever galaxy simulations. She just undermined the whole of Alan Sandage's program. And that was, you know, one of the big programs of cosmology at the time. And and completely changed how people thought about how to map out the universe and and what it indeed the fate of the universe would be,

• 37:59Speaker 02

  which was a common one of the consequences of Alan Sandage's work. And for doing that, well, she she got a lot of grief. I think Alan Sandage was not happy at all. And and it took a long time for her ideas to kind of get fully recognized within the community.

• 38:18Brian Keating?

  Yes, she certainly did. And and if I'm not mistaken, she also kind of had some impact on on this notion of of feedback and evolution, which plays such a big role in in the context of the galactic simulations that actually faithfully represent what galaxies look like. Can you talk about, yeah, subgrid rules for feedback, the importance of feedback, resolution effects? And sort of the the cutting edge, the bleeding edge of this of this field?

• 38:47Speaker 02

  You're absolutely right. So so her work, in particular with Larson, kind of was was some of the first to really talk about the the effects of of feedback. And Perhaps the easiest way to think about this at first is to go back to the weather. And if you're sitting there on a hot day and you look up and you start to see these kind of bubbling up thunderclouds start to form. And in principle, you know, there's no reason why you couldn't represent that with the equations we were talking about earlier on, the Navier-Stokes equations. But if it's this this hot day or that started out nice and clear, you find that these clouds form, they're really quite small. You know, the clouds that form, they're a kilometer or two kilometers across.

• 39:32Speaker 02

  And this process, if you're trying to map the weather across a whole country, let alone the whole world. A kilometer or two is very, very small compared to the kind of ground you're trying to cover. And so early weather simulations, even up to a few years ago, could not simultaneously keep track of a whole country or the whole world, and also have the computer power to see tiny little clouds forming across that that huge area. So what people did was that they said, okay, well, we're not we're not going to see this happen just by solving the equations. We have to put in something else. We have to put in by hand an extra rule.

• 40:17Speaker 02

  Even though we know that it it it happens for a different reason, within the simulation, we're just going to put in a rule. And the rule just says, on a hot day, when the you know, the atmosphere is in a certain state, then thunderclouds will start to form. And we'll just force the simulation to do that because it's it it ought to do it itself, but it doesn't have the computational power to actually see that happening. You put it in by hand and in the in if you're doing weather simulations, you end up with, you know, a bunch of what we were talking about earlier on, parameters. You don't know exactly how fast these clouds will form, for example, so you have a knob that you can turn on your simulation. You can you can make them form faster or slower. You don't know exactly how fast they should form. You dial the knob up and down.

• 41:02Speaker 02

  Until you find you're starting to get good weather forecasts and then you go, okay, I've I've figured out what that knob, what value that knob should be set to. It's sort of dirty, you know, it's it's it's not the vision that we like to have of physics as okay, we can just derive everything from a beautiful equation and and derive it all on a piece of paper and figure out everything that should happen. But it works, it works extremely well for weather simulations. It also works in the context of climate that you were talking about earlier on. We we know that because we can look at what's called paleoclimate data. So in other words, data from ice cores and and things like that that tell us how the climate was in the past.

• 41:46Speaker 02

  And again, you need these subgrid rules, but you you can tweak them and and then you can make predictions for for new bits of data that you're getting and so on. So that's what happens in the world of of weather and climate. And we have to do it in the universe as well. And for for us in the universe, small is not a few kilometres across. It's light years across, in fact, because the universe is so vast that a galaxy is thousands and thousands of light years across. So if you want to know what's going on across that whole galaxy, then you can't really also have the computer power to be seeing what's going on on sort of one light year scales.

• 42:28Speaker 02

  And that includes things like stars forming, you know, really important things like the actual stars that we're going to see in the galaxy form. Of course, they they, you know, a star is quite a small thing on the kind of scales that we're talking about. So you need to put in these subgrid rules for how the stars form and also what the stars then do to their surroundings. So the stars dump lots of energy in the form of radiation or in in the case of massive stars, after a few million years, they can explode as as supernovae. And that also dumps a whole bunch of energy into the galaxy, into the the very kind of large scale system that you care about. So somehow or other, you need to capture all of this and you're not going to.

• 43:12Speaker 02

  Do it just by solving the the nice neat laws of physics that you would like to solve. So you you have to put in a subgrid. Tinsley and Larson were some of the first to kind of really recognize this in in that context of galaxy formation.

• 43:26Brian Keating?

  The other thing that kind of surprised me to encounter in in the way that I did in this book, the universe in a box was sort of the notion of a universe in in a box. In other words, why be so, as past guest Andrea Linde said, so universe specific? Why not talk about the multiverse and and have that as our baseline? Talk about the way the multiverse could come into play, both in standard, you know, cosmological simulations as well as in quantum computers and the Everettian many worlds interpretation of quantum mechanics.

• 44:02Speaker 02

  In a straightforward sense, we can make multiverses within our computers in the sense that we can perform multiple simulations. So we don't just need to do one brilliant simulation. We we, of course, do many simulations, partly because we want to try different things. But also, we can do things like change some of the fundamental parameters that that describe our universe. For instance, how much matter does it have in it? So you can perform simulations that have a little bit more or a little bit less and find out how do they behave relative to to the others. In simulation world, we can, universes are cheap. We can, we can generate as many of them as we like and see, see how they behave.

• 44:45Speaker 02

  When it comes to the real universe and whether we live in a a multiverse, in other words, is our actual universe one of many? I have, you know, always been struck by this idea of of the Everettian multiverse. It's this incredible notion put forward by Hugh Everett that one way to make sense of some of the weirdnesses of quantum mechanics and and in particular, the kind of weirdness of something called wave function collapse, which is where when you do an experiment on quantum scales, it seems like what you're actually experimenting on changes when you observe it. And that and that is this this notion of wave function collapse.

• 45:30Speaker 02

  That is very uncomfortable and this idea that Hugh Everett had was that an alternative to that is to imagine that we actually live in this much, much richer multiverse, which in effect it kind of splits out into multiple distinct universes where lots of different possibilities play out in parallel. And that, you know, mathematically speaking, this kind of holds water. It does work as a way of explaining some of the oddities of of quantum mechanics. Whether it's comfortable or not is more a matter of taste. But as we kind of explore quantum systems more and more and we build more and more

• 46:13Speaker 02

  complicated quantum systems, I mean, I think it's David Deutsch who has pointed out that, you know, if you can build a quantum computer, if you can actually start building quantum computers that take advantage of these weird quantum effects, you're almost left with no alternative but to imagine that this computer works by by taking advantage of this much, the way we call it, is a much bigger state space. But, you know, colloquially, it's like that there really are kind of parallel universes doing a lot of the computational heavy lifting if those if those computers work. And when you look at the universe, you find it has quantum imprints in it.

• 46:52Speaker 02

  In fact, when we perform our simulations, we have to start from a state which is dominated by randomness, which is thought to occur from quantum effects during a period of inflation that that we mentioned briefly earlier on. If that's true, then when did the universe undergo wave function collapse to become the one universe we see today? Well, the the sort of the easiest answer in a sense, as Hugh Everett would say, is it never did. That, you know, you don't have to imagine some weird moment at which the universe suddenly collapses and becomes a classical single universe.

• 47:36Speaker 02

  It can it can just never undergo wave function collapse and these. Parallel universes can just exist side by side. That's the idea, and of course, it's fiercely controversial. But but I I've always found it actually makes sense of a huge amount of stuff that we already know.

• 47:52Brian Keating?

  Along the lines of things that were claimed to be known at least two years ago in December twenty twenty two, physicists led by friend Maria Spira-Pulu at Caltech claim that we found a quantum system that exhibits key properties of a gravitational wormhole, yet is sufficiently small to implement on today's quantum hardware. What's the current status of that claim of physicists observing a wormhole or Einstein Rosen bridge using a quantum computer?

• 48:20Speaker 02

  So I'm not familiar with the exact with that exact work. But I mean, there's a lot of work being done into quantum systems that in some sense mimic aspects of quantum gravity, including quantum black holes and and like you say, quantum wormholes. So I can't speak to that particular paper, but I mean, the key thing about all of these things is that they have quite a lot in common with simulations, in a sense. In fact, I'm involved in a collaboration which is looking at aspects of the early universe using a quantum system. And and other parts of that collaboration are looking at at quantum black holes. Now, the thing to bear in mind is that they are always analogs, right? So it it is a system that you've carefully designed

• 49:06Speaker 02

  to have some of the same properties that we think these real systems in the in the real universe should have. So if it's a quantum black hole or a quantum wormhole or so on, that they are that they're designed quite carefully to to kind of mimic some of those things. And you can learn a certain amount by by studying those in a lab. It's a really they're really fascinating experiments. What one has to be really careful to do, just the same way as we do with simulations, you know, you have to draw kind of sharp line and say, OK, yes, we can learn a lot by doing this. But it is not the same as literally having a quantum black hole that you can play with in your lab or literally having a quantum wormhole that you can play with in in your lab. The analogy that we're drawing between these two quite different bits of of physics will have limitations to it.

• 49:51Speaker 02

  And I think I'm quite confident in saying that in in a general sense, even though I I'm not like at a detailed level familiar with that particular work that you mentioned.

• 50:01Speaker 00

  Shop Black Friday appliance savings and get up to an extra $1,250 off select appliance purchases like GE Profile. Plus, free delivery on appliance purchases of $398 or more. Make life easier with the fridge that doubles as your personal kitchen assistant. Or upgrade your laundry room with the Ultra Fresh Vent System Plus Front Load Washer. So your clothes stay fresh after every single wash. Shop Black Friday appliance savings with fast free delivery at The Home Depot. Offer valid November 4 through December 2 at participating stores and at homedepot.com. U.S. only. See store online for details.

• 50:31Brian Keating?

  As we wrap up, I want to cover a couple of things that have come to mind recently. And that has to do, first of all, with the state of the art of high-performance computing. And also, this is called ethics of using them and their carbon footprints, et cetera. We just had in the United States announced a potential acquisition by Microsoft of the defunct for 40 years, Three Mile Island or more, nuclear power plant. So I've learned that the carbon impact, the energy demands of a single AI search is about nine or 10 times the conventional search query energy demands. What are the impacts of that on

• 51:13Brian Keating?

  the growth curve? Because unless I'm mistaken, some of the simulations, at least the ones that aren't for the military, they seem to saturate because as the computers get more powerful, more people want to use them. So talk about the energy impact, the ethics of doing these types of simulations, and what is the state of the art? What are you excited about?

• 51:29Speaker 02

  The energy impact is very much on people's minds who work in this field, as you might expect. We tend to now estimate what carbon footprint we are paving. And it probably won't surprise you to learn that it is a tiny fraction of what the world's data centers are using. It's a really minute fraction. So to give you an example, something called the Flamingo simulations, that were some of the largest simulations which have ever been performed. They just came out in the last few months. They have a climate impact that's roughly equivalent to two people taking a

• 52:11Speaker 02

  Transatlantic flights, so say London to New York in economy. So I think I think it's a return trip, actually. So so you know, so it's like two two people taking a return trip, London to New York and and back. That is the biggest or maybe close to the biggest cosmological simulation ever performed. That's not to say we should ignore that, um, but it's not it's not an absolutely enormous thing. And it certainly doesn't uh start to get to the point where you worry about it next to, say, the explosive growth in AI, which which you mentioned, just because of its much broader applications, of course, is is of course gonna gonna have a much bigger impact.

• 52:52Speaker 02

  People do worry about this, and I think it makes us think about making sure that we are making most efficient use of the resources that we have at our disposal, which we should be doing anyway. Right? I mean, in the end, it's taxpayers who who are funding these these things. So, you know, it's right that we look at the efficiency, but I don't think that um it it's it's sort of having a noticeable impact in in in and of itself. It is very exciting, you know, that we we live in at a time where computer power continues to grow a very, very fast. But what we're finding is that a lot of the the really state of the art computational facilities now work with quite specialized chips, quite specialized hardware,

• 53:35Speaker 02

  and and what we used to call graphics processor units, GPUs. These are the things that have become the bread and butter of AI, because they are extremely good at doing very, very repetitive operations in very large numbers. And this is proving a challenge that that, you know, the progress in computer power up to a few years ago, um, we could just sit on the back of that, and and it more or less suited our needs if we wanted to simulate the universe. Now, because it's become somewhat more specialized, a lot of the direction of growth is towards AI. That ability to do loads and loads of things repetitively is not ideally suited to simulating the universe, but mainly because when you try to simulate the universe, although there are some operations in there that are very repetitive.

• 54:21Speaker 02

  There's also a lot of communication that has to go on so that one bit of the universe talks to another bit of the universe through gravity and through sending out pressure waves and so on. So there's a lot of communication between different bits of the universe. And this is where it gets really hard to take advantage of the latest hardware because AI tends to be all about just having lots of, you know, lots of your AI queries going on in parallel, but they're completely unaware of the other AI queries that are going on. So we've got, we've got some big challenges in trying to make good use of the hardware platforms that are becoming available.

• 55:00Brian Keating?

  And last but not least, I couldn't help but think when you talk in the book about all these sort of neologisms like smarticles or smart, I think you call them smarticles, right? This is something that you've coined. I thought back to this famous poem that is obliquely perhaps referenced in your, in your book. You discuss the discovery of Neptune, which is, you know, kind of the first instance of people discovering dark matter. The poem I recently learned of reading it with all my kids, when I heard the learned astronomer, who's so depressed, Walt Whitman's very depressed about the night sky being brought to the mere fact of numbers. And then of course Feynman comes back. He also makes

• 55:44Brian Keating?

  a lot of appearances in this book, but to say, well, why can't a scientist talk about the beauty just as much as a poet? You know, why? Because he knows that Jupiter is not a god, but is made of methane. How do you react to this, you know, kind of maybe cheapening of the beauty of astronomical appearances and so forth? The cosmology, cosmos means beauty after all in Greek. Does what you do maybe take away from the beauty of understanding the universe? Or does it enhance it?

• 56:12Speaker 02

  I think it enhances it. I mean, I guess I'm with Feynman on this as you, as you might predict. I think, you know, understanding something adds layers to your appreciation of it. And it goes back to something we were discussing earlier as well, that, you know, as a cosmologist, you can go and make calculations that These things and we can build these computer simulations that are giving us different insights into the way that the universe works. That doesn't change my fundamental relation to my life. Or to the way that I look at the universe. I still kind of walk along and. When I see the night sky, it still makes me feel the same things and knowing.

• 56:52Speaker 02

  More about it, I think can only enhance one's one's appreciation for it. It doesn't. It doesn't take away and I suddenly just feel in awe of how how amazing. The universe that we've got to live in is.

• 57:08Brian Keating?

  Andrew Ponsen now at Durham. Uh, thank you for this wonderful conversation and this wonderful book, The Universe in a Box. Simulations and the quest to code the cosmos. A lot of K alliterative uh phraseology there. Andrew, thanks so much.

• 57:21Speaker 02

  Thanks. I appreciate it.

• 57:22Speaker 00

  Shop Black Friday appliance savings and get up to an extra $1,250 off select appliance purchases like GE Profile. Plus free delivery on appliance purchases of $398 or more. Make life easier with the fridge that doubles as your personal kitchen assistant. Or upgrade your laundry room with the Ultra Fresh Vent System Plus Front Load Washer. So your clothes stay fresh after every single wash. Shop Black Friday appliance savings with fast free delivery at The Home Depot. Offer valid November 4th through December 2nd at participating stores and at homedepot.com. U.S. only. See store online for details.

• 57:53Speaker 01

  Progressive's Accident Response can contact emergency services if you're in a major accident. The only thing worse than someone you love being in an accident is not being able to help them. So Progressive is here to help when you can't. So get peace of mind knowing your loved ones are protected with Progressive's Accident Response. Learn more at progressive.com. Progressive Casualty Insurance Company and Affiliates. Limited availability in California. Driver is responsible for emergency and towing charges not covered by the policy. Coverage subject to policy terms and conditions.