WEBVTT 1 00:00:03.399 --> 00:00:07.719 Welcome to Bedtime Astronomy. Explore the wonders of the cosmos 2 00:00:07.759 --> 00:00:12.279 with our soothing Bedtime Astronomie podcast. Each episode offers a 3 00:00:12.359 --> 00:00:16.320 gentle journey through the stars, planets, and beyond, perfect for 4 00:00:16.399 --> 00:00:20.239 unwinding after a long day. Let's travel through the mysteries 5 00:00:20.239 --> 00:00:22.440 of the universe as you drift off into a peaceful 6 00:00:22.480 --> 00:00:26.879 slumber under the night sky. 7 00:00:27.679 --> 00:00:29.800 So I want you to imagine something for a second. 8 00:00:29.960 --> 00:00:33.600 Try to, like, picture looking at a photograph of a galaxy. 9 00:00:33.719 --> 00:00:38.359 Oka like a massive swirling vortex of billions of stars, 10 00:00:38.759 --> 00:00:41.719 you know, laced with those glowing veins of purple gas 11 00:00:41.719 --> 00:00:44.399 and thick dark rivers of cosmic dust. 12 00:00:44.520 --> 00:00:47.200 Write something like a classic image beamed back from the 13 00:00:47.280 --> 00:00:49.079 James Webb Space Telescope. 14 00:00:49.119 --> 00:00:54.359 Exactly, a stunning high definition space image. Now imagine a 15 00:00:54.399 --> 00:00:58.079 second photograph right next to it, and it looks completely 16 00:00:58.119 --> 00:01:02.159 identical down to the pixel right, the same complex spiral arms, 17 00:01:02.200 --> 00:01:05.799 the same chaotic star forming regions, the exact same distribution 18 00:01:05.840 --> 00:01:08.640 of light and shadow. But there's a fundamental difference between 19 00:01:08.640 --> 00:01:09.319 these two images. 20 00:01:09.359 --> 00:01:10.879 And it's a pretty massive difference. 21 00:01:10.959 --> 00:01:14.439 Yeah, because one is a photograph of our actual universe, right, 22 00:01:14.480 --> 00:01:18.959 the product of fourteen billion years of messy, chaotic natural evolution, right, 23 00:01:19.519 --> 00:01:22.079 But the other one is essentially a hallucination. It is 24 00:01:22.120 --> 00:01:24.760 a product of pure math running on silicon chips in 25 00:01:24.799 --> 00:01:27.599 a basement somewhere here on Earth. It's incredible to think 26 00:01:27.599 --> 00:01:30.560 about it really is. So what if we could build 27 00:01:30.560 --> 00:01:34.799 a universe inside a computer so mathematically perfect, like, so 28 00:01:35.159 --> 00:01:38.840 unfathomably detailed that if you put those two images side 29 00:01:38.840 --> 00:01:42.680 by side, even professional PhD holding astronomers couldn't tell you 30 00:01:42.680 --> 00:01:43.400 which one was real. 31 00:01:43.480 --> 00:01:46.959 I mean, it sounds completely like science fiction, but it 32 00:01:47.040 --> 00:01:52.040 is the current, very real frontier of computational astrophysics. 33 00:01:51.400 --> 00:01:52.799 Today, which is just wild to me. 34 00:01:53.000 --> 00:01:57.200 It is we are no longer just sketching rough approximations 35 00:01:57.200 --> 00:02:01.280 of how galaxies might form. We are synthesizing entire cosmoses 36 00:02:01.359 --> 00:02:02.640 from absolute scratch. 37 00:02:02.799 --> 00:02:03.159 Wow. 38 00:02:03.280 --> 00:02:06.879 Yeah, the boundary between observed physical reality and simulated digital 39 00:02:06.879 --> 00:02:09.439 reality is becoming well incredibled porous. 40 00:02:09.479 --> 00:02:12.680 Okay, let's unpack this because simulating an entire universe from 41 00:02:12.680 --> 00:02:14.400 the literal Big Bang all the way to the present 42 00:02:14.479 --> 00:02:16.680 day sounds like a task that would just melt any 43 00:02:16.680 --> 00:02:17.479 computer on Earth. 44 00:02:17.520 --> 00:02:20.360 It definitely pushes the absolute limits of our current hardware. 45 00:02:20.400 --> 00:02:23.840 I bet. Yet this massive, decade long scientific effort has 46 00:02:23.919 --> 00:02:26.759 just achieved exactly this. We are looking at a project 47 00:02:26.840 --> 00:02:30.400 that successfully simulated the evolution of galaxies by tracking the 48 00:02:30.439 --> 00:02:34.840 absolute smallest like microscopic specs of cosmic. 49 00:02:34.520 --> 00:02:37.240 Dust, down to the fundamental chemistry. 50 00:02:37.120 --> 00:02:41.000 Right, and in doing so, they essentially rescued the entire 51 00:02:41.120 --> 00:02:44.039 standard model of the universe from being thrown in the 52 00:02:44.039 --> 00:02:45.159 scientific garbage bin. 53 00:02:45.319 --> 00:02:47.159 They really did. It was a close call for a 54 00:02:47.199 --> 00:02:47.719 minute there. 55 00:02:47.840 --> 00:02:50.639 Plus they figured out a way for us to literally 56 00:02:50.719 --> 00:02:54.439 hear this virtual universe, which we will definitely get into. 57 00:02:54.560 --> 00:02:58.080 Oh, the sonification stuff is fascinating, but the scope of 58 00:02:58.120 --> 00:03:01.240 the simulation itself is genuine only difficult to wrap your 59 00:03:01.240 --> 00:03:04.000 head around, even for the people actively. 60 00:03:03.479 --> 00:03:04.960 Working on it, I can imagine. 61 00:03:05.000 --> 00:03:08.800 But this endeavor really highlights a core philosophy of science, 62 00:03:08.840 --> 00:03:11.560 which is that knowledge is most valuable when it is 63 00:03:11.680 --> 00:03:13.960 rigorously applied to solve a crisis. 64 00:03:14.159 --> 00:03:16.000 And there was definitely a crisis. 65 00:03:15.680 --> 00:03:21.280 Oh absolutely. For decades, astrophysicists have understood the broad theoretical 66 00:03:21.360 --> 00:03:22.319 strokes of how the. 67 00:03:22.360 --> 00:03:25.159 Universe expands, that the big picture stuff, right. 68 00:03:25.360 --> 00:03:28.319 But whenever they tried to actually build a digital model 69 00:03:28.319 --> 00:03:32.000 of it. The simulated galaxies always looked well, they just looked. 70 00:03:31.800 --> 00:03:33.280 Wrong, like how wrong. 71 00:03:33.319 --> 00:03:37.240 They were too clunky, the colors were off, the sizes 72 00:03:37.360 --> 00:03:41.000 just didn't match reality at all. The breakthrough here from 73 00:03:41.000 --> 00:03:45.120 this massive international initiative known as the Collibo project is 74 00:03:45.159 --> 00:03:48.520 that it proves something fundamental, which is, if you want 75 00:03:48.560 --> 00:03:52.879 to get the massive billion light year macro universe right, 76 00:03:53.319 --> 00:03:56.240 you have to perfect the microscopic physics first. 77 00:03:56.439 --> 00:03:59.120 See. And that's the part that gets me because for 78 00:03:59.159 --> 00:04:02.080 the longest time we weren't just struggling with the microphysics right, 79 00:04:02.360 --> 00:04:04.520 we were actively blatantly ignoring them. 80 00:04:04.599 --> 00:04:06.800 Yeah, that's as one way to put it. 81 00:04:07.120 --> 00:04:08.960 I want to talk about this concept of the heat 82 00:04:09.000 --> 00:04:12.400 barrier in early simulations because when I first learned about this, 83 00:04:12.919 --> 00:04:14.759 it sounded completely absurd. 84 00:04:14.960 --> 00:04:18.519 It is a bit counterintuitive until you understand the math behind. 85 00:04:18.279 --> 00:04:20.800 It, right. But in all the older legacy simulations of 86 00:04:20.839 --> 00:04:23.639 the universe, scientists put a hard floor on the temperature 87 00:04:23.680 --> 00:04:24.879 of their virtual space. 88 00:04:24.600 --> 00:04:27.720 Didn't They They did a very strict temperature limit. 89 00:04:27.800 --> 00:04:31.680 They artificially prevented the gas inside their virtual galaxies from 90 00:04:31.720 --> 00:04:35.160 cooling down below about ten thousand degrees fahrenheit. 91 00:04:34.959 --> 00:04:37.639 Which just to put that into perspective, for you, is 92 00:04:37.680 --> 00:04:40.079 significantly hotter than the surface of our sun. 93 00:04:40.759 --> 00:04:41.480 Wait? Really? 94 00:04:41.560 --> 00:04:45.560 Oh yeah? So imagine trying to code an entire virtual 95 00:04:45.680 --> 00:04:50.399 universe where literally nothing, no cloud of gas, no planetary nebula, 96 00:04:50.920 --> 00:04:53.959 no empty void, is allowed to be cooler than a 97 00:04:54.079 --> 00:04:55.319 raging stellar surface. 98 00:04:55.639 --> 00:04:57.759 It sounds like trying to study how a delicate French 99 00:04:57.800 --> 00:05:00.800 pastry is baked, but your oven only have one setting, 100 00:05:00.800 --> 00:05:02.160 and that setting is blowtorch. 101 00:05:02.319 --> 00:05:04.199 That's a great way to describe it, Like, how do. 102 00:05:04.160 --> 00:05:06.480 You expect to bake a croissant, or in this case, 103 00:05:06.560 --> 00:05:10.279 forge a star if every ingredient in your digital kitchen 104 00:05:10.399 --> 00:05:13.519 is instantly vaporized into superheated plasma. 105 00:05:13.600 --> 00:05:15.839 Well, you don't. You simply don't. And that was the 106 00:05:15.879 --> 00:05:20.000 glaring flaw of late twentieth and early twenty first century cosmology. 107 00:05:20.120 --> 00:05:22.560 It seems like such an obvious oversight, it does. 108 00:05:22.759 --> 00:05:25.160 But to be fair to those early pioneers, it wasn't 109 00:05:25.199 --> 00:05:26.439 that astronomers didn't want their. 110 00:05:26.360 --> 00:05:28.480 Universes to be cold, Okay, so why do it? 111 00:05:28.480 --> 00:05:31.639 It was that modeling cold gas was computationally lethal to 112 00:05:31.680 --> 00:05:32.360 their machines. 113 00:05:33.079 --> 00:05:37.279 Lethal How like, is cold gas harder to calculate than 114 00:05:37.319 --> 00:05:39.920 hot gas. I would assume hot things are more chaotic. 115 00:05:40.120 --> 00:05:45.839 It's actually the exact opposite. Modeling cold gas is exponentially harder. Really, Yeah, 116 00:05:46.160 --> 00:05:48.399 when we talk about gas moving through space, we were 117 00:05:48.439 --> 00:05:53.279 really talking about hydrodynamics. In astrophysics, we treat these massive 118 00:05:53.319 --> 00:05:55.560 clouds of hydrogen and helium. 119 00:05:55.439 --> 00:05:57.519 Like a fluid, like water flowing in a river. 120 00:05:57.839 --> 00:06:01.079 Sort of. Yeah, we have to calculate their pressure, their viscosity, 121 00:06:02.240 --> 00:06:05.120 how they flow, and crucially, how they react to the 122 00:06:05.120 --> 00:06:07.040 gravitational pull of dark matter. 123 00:06:07.160 --> 00:06:08.120 Okay, that makes sense. 124 00:06:08.279 --> 00:06:08.439 Now. 125 00:06:08.480 --> 00:06:11.600 When gas is kept artificially hot at ten thousand degrees, 126 00:06:11.720 --> 00:06:14.560 it has a lot of thermal pressure. It's smooth, it 127 00:06:14.639 --> 00:06:18.879 expands outward evenly. It predictable exactly. The math to track 128 00:06:18.920 --> 00:06:23.079 a smooth expanding balloon of hot fluid is relatively straightforward. 129 00:06:23.079 --> 00:06:26.079 For a supercomputer, it can handle that easily. But what 130 00:06:26.199 --> 00:06:28.120 happens to a gas when you cool it down? 131 00:06:28.920 --> 00:06:31.560 It condenses, right, It shrinks and gets thicker, exactly. 132 00:06:31.560 --> 00:06:35.079 It loses all that outward thermal pressure. So gravity suddenly 133 00:06:35.160 --> 00:06:38.800 takes over and the gas starts clumping together into microscopic, 134 00:06:38.920 --> 00:06:42.680 incredibly dense pockets. Oh I see, it becomes chaotic. And 135 00:06:42.920 --> 00:06:48.160 highly nonlinear. To calculate the hydrodynamics of cold, clumpy gas, 136 00:06:48.600 --> 00:06:53.000 your computer grid needs a massive, massive increase in resolution. 137 00:06:52.600 --> 00:06:54.680 Because there's more stuff happening in a smaller area. 138 00:06:54.800 --> 00:06:58.839 Right, You have to track millions of tiny, independent, highly 139 00:06:58.920 --> 00:07:02.759 dense collapsing now instead of just one big smooth cloud. 140 00:07:02.879 --> 00:07:05.319 So the computers just couldn't handle the math exactly. 141 00:07:05.399 --> 00:07:08.560 If older supercomputers tried to let the gas drop below 142 00:07:08.600 --> 00:07:12.000 ten thousand degrees, the simulation would essentially choke on the 143 00:07:12.000 --> 00:07:15.560 math and crash. The algorithm simply couldn't handle the resolution 144 00:07:15.680 --> 00:07:17.639 required to track all those tiny clumps. 145 00:07:17.759 --> 00:07:21.600 But the physical reality we actually live in completely relies 146 00:07:21.720 --> 00:07:22.759 on that cold gas. 147 00:07:22.800 --> 00:07:24.240 It lies entirely on it. 148 00:07:24.279 --> 00:07:26.600 Right, Like, if the real universe couldn't cool down, the 149 00:07:26.600 --> 00:07:29.360 gas would never clump, stars would never form, and you 150 00:07:29.399 --> 00:07:31.120 and I wouldn't be sitting here talking about it. 151 00:07:31.240 --> 00:07:34.279 No, we wouldn't. And this is the grand irony of 152 00:07:34.360 --> 00:07:38.959 early cosmological simulations. Stars only form in molecular clouds that 153 00:07:39.000 --> 00:07:42.120 are incredibly cold. We are talking just a few degrees 154 00:07:42.160 --> 00:07:43.360 above absolute zero. 155 00:07:43.519 --> 00:07:44.319 Wow, that cold. 156 00:07:44.439 --> 00:07:48.199 Yes, at those freezing temperatures, the gas has virtually zero 157 00:07:48.279 --> 00:07:49.040 thermal pressure. 158 00:07:49.160 --> 00:07:51.720 Pushing outward, so gravity just completely takes over. 159 00:07:51.800 --> 00:07:55.720 Precisely without that outward push, the inward pull of gravity 160 00:07:55.839 --> 00:07:59.360 finally wins the tug of war. The cold gas collapses 161 00:07:59.399 --> 00:08:02.279 in on itself, becoming denser and denser until the pressure 162 00:08:02.319 --> 00:08:05.839 at the core is so extreme that it ignites nuclear fusion. 163 00:08:05.519 --> 00:08:06.800 And boom, a star is born. 164 00:08:07.079 --> 00:08:10.879 Exactly, But if your virtual universe is artificially locked at 165 00:08:10.920 --> 00:08:14.040 ten thousand degrees, the gas is just moving too fast. 166 00:08:14.079 --> 00:08:15.560 It has too much kinetic. 167 00:08:15.279 --> 00:08:17.800 Energy, so it just zips right past itself right. 168 00:08:17.839 --> 00:08:20.120 Gravity can never grab onto it long enough to forge 169 00:08:20.120 --> 00:08:23.600 a star. What's fascinating here is how this forced scientists 170 00:08:23.680 --> 00:08:25.879 to reevaluate their entire approach. 171 00:08:26.120 --> 00:08:29.680 So wait, if the physics in those older simulations physically 172 00:08:29.720 --> 00:08:33.600 prevented stars from format like mathematically made it impossible, how 173 00:08:33.600 --> 00:08:36.080 do they get galaxies? Do they just hack the code? H? 174 00:08:36.159 --> 00:08:36.480 Yeah? 175 00:08:36.559 --> 00:08:40.320 Essentially, they used a conceptual band aid called subgrid physics. 176 00:08:40.480 --> 00:08:42.919 Subgrid physics it sounds like a polite way of saying 177 00:08:42.960 --> 00:08:43.600 we made it. 178 00:08:43.639 --> 00:08:47.759 Up, well, kind of. Because the computer grid couldn't zoom 179 00:08:47.799 --> 00:08:51.440 in far enough to simulate actual cooling and collapsing, they 180 00:08:51.440 --> 00:08:54.279 wrote a statistical workaround an arbitrary rule. 181 00:08:54.399 --> 00:08:55.159 How did that work? 182 00:08:55.399 --> 00:08:58.440 The code basically said, Hey, if a pocket of hot 183 00:08:58.480 --> 00:09:02.200 gas reaches a certain density threshold, just pretend it magically 184 00:09:02.200 --> 00:09:05.159 cooled down, delete the gas, and spawn a star particle 185 00:09:05.200 --> 00:09:05.799 in its place. 186 00:09:06.360 --> 00:09:09.960 Wow. So they were just spawning in stars like items 187 00:09:09.960 --> 00:09:12.720 in a video game based on a rough guess. 188 00:09:12.639 --> 00:09:16.200 Exactly, just dropping them in when the math looked close enough. 189 00:09:16.279 --> 00:09:19.399 That means for years an enormous chunk of our understanding 190 00:09:19.480 --> 00:09:23.320 of galaxy evolution was built on a giant blind. 191 00:09:22.960 --> 00:09:24.679 Spot, a huge blind spot. 192 00:09:24.759 --> 00:09:27.600 We were trying to map the cosmos while completely ignoring 193 00:09:27.639 --> 00:09:30.000 the actual environment where galaxies build themselves. 194 00:09:30.200 --> 00:09:32.159 Right, it was an educated guest, But you can only 195 00:09:32.159 --> 00:09:35.159 rely on statistical band aids for so long before your 196 00:09:35.240 --> 00:09:38.440 virtual universe diverges so far from reality that it just 197 00:09:38.559 --> 00:09:40.200 stops being scientifically useful. 198 00:09:40.320 --> 00:09:41.919 Right, Eventually the cracks are going to show. 199 00:09:41.960 --> 00:09:44.200 Absolutely Eventually you have to find a way to break 200 00:09:44.200 --> 00:09:47.200 that heat barrier and let the virtual cosmos freeze. 201 00:09:47.360 --> 00:09:50.519 And breaking that barrier wasn't just a matter of like 202 00:09:50.559 --> 00:09:53.919 waiting for Apple or IBM to build a faster microchip, right, 203 00:09:54.000 --> 00:09:57.200 You actually had to write new physics into the simulation. 204 00:09:57.360 --> 00:09:59.120 Yes, entirely new physics. 205 00:09:59.159 --> 00:10:02.279 And this is where the claw reproject completely changes the game. 206 00:10:02.799 --> 00:10:05.879 To cool the universe down, they had to introduce two 207 00:10:06.519 --> 00:10:10.799 highly specific ingredients that had never been successfully modeled on 208 00:10:10.840 --> 00:10:14.960 this scale before, right, right, molecular hydrogen and cosmic dust. 209 00:10:14.759 --> 00:10:16.519 Yes, and let me stop you there, because it is 210 00:10:16.600 --> 00:10:19.120 vital to understand why those two things go hand in hand. 211 00:10:19.399 --> 00:10:23.080 You cannot have cold molecular hydrogen in space without cosmic 212 00:10:23.159 --> 00:10:27.039 dust rate. Dust is the great enabler of the universe. 213 00:10:26.840 --> 00:10:30.120 Which is such a wild paradigm shift because normally, if 214 00:10:30.120 --> 00:10:32.559 you ask an amateur astronomer about dust, they'll just groan. 215 00:10:33.039 --> 00:10:33.879 Oh, they hate it. 216 00:10:34.200 --> 00:10:36.519 Right, dust is the annoying haze that blocks the light 217 00:10:36.559 --> 00:10:39.679 from the beautiful glowing nebulae behind it. It's an obstacle. 218 00:10:40.120 --> 00:10:43.480 But here dust is essentially the vip of the universe. 219 00:10:44.600 --> 00:10:46.480 So what does this all mean? Why is it so important? 220 00:10:46.759 --> 00:10:49.600 Well, because the vacuum of space is an incredibly hostile 221 00:10:49.639 --> 00:10:53.279 place to build a molecule. The primary ingredient for a 222 00:10:53.360 --> 00:10:57.639 star is molecular hydrogen, which is just two hydrogen atoms 223 00:10:57.679 --> 00:10:58.440 bonded together. 224 00:10:58.600 --> 00:10:59.639 Okay, simple enough. 225 00:11:00.120 --> 00:11:04.159 In the vast empty expanse of an early galaxy. Trying 226 00:11:04.200 --> 00:11:07.559 to get two single hydrogen atoms to bump into each 227 00:11:07.600 --> 00:11:10.159 other and bond is nearly impossible. 228 00:11:09.639 --> 00:11:11.279 Because space is just too big. 229 00:11:11.399 --> 00:11:14.360 The space is too vast, the atoms are moving too fast, 230 00:11:14.399 --> 00:11:16.519 and if they do happen to collide, they usually just 231 00:11:16.600 --> 00:11:19.320 bounce off each other. Why do they bounce because they 232 00:11:19.360 --> 00:11:22.360 have nowhere to dump their excess kinetic energy. They hit 233 00:11:22.720 --> 00:11:25.000 and the energy just pushes them right back apart. 234 00:11:25.200 --> 00:11:27.080 It's like trying to get two pieces of velcro to 235 00:11:27.159 --> 00:11:29.759 stick together by firing them out of cannons at each 236 00:11:29.759 --> 00:11:31.919 other across an empty football stadium. 237 00:11:31.960 --> 00:11:34.480 That is Yes, that is a brilliant way to visualize it. 238 00:11:34.600 --> 00:11:37.000 They're just going to ricochet exactly. They need a mediator, 239 00:11:37.120 --> 00:11:39.679 They need a surface to slow them down. And this 240 00:11:39.879 --> 00:11:45.080 is exactly where a microscopic grain of cosmic dust comes in. Okay, Now, 241 00:11:45.080 --> 00:11:48.440 when we say dust in astrophysics, we aren't talking about 242 00:11:48.480 --> 00:11:50.360 the dust bunnies under your couch. 243 00:11:50.159 --> 00:11:51.879 Right, It's not skin cells and lint. 244 00:11:52.200 --> 00:11:56.919 No, No, we are talking about microscopic solid particles of 245 00:11:57.000 --> 00:12:00.600 silicates or carbon, Essentially tiny grains of sand or soot 246 00:12:00.919 --> 00:12:03.279 forged in the dying breaths of older stars. 247 00:12:03.399 --> 00:12:05.639 Okay, so stellar's soot. 248 00:12:05.399 --> 00:12:09.440 Yes, and these solid grains act as a catalytic docking station. 249 00:12:10.240 --> 00:12:12.960 So a single hydrogen atom is flying through the stadium 250 00:12:13.039 --> 00:12:16.480 and instead of hitting another atom, it crashes into this giant, 251 00:12:16.759 --> 00:12:19.039 well relatively speaking, grain of soot yes. 252 00:12:19.600 --> 00:12:22.480 It hits the dust grain and actually sticks to its surface. 253 00:12:22.879 --> 00:12:26.120 The dust grain absorbs the atom's kinetic energy, acting as 254 00:12:26.120 --> 00:12:26.840 a thermal sink. 255 00:12:27.039 --> 00:12:29.120 Oh, so it bleeds off the speed exactly. 256 00:12:29.519 --> 00:12:32.840 Then another hydrogen atom hits that exact same dust grain. 257 00:12:33.159 --> 00:12:35.360 It wanders around the solid surface until it finds the 258 00:12:35.360 --> 00:12:36.200 first atom. 259 00:12:36.000 --> 00:12:37.960 And because they're both slowed down, they bond. 260 00:12:38.279 --> 00:12:41.039 They form molecular hydrogen and then drift off the dust 261 00:12:41.039 --> 00:12:42.240 grain together into space. 262 00:12:42.679 --> 00:12:45.240 So the dust is literally acting like a cosmic anvil 263 00:12:45.279 --> 00:12:47.279 where the fuel for stars is hammered together. 264 00:12:47.519 --> 00:12:51.200 It is the factory floor of star formation. But its 265 00:12:51.279 --> 00:12:52.159 job doesn't end there. 266 00:12:52.200 --> 00:12:52.679 There's more. 267 00:12:52.840 --> 00:12:56.919 Oh yeah. Once that molecular hydrogen forms, it is incredibly fragile. 268 00:12:57.679 --> 00:13:01.440 A galaxy is a violent place, fle with harsh, destructive 269 00:13:01.559 --> 00:13:04.919 ultraviolet radiation pouring out of young hot stars. 270 00:13:05.000 --> 00:13:08.120 So it's basically a hostile environment for these new molecules. 271 00:13:07.639 --> 00:13:11.720 Very hostile if that UV radiation hits our newly formed 272 00:13:11.759 --> 00:13:15.080 molecular hydrogen, it excites the molecule, heats it up, and 273 00:13:15.200 --> 00:13:18.480 literally blasts it apart. Oh, it destroys the fuel supply 274 00:13:18.559 --> 00:13:21.960 for any future stars. It just sterilizes the whole region precisely, 275 00:13:22.240 --> 00:13:24.320 unless the dust steps in again. 276 00:13:24.240 --> 00:13:26.080 The VIP returns exactly. 277 00:13:26.440 --> 00:13:30.519 Cosmic dust is incredibly effective at absorbing ultraviolet light. It 278 00:13:30.639 --> 00:13:34.559 blankets these cold molecular clouds, acting as a physical shield, 279 00:13:34.720 --> 00:13:37.360 so it takes the hit right. The UV radiation hits 280 00:13:37.360 --> 00:13:38.840 the dust on the outside of the cloud, while the 281 00:13:38.879 --> 00:13:42.519 fragile hydrogen gas safely nestled deep inside remains freezing cold. 282 00:13:42.600 --> 00:13:45.480 It's like the universe's own SBF ten thousand sunscreen. 283 00:13:45.559 --> 00:13:46.639 That's a perfect analogy. 284 00:13:46.919 --> 00:13:49.279 The dust takes the hit so the baby stars can 285 00:13:49.360 --> 00:13:53.000 incubate in the dark cold center. And the clean Aber 286 00:13:53.039 --> 00:13:56.320 project was the first time anyone managed to mathematically track 287 00:13:56.679 --> 00:14:00.720 these microscopic sit grains, like their chemical reactions and their 288 00:14:00.720 --> 00:14:03.919 shielding effects across an entire simulated. 289 00:14:03.480 --> 00:14:05.799 Universe on this massive of a scale. 290 00:14:05.960 --> 00:14:07.799 Yes, that is mind blowing. 291 00:14:08.159 --> 00:14:11.600 And what's fascinating here is that introducing dust didn't just 292 00:14:11.639 --> 00:14:15.200 fix the physics. It completely revolutionized the visuals of the 293 00:14:15.200 --> 00:14:18.200 simulation too. How so, Well, think about it. If the 294 00:14:18.320 --> 00:14:22.759 dust is absorbing all that intense ultraviolet starlight, that energy 295 00:14:22.799 --> 00:14:26.440 can't just vanish. The laws of thermodynamics dictate it has 296 00:14:26.519 --> 00:14:27.759 to go somewhere. 297 00:14:27.399 --> 00:14:30.279 Right, energy cannot be created or destroyed. So the dust 298 00:14:30.320 --> 00:14:33.000 heats up and starts glowing in a different wavelength exactly. 299 00:14:33.039 --> 00:14:35.600 It re emits the energy as intrared light. 300 00:14:35.639 --> 00:14:39.799 Exactly, and that completely alters how a simulated galaxy appears 301 00:14:39.799 --> 00:14:41.840 when we look at it. Yeah, for the first time, 302 00:14:42.039 --> 00:14:45.120 researchers were creating virtual galaxies that didn't just have the 303 00:14:45.200 --> 00:14:46.279 right mass or gravity. 304 00:14:46.360 --> 00:14:47.240 They actually looked real. 305 00:14:47.360 --> 00:14:50.720 They actually look like the galaxies we observed through infrared telescopes. 306 00:14:50.840 --> 00:14:53.399 I was actually looking at the visual outputs the Collier 307 00:14:53.440 --> 00:14:56.960 team published. They showed a rendering of a virtual disk 308 00:14:57.159 --> 00:15:00.919 galaxy structurally very similar to our ow Milky Way, right, 309 00:15:01.240 --> 00:15:02.600 from two different perspectives. 310 00:15:02.720 --> 00:15:04.879 Yes, the face on on edge on views right. 311 00:15:05.480 --> 00:15:08.080 When you look at it face on from the top down, 312 00:15:08.440 --> 00:15:12.559 it is this brilliant glowing spiral of starlight. But when 313 00:15:12.559 --> 00:15:15.200 they rotate the simulation and show it to you edge 314 00:15:15.200 --> 00:15:18.519 on from the side, the starlight is suddenly choked out 315 00:15:18.519 --> 00:15:22.399 by this dense, dark, suffocating band of dust slicing right 316 00:15:22.440 --> 00:15:23.679 through the galactic equator. 317 00:15:23.759 --> 00:15:26.159 It's a dramatic visual shift, it really is. 318 00:15:26.200 --> 00:15:29.639 It perfectly mimics the real optical telescope images we get 319 00:15:29.679 --> 00:15:31.720 of edge on galaxies, and that. 320 00:15:31.759 --> 00:15:35.320 Visual fidelity is what makes this a monumental achievement. We 321 00:15:35.399 --> 00:15:39.200 are no longer limited to comparing abstract graphs of theoretical mass. 322 00:15:39.279 --> 00:15:41.480 You know, we're comparing actual pictures exactly. 323 00:15:41.519 --> 00:15:44.320 We can take a real observed image from a space telescope, 324 00:15:44.360 --> 00:15:47.159 put it next to a simulated image from collar, run 325 00:15:47.240 --> 00:15:50.960 both images through the exact same analytical software, and test 326 00:15:51.000 --> 00:15:54.559 if they match. We can measure the exact obscuration of light. 327 00:15:54.679 --> 00:15:58.360 But hold on. The sheer computational brute force required to 328 00:15:58.440 --> 00:16:01.679 do that is I mean, I can't even process it. 329 00:16:01.679 --> 00:16:02.679 It's staggering. 330 00:16:02.960 --> 00:16:06.279 You are asking a computer to track microscopic grains of soot, 331 00:16:06.679 --> 00:16:10.679 calculate the fluid dynamics of freezing gas, apply the gravitational 332 00:16:10.679 --> 00:16:13.919 pull of billions of dark matter halos, and compute the 333 00:16:14.000 --> 00:16:18.159 radiation physics of starlight, all across a volume of space 334 00:16:18.200 --> 00:16:20.559 that spans hundreds of millions of light years. 335 00:16:20.639 --> 00:16:21.399 It's a lot of math. 336 00:16:21.559 --> 00:16:24.759 What kind of machine is actually capable of calculating that matrix? 337 00:16:25.120 --> 00:16:28.600 Well, it requires an architecture that is almost as complex 338 00:16:28.679 --> 00:16:32.519 as the simulation itself. The physical engine running this was 339 00:16:32.600 --> 00:16:36.159 the cost Am Supercomputer, which is hosted by the DRAC 340 00:16:36.360 --> 00:16:39.720 National Facility at Durham University in the UK. Oh but 341 00:16:39.879 --> 00:16:43.159 hardware alone isn't enough. You can have the absolute fastest 342 00:16:43.159 --> 00:16:45.799 supercomputer in the world and if your code is inefficient, 343 00:16:45.919 --> 00:16:48.639 it will still crash when trying to simulate a freezing. 344 00:16:48.279 --> 00:16:51.320 Galaxy, right because of that resolution bottleneck we talked about 345 00:16:51.320 --> 00:16:52.240 earlier Exactly. 346 00:16:52.320 --> 00:16:54.799 The real hero here is the software code they wrote, 347 00:16:54.799 --> 00:16:55.840 which is called swift. 348 00:16:56.320 --> 00:17:00.000 Swift Okay, so how does swift handle the math differently 349 00:17:00.679 --> 00:17:02.720 then the older codes that used to crash when the 350 00:17:02.759 --> 00:17:03.480 gas got cold. 351 00:17:03.919 --> 00:17:07.359 Older cosmological codes usually operated on rigid time steps. 352 00:17:07.559 --> 00:17:08.160 What does that mean? 353 00:17:08.400 --> 00:17:11.519 It means the computer would calculate the gravity, temperature, and 354 00:17:11.599 --> 00:17:15.000 velocity of every single particle in the entire universe, move 355 00:17:15.039 --> 00:17:18.000 the simulation forward by one tick of the cosmic clock, 356 00:17:18.119 --> 00:17:20.440 and then recalculate everything all over again. 357 00:17:20.519 --> 00:17:23.200 Oh wow, So which means the computer has to wait 358 00:17:23.240 --> 00:17:27.519 for the absolute slowest, most complicated calculation to finish before 359 00:17:27.559 --> 00:17:30.799 the whole universe is allowed to move forward one second exactly. 360 00:17:30.880 --> 00:17:33.079 It creates massive processing. 361 00:17:32.599 --> 00:17:34.000 Bottlenecks like a traffic jam. 362 00:17:34.119 --> 00:17:37.680 Very much like a traffic jam, Swift entirely abandoned that model. 363 00:17:37.839 --> 00:17:42.440 It uses task based asynchronous parallelism, meaning it breaks the 364 00:17:42.559 --> 00:17:46.039 universe down into millions of independent tasks. If a cold, 365 00:17:46.119 --> 00:17:48.799 dense cloud of gas in one corner of the simulation 366 00:17:48.920 --> 00:17:51.720 is highly chaotic and needs to be calculated in incredibly 367 00:17:51.759 --> 00:17:54.720 tiny fractions of a second, the processor handles that. 368 00:17:54.680 --> 00:17:56.880 Locally without making the rest of the universe. 369 00:17:56.559 --> 00:17:59.720 Wait right, without forcing a quiet, empty void on the 370 00:17:59.759 --> 00:18:02.279 other side of the virtual universe to slow down and 371 00:18:02.319 --> 00:18:06.559 wait for it. The code dynamically shifts computational power exactly 372 00:18:06.559 --> 00:18:08.920 where it is needed the millisecond it is needed. 373 00:18:08.960 --> 00:18:11.880 So it's essentially a self optimizing universe. It is, And 374 00:18:11.920 --> 00:18:14.720 even with that hyper efficient code, the scale of the 375 00:18:14.720 --> 00:18:18.599 simulation they ran is just staggering. I want you, the listener, 376 00:18:18.599 --> 00:18:21.680 to really internalize these numbers for a second. The largest 377 00:18:21.680 --> 00:18:25.759 simulation in the Calibra suite required seventy two million CPU 378 00:18:25.839 --> 00:18:26.799 hours to complete. 379 00:18:26.920 --> 00:18:31.079 That's seventy two million hours of continuous processing. 380 00:18:30.720 --> 00:18:33.920 Seventy two million. If you somehow manage to install this 381 00:18:34.000 --> 00:18:36.880 swift code onto a top of the line twenty twenty 382 00:18:36.920 --> 00:18:40.160 six gaming laptop and you hit run, your computer would 383 00:18:40.200 --> 00:18:42.519 have to sit there processing it one hundred percent maximum 384 00:18:42.559 --> 00:18:45.240 capacity twenty four hours a day for over eight thousand 385 00:18:45.279 --> 00:18:48.079 years before it finished calculating what calls an A eight did. 386 00:18:47.920 --> 00:18:50.559 And that seventy two million hours produced a volume of 387 00:18:50.559 --> 00:18:54.400 space that is unprecedented in its resolution. They build multiple 388 00:18:54.400 --> 00:18:57.119 simulated universes, but their largest box is a side length 389 00:18:57.160 --> 00:18:59.519 of four hundred coup moving megaparsex. 390 00:19:00.079 --> 00:19:03.920 Now wait, co moving megaparsex. A megaparsec is about three 391 00:19:04.000 --> 00:19:07.640 point twenty six million light years. But what does comoving mean? 392 00:19:08.160 --> 00:19:09.839 Why not just say a billion light years? 393 00:19:09.880 --> 00:19:12.680 Well, because the universe is expanding, right. If you measure 394 00:19:12.720 --> 00:19:16.519 the distance between two galaxies today and then wait a 395 00:19:16.559 --> 00:19:19.799 billion years, the distance between them will be larger, even 396 00:19:19.839 --> 00:19:22.920 if the galaxies themselves haven't actually traveled through space. 397 00:19:23.039 --> 00:19:24.720 The space itself stretched. 398 00:19:24.359 --> 00:19:28.119 Exactly, space is self stretched. Co moving is a coordinate 399 00:19:28.119 --> 00:19:30.279 system that expands along with the universe. 400 00:19:30.440 --> 00:19:31.200 How does that work? 401 00:19:31.359 --> 00:19:34.440