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 Astronomy 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 --> 00:00:23.800 slumber under the night sky. 7 00:00:27.000 --> 00:00:31.359 Have you ever considered the inherent bias in how we 8 00:00:31.440 --> 00:00:32.560 observe the universe. 9 00:00:32.640 --> 00:00:34.560 It's a massive bias, right. 10 00:00:34.359 --> 00:00:36.719 I mean, when you look up at the night sky, 11 00:00:36.799 --> 00:00:40.880 you are basically experiencing a severely filtered version of reality, just. 12 00:00:40.840 --> 00:00:43.039 A tiny fraction of what's out there exactly. 13 00:00:43.079 --> 00:00:46.240 You see the stars, the planets, maybe the Andromeda Galaxy. 14 00:00:46.240 --> 00:00:48.759 If you are in a dark enough location, you're lucky. Yeah, 15 00:00:48.759 --> 00:00:52.600 if you're lucky. But fundamentally, human eyes, and by extension, 16 00:00:52.640 --> 00:00:56.719 the vast majority of our optical telescopes, they are drawn 17 00:00:56.840 --> 00:00:58.840 exclusively to the brightest sources of light. 18 00:00:59.039 --> 00:01:02.200 It's a foundational paradox in astrophysics, really. 19 00:01:02.079 --> 00:01:05.040 It is. We build our entire models of the cosmos 20 00:01:05.040 --> 00:01:08.439 based on these blazing beacons in the dark. You know, 21 00:01:08.480 --> 00:01:11.359 the massive, highly luminous galaxies and quasars. 22 00:01:11.400 --> 00:01:13.959 But doing that, relying just on the bright spots leaves 23 00:01:13.959 --> 00:01:15.480 a monumental blind spot, a. 24 00:01:15.480 --> 00:01:19.120 Huge one, because what about the vast spaces between those objects, 25 00:01:19.120 --> 00:01:22.400 seemingly empty voids, because they are not empty at all, 26 00:01:22.480 --> 00:01:25.359 far from it. So today we are doing a deep 27 00:01:25.400 --> 00:01:29.200 dive into a groundbreaking publication from the Astrophysical Journal. This 28 00:01:29.359 --> 00:01:32.640 was released very recently March three, twenty twenty six. 29 00:01:32.840 --> 00:01:35.159 Based on some incredible. 30 00:01:34.599 --> 00:01:39.120 Data, yes, data from the hobby Everly Telescope Dark Energy Experiment, 31 00:01:39.439 --> 00:01:43.280 which will refer to as HIA dectics. Our mission today 32 00:01:43.319 --> 00:01:46.560 for you, the listener, is to understand how astronomers are 33 00:01:46.599 --> 00:01:48.920 finally looking past those bright stars. 34 00:01:49.040 --> 00:01:51.239 They're finding a hidden ocean, basically. 35 00:01:50.879 --> 00:01:54.519 A vast, hidden sea of light residing right between really galaxies. 36 00:01:55.040 --> 00:01:59.000 And we'll explore what mapping this invisible cosmic web actually 37 00:01:59.079 --> 00:02:02.200 means for our fundamental understanding of the universe. 38 00:02:02.560 --> 00:02:07.000 This research really represents a complete paradigm shift in observational cosmology. 39 00:02:07.120 --> 00:02:10.520 Ow so well, we are looking at a highly critical 40 00:02:10.560 --> 00:02:14.800 epoch in the universe's history, specifically the period between nine 41 00:02:14.840 --> 00:02:16.240 and eleven billion years ago. 42 00:02:16.319 --> 00:02:19.639 Okay, let's unpack this. Why are astronomers so heavily focused 43 00:02:19.680 --> 00:02:22.400 on this particular window. What makes the universe of ten 44 00:02:22.479 --> 00:02:25.719 billion years ago so crucial for understanding the galaxies we 45 00:02:25.759 --> 00:02:26.759 see around us today. 46 00:02:26.960 --> 00:02:31.280 That timeframe, it corresponds to what astrophysicists often call cosmic. 47 00:02:30.879 --> 00:02:33.479 Noon, cosmic noon. I love that term. 48 00:02:33.560 --> 00:02:36.039 It's very descriptive. If you look at the cosmic star 49 00:02:36.280 --> 00:02:37.560 formation rate history, it. 50 00:02:37.560 --> 00:02:39.560 Isn't a flatline, right, not at all. 51 00:02:40.120 --> 00:02:43.159 The universe did not produce stars at a constant rate. 52 00:02:44.000 --> 00:02:47.199 About nine to eleven billion years ago, the universe was 53 00:02:47.280 --> 00:02:50.400 experiencing its absolute peak of star formation. 54 00:02:50.360 --> 00:02:52.199 So it was just churning out stars. 55 00:02:52.319 --> 00:02:56.960 It was an incredibly dynamic, violently active epoch. The galaxies 56 00:02:57.039 --> 00:03:01.120 during this period, they weren't the settled, beautiful spiral galaxies 57 00:03:01.199 --> 00:03:02.240 we see today. 58 00:03:02.039 --> 00:03:03.199 Like the Milky Way, right. 59 00:03:03.199 --> 00:03:05.840 They weren't like the Milky Way. They were chaotic. They 60 00:03:05.840 --> 00:03:10.080 were actively pulling in or accreting massive amounts of primordial 61 00:03:10.120 --> 00:03:13.479 gas from the intergalactic medium, just feeding on this gas 62 00:03:13.520 --> 00:03:17.680 exactly and igniting stars at rates hundreds or sometimes thousands 63 00:03:17.719 --> 00:03:19.919 of times higher than what our galaxy does today. 64 00:03:20.000 --> 00:03:24.719 So it's essentially the peak of cosmic construction. But observing 65 00:03:24.759 --> 00:03:29.280 that construction comes with significant physical limitations, doesn't it very significant? 66 00:03:29.479 --> 00:03:31.479 The primary limitation is just surface. 67 00:03:31.199 --> 00:03:34.080 Brightness because it's so far away, right, Because. 68 00:03:33.879 --> 00:03:36.800 We're looking at objects nine to eleven billion light years away, 69 00:03:37.240 --> 00:03:40.560 the inverse square law dictates that the light reaching us 70 00:03:40.680 --> 00:03:41.719 is exceptionally faint. 71 00:03:41.879 --> 00:03:44.000 Okay, but we can see some galaxies from back then. 72 00:03:44.319 --> 00:03:48.759 We can resolve the most massive hyperluminous galaxies from that era, yes, 73 00:03:49.599 --> 00:03:54.560 because their localized starburst activity is so incredibly intense. But 74 00:03:54.639 --> 00:03:58.240 the smaller stuff, the fainter dwarf galaxies, and more importantly, 75 00:03:58.560 --> 00:04:02.240 the sprawling filaments of fuse hydrogen gas that form the 76 00:04:02.319 --> 00:04:03.360 cosmic web. 77 00:04:03.280 --> 00:04:04.000 The fuel lines. 78 00:04:04.039 --> 00:04:06.439 Basically right the fuel lines, they fall well below the 79 00:04:06.479 --> 00:04:10.039 detection limits of standard optical imaging. The light from that 80 00:04:10.120 --> 00:04:12.960 diffuse gas has spread out over billions of light years. 81 00:04:13.159 --> 00:04:15.800 By the time it hits a telescope mirror on Earth. 82 00:04:15.560 --> 00:04:19.600 It's effectively indistinguishable from the ambient background noise of the 83 00:04:19.720 --> 00:04:22.839 night sky, or even the thermal noise of the camera 84 00:04:22.879 --> 00:04:23.879 instruments themselves. 85 00:04:24.079 --> 00:04:26.319 The time machine aspect of this is something that always 86 00:04:26.360 --> 00:04:29.720 strikes me. We use that term colloquially, but mathematically that 87 00:04:29.879 --> 00:04:33.399 is exactly what telescope data from this era represents. 88 00:04:32.959 --> 00:04:34.759 It's literally a time machine, right. 89 00:04:34.680 --> 00:04:37.480 Because the speed of light is a hard limit. When 90 00:04:37.519 --> 00:04:40.920 we pull data from ha tex that originated ten billion 91 00:04:41.000 --> 00:04:44.399 years ago, we are capturing photons that have been traveling 92 00:04:44.439 --> 00:04:48.319 through the vacuum of space since before our sun even existed. 93 00:04:48.160 --> 00:04:51.279 Long before the Earth is roughly four and a half 94 00:04:51.319 --> 00:04:52.319 billion years old. 95 00:04:52.639 --> 00:04:54.920 So this light had already been traveling for over five 96 00:04:55.000 --> 00:04:57.720 billion years before our planet even coalesced from a cloud 97 00:04:57.759 --> 00:05:00.800 of dust. It's staggering to think about it really is, 98 00:05:00.920 --> 00:05:04.839 and capturing that specific ancient light to reconstruct those early 99 00:05:04.920 --> 00:05:08.519 fuel lines, it requires an entirely different approach than just 100 00:05:08.560 --> 00:05:10.480 taking a long exposure photograph. 101 00:05:10.600 --> 00:05:13.360 Right, you get to take a picture. Taking a standard 102 00:05:13.439 --> 00:05:17.079 optical image of this epoch to find diffuse gas is 103 00:05:17.199 --> 00:05:18.399 largely feudile. 104 00:05:18.319 --> 00:05:20.079 Because the background washes it out. 105 00:05:20.199 --> 00:05:23.920 Exactly The broad band filters used in standard photography or 106 00:05:23.920 --> 00:05:27.720 photometry simply let into much background light. It washes out 107 00:05:27.720 --> 00:05:28.680 any faint structures. 108 00:05:29.120 --> 00:05:30.839 So what's the alternative to. 109 00:05:30.839 --> 00:05:33.959 Map the intergalactic gas ten billion years ago? We have 110 00:05:34.000 --> 00:05:38.199 to abandon images completely. We rely almost entirely on spectroscopy. 111 00:05:38.279 --> 00:05:41.040 We're not looking for the physical shape of a galaxy anymore. 112 00:05:41.199 --> 00:05:44.439 No, we are looking for the highly specific physical signatures 113 00:05:44.600 --> 00:05:49.839 hidden within the electromagnetic radiation itself. We isolate distinct wavelengths 114 00:05:49.839 --> 00:05:52.120 of light that prove the presence of specific matter. 115 00:05:52.439 --> 00:05:55.759 That requires transitioning our focus to the actual language of light, 116 00:05:55.839 --> 00:05:59.319 the spectrum. Now for you listening, breaking light down into 117 00:05:59.399 --> 00:06:02.720 a spectrum reveals the emission and absorption lines of chemical 118 00:06:02.720 --> 00:06:03.399 elements like. 119 00:06:03.399 --> 00:06:05.600 A cosic barcode, right, a barcode. 120 00:06:05.800 --> 00:06:08.399 But I want to zero in on the specific wavelength 121 00:06:08.439 --> 00:06:13.800 that makes this entire HGDX map possible. The Liman alpha emission. 122 00:06:13.439 --> 00:06:15.319 Line the holy grail for this era. 123 00:06:15.839 --> 00:06:19.519 Why is this specific quantum transition the ultimate tool for 124 00:06:19.600 --> 00:06:20.720 looking at the cosmic noon? 125 00:06:21.319 --> 00:06:23.920 Well, the Limon alpha line is the cornerstone of high 126 00:06:23.959 --> 00:06:27.519 redshift observational astronomy. To understand why, we just need a 127 00:06:27.600 --> 00:06:30.040 quick look at the quantum mechanics of the hydrogen atom. 128 00:06:29.879 --> 00:06:32.680 Which is the most abundant element out there by far. 129 00:06:33.240 --> 00:06:35.759 So when a hydrogen atom sits near a region of 130 00:06:35.800 --> 00:06:39.720 intense star formation, like those massive chaotic galaxies we talked about, 131 00:06:39.959 --> 00:06:43.000 it gets bombarded by extreme ultraviolet. 132 00:06:42.480 --> 00:06:44.480 Radiation from the young hot stars. 133 00:06:44.639 --> 00:06:48.439 Yes, specifically massive o and B type stars. This radiation 134 00:06:48.560 --> 00:06:52.240 is so energetic that ionizes the hydrogen. It physically strips 135 00:06:52.240 --> 00:06:54.000 the electron away from the proton, and. 136 00:06:54.000 --> 00:06:56.759 The signal we're looking for is created when that electron 137 00:06:56.800 --> 00:06:57.639 finds its way. 138 00:06:57.439 --> 00:07:02.639 Back right correct, when the pro eventually recombines with an electron, 139 00:07:02.879 --> 00:07:08.399 that electron cascades down through the atom's specific quantized energy. 140 00:07:08.160 --> 00:07:10.600 Level, well stepping down a ladder, exactly. 141 00:07:10.160 --> 00:07:12.360 Like stepping down a ladder, and when it drops from 142 00:07:12.399 --> 00:07:15.319 the first excited state down to the grounds the bottom 143 00:07:15.360 --> 00:07:18.800 run of the ladder, it releases a photon with a 144 00:07:18.920 --> 00:07:21.399 very specific, unchangeable. 145 00:07:20.800 --> 00:07:23.399 Wavelength, and that wavelength is one hundred. 146 00:07:23.079 --> 00:07:25.839 And twenty one point six nanometers. That is the Liman 147 00:07:25.879 --> 00:07:26.439 alpha line. 148 00:07:26.439 --> 00:07:28.360 We one hundred and twenty one point six. 149 00:07:28.199 --> 00:07:31.319 Okay, Because the early universe was absolutely dominated by hydrogen 150 00:07:31.600 --> 00:07:34.920 and the star formation rates were so extreme, these galaxies 151 00:07:34.959 --> 00:07:38.519 act as colossal Liman alpha factories. They pump out an 152 00:07:38.519 --> 00:07:40.800 astonishing number of these specific photons. 153 00:07:40.879 --> 00:07:43.360 But wait, one hundred twenty one point six nanometers is 154 00:07:43.399 --> 00:07:46.319 deep in the ultraviolet spectrum that's invisible to the human eye. 155 00:07:46.399 --> 00:07:48.120 It is, and it doesn't stay at one hundred and 156 00:07:48.160 --> 00:07:51.040 twenty one point six nanimeters either because of the expansion. 157 00:07:50.560 --> 00:07:52.120 Of the universe right the red shift. 158 00:07:52.240 --> 00:07:55.319 As those photons travel through space for ten billion years, 159 00:07:55.680 --> 00:07:58.639 space itself expands, it physically stretches the light. 160 00:07:58.600 --> 00:08:01.319 Waves, so by the time they reach the HGTX instruments 161 00:08:01.360 --> 00:08:05.040 in Texas, that ultraviolet light has been red shifted straight 162 00:08:05.079 --> 00:08:07.160 into the visible optical bands right. 163 00:08:07.040 --> 00:08:09.120 Around three hundred and fifty to five hundred and fifteen 164 00:08:09.160 --> 00:08:13.240 nanimeters greenish blue light exactly. The source material notes this 165 00:08:13.279 --> 00:08:16.839 shows up as a dramatic peak in the data. If 166 00:08:16.879 --> 00:08:20.720 you picture the spectrographic feed, you have this relatively flat 167 00:08:20.800 --> 00:08:26.279 continuum of background emission and then boom, a violent, unmistakable. 168 00:08:25.600 --> 00:08:28.519 Spike at that specific red shifted wavelength. 169 00:08:28.759 --> 00:08:32.440 It's the undeniable fingerprint of excited hydrogen at that exact 170 00:08:32.440 --> 00:08:33.440 distance in the universe. 171 00:08:33.799 --> 00:08:37.000 We call that dramatic peak a Liman alpha emitter or 172 00:08:37.039 --> 00:08:40.279 an lae An lee. Finding those massive spikes is the 173 00:08:40.320 --> 00:08:44.159 traditional methodology for locating high red shift galaxies. If you 174 00:08:44.200 --> 00:08:48.320 see that peak, you have definitively located a bright active 175 00:08:48.320 --> 00:08:50.000 galaxy from that specific epoch. 176 00:08:50.120 --> 00:08:51.759 But the galaxies aren't the whole story. 177 00:08:51.919 --> 00:08:55.240 No, theoretical models of the cosmic web have always suggested 178 00:08:55.240 --> 00:08:57.720 that lemon alpha mission shouldn't just be restricted to the 179 00:08:57.759 --> 00:08:58.600 massive galaxies. 180 00:08:58.600 --> 00:09:00.360 The gas in between them should be glowing too. 181 00:09:00.600 --> 00:09:04.000 The vast filaments of intergalactic gas drifting between the galaxies 182 00:09:04.120 --> 00:09:07.480 should also be emitting these photons, either through recombination from 183 00:09:07.519 --> 00:09:10.279 the background radiation or just from gas falling into dark 184 00:09:10.320 --> 00:09:11.600 matter halos and heating up. 185 00:09:11.840 --> 00:09:15.399 But the emission from those gas filaments that would be 186 00:09:15.960 --> 00:09:19.840 orders of magnitude weaker than the galaxies themselves. 187 00:09:19.360 --> 00:09:21.600 Vastly weaker. I mean, the peaks from the galaxies are 188 00:09:21.679 --> 00:09:24.600 highly localized, they're relatively easy to extract from the data. 189 00:09:24.759 --> 00:09:25.720 They stand out. 190 00:09:25.639 --> 00:09:28.600 They do, but the emission from the intergalactic gas is 191 00:09:28.679 --> 00:09:34.200 exceptionally diffuse. It's just a subtle, incredibly faint glow spread 192 00:09:34.240 --> 00:09:36.440 across massive cosmic volumes. 193 00:09:36.039 --> 00:09:38.480 And for decades, finding that faint signal was thought to 194 00:09:38.480 --> 00:09:39.840 be impossible. 195 00:09:39.360 --> 00:09:42.879 Virtually impossible on a large scale. The instrumental noise, the 196 00:09:42.919 --> 00:09:46.320 foreground light from our own solar system, it all drowns it. 197 00:09:46.320 --> 00:09:48.879 Out, which brings us to the sheer scale of the 198 00:09:48.919 --> 00:09:52.639 instrument required to even attempt this. You cannot just point 199 00:09:52.639 --> 00:09:55.519 a standard observatory telescope at the sky and hope to 200 00:09:55.519 --> 00:09:56.360 map this stuff. 201 00:09:56.440 --> 00:09:57.480 No, you need a behemoth. 202 00:09:57.600 --> 00:10:00.879 You need the hobby Eberly Telescope Dark e Energy Experiment 203 00:10:00.919 --> 00:10:04.000 at the McDonald Observatory in West Texas. Let's discuss the 204 00:10:04.080 --> 00:10:07.240 volume of this survey, because the engineering reality of headdecks 205 00:10:07.440 --> 00:10:09.080 is staggering the hobby. 206 00:10:09.080 --> 00:10:13.320 Every telescope. The HET is a totally unique piece of engineering. 207 00:10:13.600 --> 00:10:16.440 Most big telescopes move on dual axis, right, They tilt 208 00:10:16.519 --> 00:10:19.039 up and down and spin around to track the sky. 209 00:10:19.399 --> 00:10:23.159 Right. But the HT has a fixed elevation angle. It 210 00:10:23.240 --> 00:10:25.159 sits permanently at fifty five degrees. 211 00:10:25.320 --> 00:10:25.639 Okay. 212 00:10:26.000 --> 00:10:29.720 It simply rotates in asimuth around in a circle while 213 00:10:29.919 --> 00:10:33.639 a highly complex tracker moves across the focal plane at 214 00:10:33.639 --> 00:10:37.120 the top to follow the astronomical targets as the Earth turns. 215 00:10:37.279 --> 00:10:38.559 That's incredibly clever. 216 00:10:38.799 --> 00:10:42.000 It's very efficient, and for the dark energy experiment, the 217 00:10:42.039 --> 00:10:45.159 telescope was upgraded with a massive array of spectrographs. 218 00:10:45.200 --> 00:10:49.600 They call it VIRUS, the Visible Integral Field Replicable Unit spectrograph. 219 00:10:49.639 --> 00:10:52.639 That's the one, and VIRUS isn't just one single instrument, No. 220 00:10:52.600 --> 00:10:57.159 It's a massive replication strategy. Instead of building one giant spectrograph. 221 00:10:57.279 --> 00:10:59.879 They build dozens of identical units and fed them with 222 00:11:00.159 --> 00:11:02.039 thousands of optical fibers. 223 00:11:01.679 --> 00:11:04.879 Over thirty thousand optical fiber thirty thousand. This allows eighth 224 00:11:05.080 --> 00:11:09.200 decics to perform integral field spectroscopy on an industrial scale. 225 00:11:09.480 --> 00:11:12.360 They're taking spectra of thousands of discrete points on the 226 00:11:12.360 --> 00:11:14.679 sky simultaneously. 227 00:11:13.919 --> 00:11:17.320 And their primary mission, as the name implies, is mapping 228 00:11:17.399 --> 00:11:20.399 the expansion history of the universe to constrain dark energy. 229 00:11:20.720 --> 00:11:23.480 Yes, their stated goal was to chart the three D 230 00:11:23.639 --> 00:11:29.159 positions of over one million bright Liman alpha emitting galaxies. 231 00:11:29.480 --> 00:11:32.360 To get that catalog of a million galaxies, they had 232 00:11:32.360 --> 00:11:35.399 to cover a massive area of the sky, an area 233 00:11:35.480 --> 00:11:37.399 measuring over two thousand full moons. 234 00:11:37.440 --> 00:11:39.440 It's a huge swathe of the celestial sphere. 235 00:11:39.519 --> 00:11:42.399 Let's help you visualize that the angular diameter of the 236 00:11:42.399 --> 00:11:45.320 full moon is about half a degree, so taking up 237 00:11:45.360 --> 00:11:49.679 two thousand full moons, that is a massive, sweeping expanse 238 00:11:49.759 --> 00:11:50.120 of space. 239 00:11:50.159 --> 00:11:53.120 They're just blindly pointing these thirty thousand fibers at the sky, 240 00:11:53.279 --> 00:11:57.840 pulling in light, separating into wavelengths, and generating an unbelievable 241 00:11:57.919 --> 00:11:59.879 six hundred million individual spectra. 242 00:12:00.080 --> 00:12:03.000 He Here's where it gets really interesting. Carl Gipart, the 243 00:12:03.039 --> 00:12:06.480 principal investigator for at DENEX, revealed a metric about this 244 00:12:06.559 --> 00:12:10.039 data collection that fundamentally alters how we view these surveys. 245 00:12:10.159 --> 00:12:10.840 It really does. 246 00:12:11.000 --> 00:12:15.320 Despite gathering six hundred million spectra that primary dark energy mission, 247 00:12:15.360 --> 00:12:18.639 the effort to map the one million bright galaxies, it 248 00:12:18.720 --> 00:12:21.399 only utilizes roughly five percent of the collected data. 249 00:12:21.440 --> 00:12:24.240 Five percent. That is the crucial pivot point of this 250 00:12:24.440 --> 00:12:27.200 entire deep dive. It's what the primary pipeline for h 251 00:12:27.279 --> 00:12:31.759 DEX is designed for point source extraction. It scans all 252 00:12:32.120 --> 00:12:37.279 six hundred million spectra looking for high signal to noise ratio. 253 00:12:37.000 --> 00:12:39.440 Peaks the bright galaxies. 254 00:12:38.960 --> 00:12:43.200 Exactly once it identifies in catalogs at Galaxy. The rest 255 00:12:43.240 --> 00:12:45.960 of the data surrounding that peak, which makes up ninety 256 00:12:46.000 --> 00:12:49.519 five percent of the total data set, is mathematically categorized 257 00:12:49.559 --> 00:12:50.440 as background noise. 258 00:12:50.720 --> 00:12:52.720 They threw out ninety five percent of the data, I 259 00:12:52.720 --> 00:12:54.840 mean not literally deleted it from the hard drives, but 260 00:12:55.240 --> 00:12:58.480 scientifically it was sideline was ignored. You build this incredibly 261 00:12:58.480 --> 00:13:02.279 complex array of thirty five optical fibers, you survey two 262 00:13:02.320 --> 00:13:05.399 thousand full moons of sky and ninety five percent of 263 00:13:05.440 --> 00:13:08.720 the photons you catch are deemed irrelevant just because they 264 00:13:08.720 --> 00:13:11.000 don't cross a specific brightness threshold. 265 00:13:11.159 --> 00:13:16.320 Well, from a traditional survey perspective, that is standard operating procedure. Really, yes, 266 00:13:17.080 --> 00:13:20.519 if your objective is a highly pure catalog of discrete 267 00:13:20.679 --> 00:13:24.840 individual objects, anything that cannot be confidently resolved as an 268 00:13:24.840 --> 00:13:26.399 object is an impediment. 269 00:13:26.480 --> 00:13:28.039 It's just getting in the way, exactly. 270 00:13:28.200 --> 00:13:31.039 It's four ground light, it's atmospheric air glow, it's thermal 271 00:13:31.080 --> 00:13:32.600 noise in the CCD detectors. 272 00:13:32.799 --> 00:13:35.759 But Masha Lujah Niemeyer and the team behind this new 273 00:13:35.799 --> 00:13:41.240 publication they recognized a profound philosophical flaw in that approach. 274 00:13:41.600 --> 00:13:44.639 They realized that ninety five percent is not empty noise, 275 00:13:44.960 --> 00:13:46.240 no among. 276 00:13:45.919 --> 00:13:49.519 The instrument artifacts, and the air glow is the literal 277 00:13:49.600 --> 00:13:51.559 sea of light from the cosmic web. 278 00:13:51.799 --> 00:13:54.720 It contains the aggregate liman alpha emissions of all the 279 00:13:54.799 --> 00:13:58.200 dwarf galaxies that were too faint to trigger the detection algorithms, 280 00:13:58.799 --> 00:14:01.519 plus the glowing films of intergalactic gas. 281 00:14:02.000 --> 00:14:04.559 It's the difference between mapping the peaks of a mountain 282 00:14:04.639 --> 00:14:07.879 range and mapping the entire tectonic plate underneath it. 283 00:14:07.919 --> 00:14:08.919 That's a great way to put it. 284 00:14:08.960 --> 00:14:12.080 The bright galaxies are just the most luminous nodes of 285 00:14:12.120 --> 00:14:16.519 a much larger, interconnected structure. But the challenge wasn't getting 286 00:14:16.600 --> 00:14:20.120 the data hadx already banked half a petabite of it. 287 00:14:20.200 --> 00:14:24.000 The challenge was statistical, highly statistical. How do you extract 288 00:14:24.039 --> 00:14:27.159 an incredibly faint, highly diffused signal from a data set 289 00:14:27.200 --> 00:14:30.080 where the noise is orders of magnitude louder than the signal. 290 00:14:30.200 --> 00:14:33.320 You have to completely abandon the concept of object resolution. 291 00:14:33.080 --> 00:14:35.639 Stop looking for individual things exactly. 292 00:14:35.799 --> 00:14:38.000 You can no longer ask the data pipeline to find 293 00:14:38.000 --> 00:14:42.039 a specific galaxy. This requires transitioning to a technique known 294 00:14:42.080 --> 00:14:43.919 as line intensity mapping. 295 00:14:44.360 --> 00:14:46.759 Line intensity mapping or LIMB. 296 00:14:46.600 --> 00:14:50.320 WIM fundamentally redefines the objective of the survey. Instead of 297 00:14:50.360 --> 00:14:54.320 searching for spatial coordinates of bright peaks, LIMB measures the 298 00:14:54.360 --> 00:14:58.399 integrated surface brightness of a specific spectral line across large 299 00:14:58.399 --> 00:14:59.360 cosmic volumes. 300 00:14:59.600 --> 00:15:02.399 To make that concrete for you, listening, Shuleian Mignot's, a 301 00:15:02.480 --> 00:15:05.440 co author on the paper, offered an excellent analogy regarding 302 00:15:05.440 --> 00:15:07.559 how we view the spatial distribution. 303 00:15:07.159 --> 00:15:08.720 Of light, the airplane analogy. 304 00:15:08.879 --> 00:15:12.360 Yes, he compared the traditional cataloging method to flying in 305 00:15:12.399 --> 00:15:14.840 an airplane at night and trying to map a country's 306 00:15:14.879 --> 00:15:17.559 population by only looking at the brightest city centers. 307 00:15:17.600 --> 00:15:20.879 It's highly illustrative. If your optical censor on the aircraft 308 00:15:20.919 --> 00:15:24.600 is calibrated to only register the intense light output of 309 00:15:24.759 --> 00:15:28.039 major metropolitan areas like New York, Chicago. 310 00:15:27.639 --> 00:15:31.159 Los Angeles, then your resulting map implies a binary distribution. 311 00:15:31.360 --> 00:15:34.480 Exactly, it implies there are intense points of existence surrounded 312 00:15:34.480 --> 00:15:36.960 by total empty voids. 313 00:15:36.679 --> 00:15:39.919 But we know demographically that the population is continuous. 314 00:15:40.200 --> 00:15:44.919 There are sprawling suburbs, rural highway corridors, and small towns 315 00:15:44.960 --> 00:15:46.720 connecting those major hubs. 316 00:15:46.559 --> 00:15:51.480 And the traditional point source extraction of ATX was mapping 317 00:15:51.519 --> 00:15:54.519 the cosmic cities, but it was completely blind to the 318 00:15:54.559 --> 00:15:58.159 cosmic suburbs and the interstate highways of gas connecting. 319 00:15:57.720 --> 00:16:01.039 Them, because the light from the suburbs is isn't concentrated 320 00:16:01.159 --> 00:16:02.399 enough to trigger the sensor. 321 00:16:02.840 --> 00:16:06.879 So to map the entire landscape, Muno's suggests keeping the 322 00:16:06.919 --> 00:16:09.759 airplane at the exact same altitude, looking at the exact 323 00:16:09.799 --> 00:16:13.200 same landscape, but changing the optical properties of the sensor. 324 00:16:13.320 --> 00:16:15.679 You look through a deliberately smudged window. 325 00:16:15.519 --> 00:16:16.559 A smudged window. 326 00:16:16.799 --> 00:16:20.679 The smudged window represents the spatial and spectral smoothing inherent 327 00:16:20.759 --> 00:16:24.320 in line intensity mapping. When you apply a smoothing kernel 328 00:16:24.320 --> 00:16:27.840 to the data, you intentionally degrade the resolution. You make 329 00:16:27.879 --> 00:16:30.120 it burry, You make it very blurry. You can no 330 00:16:30.200 --> 00:16:33.279 longer distinguish the sharp boundaries of the major cities. The 331 00:16:33.360 --> 00:16:35.120 points of light blur and expand. 332 00:16:35.159 --> 00:16:38.440 But there's a mathematical advantage to that, right, a critical one. 333 00:16:38.919 --> 00:16:42.559 While resolution is lost, the total photon count is conserved, 334 00:16:42.639 --> 00:16:46.919 it's aggregated. The faint sub threshold light from the cosmic 335 00:16:46.960 --> 00:16:51.279 suburbs is integrated into larger volumetric pixels, which we call. 336 00:16:51.159 --> 00:16:53.879 Voxels vouxeles three D pixels right. 337 00:16:53.960 --> 00:16:57.240 And by integrating over larger volumes, the faint signal of 338 00:16:57.279 --> 00:17:02.039 the diffuse limon al for emission constructively interferes. It naturally 339 00:17:02.159 --> 00:17:06.200 rises above the threshold of the random, uncorrelated instrumental noise. 340 00:17:06.400 --> 00:17:10.319 It's a brilliant statistical maneuver. You sacrifice the ability to 341 00:17:10.359 --> 00:17:14.079 say there is a distinct dwarf galaxy exactly at coordinate X, 342 00:17:14.559 --> 00:17:17.039 but you gain the ability to say this entire region 343 00:17:17.079 --> 00:17:21.000 of space is radiating a faint Liman alpha glow. The 344 00:17:21.039 --> 00:17:24.880 blurry picture actually contains more cosmological information about the distribution 345 00:17:24.920 --> 00:17:27.680 of matter than the sharp, highly filtered picture did. 346 00:17:27.880 --> 00:17:30.799 It's a more complete truth. Now the source material does 347 00:17:30.799 --> 00:17:33.440 clarify that line intensity mapping itself is not a newly 348 00:17:33.480 --> 00:17:34.319 invented concept. 349 00:17:34.480 --> 00:17:36.000 Right radio astronomers have used it. 350 00:17:36.000 --> 00:17:37.799 They've used it for years to map the twenty one 351 00:17:37.880 --> 00:17:41.519 centimeter line of neutral hydrogen. But applying this technique to 352 00:17:41.599 --> 00:17:44.599 the ultraviolet Liman alpha e missions in the optical band. 353 00:17:44.680 --> 00:17:46.839 Over a survey area of two. 354 00:17:46.680 --> 00:17:50.359 Thousand full moons, that is entirely unprecedented. 355 00:17:49.720 --> 00:17:53.759 And the application at this scale introduces formidable computational complexities. 356 00:17:53.839 --> 00:17:57.680 Massive complexities in twenty one centimeter mapping. The foregrounds are intense, 357 00:17:57.960 --> 00:18:00.799 but the spectral line itself is relative straightforward. 358 00:18:00.839 --> 00:18:02.799 But Lyman alpha is messy. 359 00:18:03.039 --> 00:18:07.000 It's a resonant line. The focons scatter repeatedly off neutral 360 00:18:07.079 --> 00:18:10.519 hydrogen atoms before they ever escape the galactic halo. It 361 00:18:10.559 --> 00:18:13.839 makes the radiative transfer extremely complex to model. 362 00:18:14.160 --> 00:18:18.599 Plus the data set itself is gargantuan. To apply line 363 00:18:18.599 --> 00:18:22.079 intensity mapping to the discarded ninety five percent of the 364 00:18:22.240 --> 00:18:25.839 HX data required processing roughly half a petabyte of raw 365 00:18:25.880 --> 00:18:29.160 spectroscopic files. Half a petabyte, let's give that some scale. 366 00:18:29.519 --> 00:18:31.880 If you consider a high definition movie to be roughly 367 00:18:31.960 --> 00:18:35.240 five gigabytes, half a petabyte is equivalent to one hundred 368 00:18:35.359 --> 00:18:37.000 thousand high definition movies. 369 00:18:37.079 --> 00:18:38.480 You're not doing that on all laptop. 370 00:18:38.680 --> 00:18:41.480 No. Processing that volume of data isn't something you do 371 00:18:41.559 --> 00:18:44.400 on a workstation in a university lab. The team had 372 00:18:44.400 --> 00:18:48.079 to rely on the Texas Advanced Computing Center or TACC. 373 00:18:48.319 --> 00:18:52.359 They were utilizing supercomputers like Fronterra and Stampede. 374 00:18:51.759 --> 00:18:55.839 Running completely customed pipelines. They had to mathematically strip away 375 00:18:55.839 --> 00:18:59.440 the atmospheric emission lines, the foreground zodiacal light from our 376 00:18:59.440 --> 00:19:02.440 solar system, the galactic cirrus from the Milky Way, and 377 00:19:02.599 --> 00:19:04.880 all the instrumental artifacts. 378 00:19:04.440 --> 00:19:06.559 From six hundred million spectra, all. 379 00:19:06.400 --> 00:19:11.640 Without accidentally erasing the incredibly fragile ultra faintlyman alpha signal 380 00:19:11.839 --> 00:19:12.759 hidden underneath. 381 00:19:12.920 --> 00:19:16.119 The data reduction pipeline is an engineering marvel in itself, 382 00:19:16.839 --> 00:19:21.400 but raw computational power is meaningless without a rigorous physical 383 00:19:21.440 --> 00:19:24.279 framework to guide it. If we connect this to the 384 00:19:24.319 --> 00:19:27.920 bigger picture, the actual methodology used to reveal the cosmic 385 00:19:27.960 --> 00:19:30.599 web relies on a foundational property of. 386 00:19:30.559 --> 00:19:32.640 Cosmology, gravitational clustering. 387 00:19:32.960 --> 00:19:36.680 Exactly, the universe is not a uniform soup of matter. 388 00:19:37.039 --> 00:19:40.119 It is heavily structured by the gravitational potential wells of 389 00:19:40.200 --> 00:19:41.559 dark matter halos. 390 00:19:41.279 --> 00:19:45.119 And this is where Chro Komatsu's signpost technique comes into play. 391 00:19:45.400 --> 00:19:48.559 Kamatsu is a highly respected cosmologist at the Max Planck 392 00:19:48.680 --> 00:19:50.119 Institute for Astrophysics. 393 00:19:50.279 --> 00:19:53.000 His contribution here is brilliant. It bridges the gap between 394 00:19:53.000 --> 00:19:55.799 the five percent catalog and the ninety five percent noise. 395 00:19:56.160 --> 00:20:00.599 He utilizes the concept of cross correlation. Since gravity dictates 396 00:20:00.640 --> 00:20:04.319 that matter will pool inside these dark matter halos, we 397 00:20:04.400 --> 00:20:07.799 know that massive bright galaxies do not exist in isolation. 398 00:20:08.000 --> 00:20:11.920 They seated the densest nodes of the cosmic web exactly. 399 00:20:11.680 --> 00:20:16.000 The one million bright Lineman alpha emitters the five percent 400 00:20:16.039 --> 00:20:20.640 of the data already cataloged by HDX. They aren't discarded 401 00:20:20.680 --> 00:20:21.400