1 00:00:08,410 --> 00:00:11,010 Okay. So that's 3:00 and time to begin. 2 00:00:11,020 --> 00:00:19,360 So it is a pleasure to introduce to you today Speaker Professor Gabriella Gonzalez of Louisiana State University. 3 00:00:20,350 --> 00:00:31,360 Gaby has been a member of the Ligo's scientific collaboration since 1997, and in 2011 she was elected as its spokesperson. 4 00:00:32,500 --> 00:00:38,230 Her group is involved in the characterisation of the noise of the like her detectors. 5 00:00:39,410 --> 00:00:44,600 But the calibration of the detectors themselves. And with the analysis of the data. 6 00:00:46,170 --> 00:00:55,160 On September the 14th last year. The two, like gravitational detectors in Washington State and Louisiana, 7 00:00:55,260 --> 00:01:01,110 later said nearly simultaneous signals that time frequency properties consistent 8 00:01:01,560 --> 00:01:06,840 with gravitational wave ignition by the merger of two massive compact objects. 9 00:01:07,860 --> 00:01:13,620 Further analysis of those signals by Liger and Virgo revealed that the gravitational 10 00:01:13,620 --> 00:01:19,080 waves detected by liger came from the merger of a binary black hole system. 11 00:01:20,250 --> 00:01:24,390 This observation, followed by another in December of the same year, 12 00:01:25,020 --> 00:01:30,749 is one of the most remarkable and significant discoveries in all of the physical 13 00:01:30,750 --> 00:01:37,080 sciences in the 21st century has captured the imagination of citizens around the world, 14 00:01:37,740 --> 00:01:42,600 and it marks the beginning of the new field of gravitational wave astronomy. 15 00:01:43,590 --> 00:01:48,149 Asked about how she would like to be introduced today and she told me just saying, 16 00:01:48,150 --> 00:01:56,970 a professor from Louisiana State who will present today a talk to you on behalf of more than 1000 colleagues and the like, a collaboration. 17 00:01:57,000 --> 00:02:08,850 Please welcome Gabby Gonzalez. Thank you. 18 00:02:09,210 --> 00:02:12,510 Can you hear me? Yes. Okay. Yes. 19 00:02:12,510 --> 00:02:17,700 It's a pleasure to be here. Thank you for inviting me. It's my first time in Oxford and it's lovely. 20 00:02:19,470 --> 00:02:27,180 And yes, I want to tell you about this beautiful observation we had, which is the first of many. 21 00:02:27,510 --> 00:02:36,809 And I hope since the 21st century is still early, that it will be also the first of many significant discoveries we already had to the Higgs. 22 00:02:36,810 --> 00:02:39,180 And these I'm looking for many more. 23 00:02:40,740 --> 00:02:49,680 Let me remind you, maybe not all of you have studied general relativity or have learned about the basics of gravitational waves. 24 00:02:49,680 --> 00:02:58,320 And we just give you a very quick tour. These are Einstein's equations, and they look very simple, but they are not. 25 00:02:59,130 --> 00:03:06,660 On the left hand side, we have Einstein's stance, sort of, which is constructed from a spacetime metric. 26 00:03:07,110 --> 00:03:15,000 So a spacetime metric is a sensor that tells us how to measure distances on time. 27 00:03:15,020 --> 00:03:19,020 So I imagine he disagreed on rulers and clocks. 28 00:03:19,380 --> 00:03:26,610 That's a spacetime metric. And what we have here is all sort of derivatives that are non-linear, 29 00:03:26,610 --> 00:03:32,640 like combined multiplications of derivatives with respect to all the spacetime variables. 30 00:03:33,030 --> 00:03:39,210 So that is the mass that you have on the left hand side having to do with spacetime. 31 00:03:39,930 --> 00:03:43,590 The right hand side is the stress energy tensor. 32 00:03:43,800 --> 00:03:50,460 That's what's telling you where the mass is and what the masses are doing, how are they moving? 33 00:03:51,120 --> 00:03:58,350 And what this equation is telling you is that the metric is telling the masses how to move. 34 00:03:58,800 --> 00:04:07,500 So that's why this is a theory of gravity. It tells you that the sun and the earth are this system where the earth goes around the sun, 35 00:04:07,800 --> 00:04:11,370 not because there is a force of gravity that attracts them, 36 00:04:11,670 --> 00:04:20,370 but because the sun curves the spacetime in such a way that for the Earth it's easier to go in a simple than to go straight. 37 00:04:21,030 --> 00:04:26,310 So that is how the curvature of spacetime determines how that matters most. 38 00:04:26,970 --> 00:04:33,750 But because the matter moves, then the matter is also telling the spacetime how to cope. 39 00:04:34,050 --> 00:04:41,190 So the two things go together. And that is the beauty and the difficulty of Einstein's equations. 40 00:04:41,340 --> 00:04:48,360 They are actually quite complicated. But even Einstein himself, after passing this in 1915. 41 00:04:49,170 --> 00:04:59,500 So that if you if you said the right hand side equal to zero, and then you assume that the space time metric isn't just the regular space, 42 00:04:59,640 --> 00:05:06,240 flat space and metric Minkowski metric that we use all the time to solve electromagnetic equations. 43 00:05:06,240 --> 00:05:12,480 Then in other words, you assume it's a flat metric plus a small perturbation, 44 00:05:13,320 --> 00:05:22,170 and small in this case means that you are measuring things that are don't have dimensions 45 00:05:22,170 --> 00:05:26,670 in here because you are combining space and time or they have the same dimensions. 46 00:05:27,000 --> 00:05:33,990 So this this perturbation is measured as changes in distance, over distance. 47 00:05:35,040 --> 00:05:37,680 You can also transform that into changes in time. 48 00:05:38,410 --> 00:05:43,650 But for our purposes, we're going to interpret that as changes in distance, fractional changes in distance. 49 00:05:44,190 --> 00:05:54,300 So if you put in if you make this equal to zero and you put you put this approximation in here and you keep only terms linear in h, 50 00:05:54,540 --> 00:06:05,630 what you get is a wave equation. So what that tells you is that spacetime needs Einstein's equations a neat way solutions. 51 00:06:05,650 --> 00:06:15,040 And those are gravitational waves. And that he got in 1916 with some math errors in there, actually math and physics, everything down. 52 00:06:15,050 --> 00:06:23,700 But those were sorted out later. Those waves would be transverse and they would have two polarisations. 53 00:06:24,170 --> 00:06:33,260 The idea is that if you have a ring of death particles that are freely falling and that's not such a thought experiment. 54 00:06:33,500 --> 00:06:44,239 I like to think about these people who like jumping off planes in parachutes before they open the parachutes and they are all arranged in a circle. 55 00:06:44,240 --> 00:06:50,060 Sometimes they hold hands, but if they don't hold hands, what you have is particles in freefall. 56 00:06:50,900 --> 00:06:57,560 So imagine that you have such a circle of this particles and there's a gravitational wave propagating in this direction. 57 00:06:57,950 --> 00:07:02,440 Then distances that are going to chase particles are not necessarily going to move. 58 00:07:02,450 --> 00:07:07,370 That depends on your coordinate systems, but the distances are going to change that. 59 00:07:07,370 --> 00:07:14,089 That would seem to be equal distances here. Now it's longer here and shorter there, and then they are the same again. 60 00:07:14,090 --> 00:07:22,010 And then this one is longer and this one is shorter. So does the distortion of distances due to the gravitational wave. 61 00:07:22,490 --> 00:07:26,000 And there are two polarisation. These are quadrupole gravitational waves. 62 00:07:26,000 --> 00:07:32,900 So in this polarisation, these ellipses form a plus sign in this when you have the ellipses forming a cross. 63 00:07:32,910 --> 00:07:36,979 So those are the two polarisations in the theory of relativity. 64 00:07:36,980 --> 00:07:43,160 Alternative theories have, of course, other assumptions. Now, how are these waves produced? 65 00:07:43,850 --> 00:07:53,179 Well, in 1980 he had another paper in which he used similar approximations, but then he solved the equations, 66 00:07:53,180 --> 00:07:58,370 putting some matter in here for the wave propagation, for the wave formation. 67 00:07:58,370 --> 00:08:08,120 And and what he saw is what we call now the quadrupole formula that this perturbation of distance is going to be produced by your mass quadrupole. 68 00:08:08,270 --> 00:08:15,110 So you need the non vertical distribution you need to have it must quadrupole, you need to have time derivatives. 69 00:08:15,350 --> 00:08:23,959 In fact, we want three time derivatives because to just give you a constant H and what you want is a periodic H or two time dependent H, 70 00:08:23,960 --> 00:08:29,540 so you want three derivatives of come here and there are very unfortunate constants in here. 71 00:08:30,140 --> 00:08:34,190 This is Newton's constant because this is a theory of gravity. This is a speed of light, 72 00:08:34,190 --> 00:08:39,440 because this is a relativistic theory of gravity and you have distance to the source 73 00:08:39,440 --> 00:08:44,480 non distance squared because this is the field effect we had calculating the field, 74 00:08:45,410 --> 00:08:52,549 the field and not the not the energy. And this is unfortunate because if you think about experiment, 75 00:08:52,550 --> 00:09:00,800 so now if you think about I do not do very heavy masses in a ring rotating at some kilohertz frequencies. 76 00:09:01,210 --> 00:09:05,120 The Hertz experiment that this aid is just negligible. 77 00:09:05,660 --> 00:09:14,750 And Einstein had in that paper this was a tournament which they were calculating observational effects of this theory to to to prove it, 78 00:09:14,750 --> 00:09:16,940 to see whether there were experiments that could prove it. 79 00:09:17,480 --> 00:09:22,160 He said this was never going to be measured, and that's what we thought for a very, very long time. 80 00:09:23,450 --> 00:09:29,209 Now we know better. We know that this could be you have large quadrupole. 81 00:09:29,210 --> 00:09:36,320 So if you have on time dependent if you have for example to masses moving around the centre of mass like the earth and the sun. 82 00:09:36,650 --> 00:09:42,590 But if you have compact objects like neutron stars, then not only they are going to rotate, 83 00:09:42,950 --> 00:09:46,489 but they're going to be losing energy to these gravitational waves. 84 00:09:46,490 --> 00:09:53,390 So they're going to get closer and closer than the if they are compact, they can get very close and then move very fast. 85 00:09:53,720 --> 00:09:57,940 So that's the way to get a large quadrupole and we know that exists. 86 00:09:57,950 --> 00:10:04,120 You probably remember that in the seventies, the first binary system of neutron stars was discovered. 87 00:10:04,120 --> 00:10:09,500 That was a wholesaler system and it was in our galaxy not that far away. 88 00:10:09,830 --> 00:10:14,870 And they had the stars were quite close. This was an eight hour period. 89 00:10:15,200 --> 00:10:22,819 Imagine you have each star, the mass of the sun they had rotating with an eight hour period. 90 00:10:22,820 --> 00:10:26,149 The earth rotates with a one year period around the sun. 91 00:10:26,150 --> 00:10:31,690 And here you have these two copied stars a lot closer to it. 92 00:10:32,120 --> 00:10:36,140 It's actually, to me, mind boggling to think about this system. 93 00:10:36,350 --> 00:10:38,870 It was the first binary system discovered since then. 94 00:10:38,870 --> 00:10:48,259 There's been quite a few more, but because they were close, if you want, then the laws of energy could be seen. 95 00:10:48,260 --> 00:10:56,270 And that was the first proof that gravitational waves exist because they take energy of the system, just like the theory predicts. 96 00:10:57,140 --> 00:11:02,570 So that was absurd. And then we knew that those gravitational waves would have this form. 97 00:11:02,570 --> 00:11:08,090 As the stars get closer and closer, then the period gets shorter, the frequency gets higher. 98 00:11:08,450 --> 00:11:17,750 You can use an approximation. And then this dimensionless quantity is more or less proportional to the product of the radio of the stars. 99 00:11:17,930 --> 00:11:23,239 It's about ten kilometres for neutron stars divided by the distance between the stars. 100 00:11:23,240 --> 00:11:28,160 So the closer they get, the larger the gravitational wave. Divided by the distance to the source. 101 00:11:29,120 --> 00:11:35,840 So how far can these sources be? Well, these are in our galaxy, so they're not that far, but they're very far apart. 102 00:11:36,380 --> 00:11:43,460 So this actually the number for this particular system on earth is about ten to the -26. 103 00:11:44,450 --> 00:11:51,950 Fractional reserve. And that's way too small for any instrument we know, even space instruments. 104 00:11:52,370 --> 00:11:56,510 And this is, again, a very low, very long period, low frequency. 105 00:11:57,230 --> 00:12:06,380 So let's think about and this would be stronger if they had close to coalescence, but that wouldn't happen for about 300 million years. 106 00:12:06,560 --> 00:12:12,970 We don't want to wait that long. So let's imagine that we're looking at the larger volume of space. 107 00:12:13,310 --> 00:12:17,960 Let's imagine that we are sensitive to the Virgo cluster in the Virgo cluster. 108 00:12:18,320 --> 00:12:29,149 Systems like these are probably coalescing not once every 10,000 years, like on let go in our galaxy, but once every 50 years. 109 00:12:29,150 --> 00:12:33,800 And that's getting reasonable, especially taking astronomy errors into account. 110 00:12:34,130 --> 00:12:38,240 So let's imagine we have a detector that's sensitive to that distance. 111 00:12:38,690 --> 00:12:45,320 How strong would have detected have to be? And the answer is ten to the -21. 112 00:12:46,070 --> 00:12:54,440 So a binary system of neutron stars at about 26 produce a fractional change. 113 00:12:54,440 --> 00:12:57,760 And this is near the earth of it. But in ten to the 21. 114 00:12:58,970 --> 00:13:03,890 How small is that? Well, very small. And of course, let's put some numbers in there. 115 00:13:04,400 --> 00:13:14,810 If we are thinking about the baseline of the distance between the earth and the sun, a part in terms of the 21 is about an atomic diameter. 116 00:13:15,590 --> 00:13:22,880 So we have looking at the wave that's moving the earth from the sun closer and farther away by an atomic diameter. 117 00:13:23,710 --> 00:13:27,800 That's not suitable. Can we make it shorter? 118 00:13:27,830 --> 00:13:31,810 Well, if we make it shorter, then we need to make it more sensitive. 119 00:13:31,960 --> 00:13:35,080 We can measure on that much better than an atomic diameter. 120 00:13:35,080 --> 00:13:37,030 Right. How much better can we measure? 121 00:13:37,630 --> 00:13:47,050 Well, that's what people in the seventies have begun trying to calculate how small it could be, and they'll hold the suspense until a bit later. 122 00:13:47,080 --> 00:13:50,440 But, you know, now that this number we can measure. 123 00:13:52,270 --> 00:13:54,900 So let's get to the sensitivity of it just a bit later. 124 00:13:54,910 --> 00:14:03,790 Let me tell you now about another calculation that was also not done until about ten years ago, only ten years ago. 125 00:14:04,150 --> 00:14:12,960 And that is the solution for the amplitude of gravitational waves on the form of gravitational waves produced by black holes. 126 00:14:13,660 --> 00:14:22,660 If you really want to calculate how spacetime, how gravitational waves that form, the simplest system you can think of is black holes. 127 00:14:22,780 --> 00:14:28,149 Because then you don't have matter, you don't have radiation, you don't have atoms, you only have space. 128 00:14:28,150 --> 00:14:32,020 That sounds easy. It wasn't that easy, actually. 129 00:14:32,020 --> 00:14:36,730 You cannot get analytical solutions. You can only get numerical solutions. 130 00:14:37,060 --> 00:14:41,200 And that's what people need. Again, not more than ten years ago. 131 00:14:41,710 --> 00:14:46,510 Now they are done regularly, but they take a lot of computer time. 132 00:14:47,230 --> 00:14:57,640 And I'm showing you I'm going to show you here the movie of just just because two black holes that have about 30 solar masses. 133 00:14:58,990 --> 00:15:05,830 And what I am going to show you is a colour representing curvature near the black holes. 134 00:15:06,280 --> 00:15:13,300 And then the solution to the gravitational wave that you would measured far away from the black holes in the radiation zone. 135 00:15:13,810 --> 00:15:21,520 So this is going to be the gravitational wave measure, let's say universe, and this is going to be the curvature near the black holes. 136 00:15:22,900 --> 00:15:26,080 And let me show you. 137 00:15:26,080 --> 00:15:31,090 It's actually quite nice. I know how to make that thing go away. 138 00:15:32,120 --> 00:15:37,490 There are. Notice that the black holes are relatively far away. 139 00:15:37,510 --> 00:15:42,960 This is half a second before coalescence. There's a very large curvature near each black hole. 140 00:15:43,000 --> 00:15:48,820 This is a cat on a plane, of course. Could virtually anything in all four dimensions, not just three. 141 00:15:50,920 --> 00:15:55,990 But this is just on a plane. But the solution is calculated all over space time. 142 00:15:56,350 --> 00:16:01,059 So they're getting closer and closer. Now we are 250 milliseconds before coalescence. 143 00:16:01,060 --> 00:16:05,190 We see that as they get closer, they have rotating faster. 144 00:16:05,200 --> 00:16:14,349 The amplitude of gravitational waves is larger. Now the curvatures are beginning to mix up and we knew these that it was going 145 00:16:14,350 --> 00:16:20,020 to get very messy here and how it is now that the horizon's got together. 146 00:16:20,050 --> 00:16:26,170 That's in order to look at all the detail in there. But now look at how simple life form is here. 147 00:16:27,410 --> 00:16:36,360 This relativity, this numerical relativity simulation tells you what happens near the black hole and far from the black hole. 148 00:16:36,380 --> 00:16:43,010 This is what we will measure. We can infer from what we measured what happened pretty near the black hole. 149 00:16:43,280 --> 00:16:50,700 But what we need to measure is relatively simple. It's a way forward that shows increasing amplitude and frequency. 150 00:16:50,720 --> 00:16:54,020 The frequency depends on the masses of the black holes. 151 00:16:54,380 --> 00:17:07,000 And then a ring down. So in the seventies, people thought that ten to the -21 was something that was achievable with an interferometer. 152 00:17:08,850 --> 00:17:17,850 They had to calculate what their sources of noise could be in an interferometer if you didn't to see how long it could be down. 153 00:17:17,850 --> 00:17:24,750 And they calculated that rate wise, it's one of the first ones that put not just that concept of any that phenomena that had appeared earlier, 154 00:17:24,960 --> 00:17:28,350 but also the noises, sensitivity of a detector. 155 00:17:29,610 --> 00:17:34,860 So he imagined kilometre sized detectors for kilometres. 156 00:17:35,190 --> 00:17:39,330 So that means that you would have to measure about ten to the -18 metres. 157 00:17:40,640 --> 00:17:51,320 1810 to the -18, an atom is sent to the minus two as a proton is sent to the -15 metres. 158 00:17:51,530 --> 00:18:02,900 We are talking about ten to the -18, a part in a thousand of a proton diameter in the difference between this length and that length. 159 00:18:03,760 --> 00:18:08,639 Each of these lands four kilometres long. And he put some noise. 160 00:18:08,640 --> 00:18:12,470 Kirsten down, made some estimates with some other people that hurt him. 161 00:18:12,770 --> 00:18:17,090 He said it's possible. Took some time to convince. 162 00:18:17,810 --> 00:18:21,320 So this was the idea is detectors four kilometres long. 163 00:18:21,320 --> 00:18:26,060 It took some time to convince the National Science Foundation, but that was done. 164 00:18:26,480 --> 00:18:28,250 It was done in a very special way. 165 00:18:28,260 --> 00:18:35,890 It said that if you build it four kilometres long and you installed the technology we have now, that's in the nineties. 166 00:18:35,900 --> 00:18:46,400 I was starting my to that. Then you can measure ten to the -21, but that would probably not be enough to detect gravitational waves. 167 00:18:46,400 --> 00:18:50,990 You would have to install technology to do ten times better. 168 00:18:51,920 --> 00:18:56,750 And they from the beginning, this was thought to face experiment. 169 00:18:57,470 --> 00:19:02,959 So the funding in the nineties was to build these two facilities to put in their what 170 00:19:02,960 --> 00:19:08,540 was called the initial legal detectors that would have noise at about ten to the -21. 171 00:19:08,990 --> 00:19:13,610 Oh, that was a facility in Livingston, Louisiana. 172 00:19:16,540 --> 00:19:20,780 That's near Baton Rouge, where they live. Again, you can go to Google Maps. 173 00:19:20,800 --> 00:19:28,810 I love doing that. If you go to the Google Maps, you see the real thing and then you can fly 3000 kilometres. 174 00:19:29,770 --> 00:19:38,470 If you go straight 10 milliseconds, if you travel at the speed of light to Hanford and that's in the middle of the desert, 175 00:19:38,710 --> 00:19:42,520 you go from the swamp, from a logging forest to the middle of the desert. 176 00:19:42,820 --> 00:19:49,040 And those are the to lie detectors. That was in the early nineties. 177 00:19:49,340 --> 00:19:54,310 In the 2000, we began measuring noise in those detectors. 178 00:19:54,320 --> 00:20:02,960 It took about five years to get to the noise sensitivity to the design sensitivity that those detectors had been built for. 179 00:20:03,230 --> 00:20:10,910 And that is about this. So this is a way in which we represent the noise in the detectors, not the signal, but the noise. 180 00:20:10,910 --> 00:20:18,469 But the noise is what determines the sensitivity. So we have we call this a noise, spectral density, 181 00:20:18,470 --> 00:20:27,740 because it tells us about the noise measured us through a fractional distance measurement as a function of frequency. 182 00:20:27,950 --> 00:20:31,850 So it's the noise that we have at each frequency because this is noise. 183 00:20:31,880 --> 00:20:35,780 Low is good. So this is our sweet spot. 184 00:20:35,960 --> 00:20:45,470 This was this was our sweet spot. The initial liger at low frequencies, we are limited by seismic noise and high frequencies. 185 00:20:45,470 --> 00:20:50,870 We were limited by the quantum noise in the laser. 186 00:20:50,900 --> 00:21:00,350 We are measuring photons. So the uncertainty, that number of photons, that is short noise, we call that quantum noise, but that is just show noise. 187 00:21:00,770 --> 00:21:13,700 And in the middle, we don't quite know what the actual noise was, but it was consistent or very close to the Brownian motion of the mirrors. 188 00:21:14,030 --> 00:21:20,650 The mirrors are in vacuum, so they're not pushed around by air molecules, but they are made of atoms that you had at warm temperatures. 189 00:21:20,840 --> 00:21:26,299 They are vibrating. And we measured the position of the front face of the mirrors. 190 00:21:26,300 --> 00:21:29,450 So we do have some sensitivity to that. 191 00:21:29,750 --> 00:21:33,950 We have to calculate the frequency dependence of that of that thermal noise. 192 00:21:33,950 --> 00:21:41,160 And that is this one. Notice that these numbers are ten to the -22, ten to the -23. 193 00:21:41,310 --> 00:21:44,890 You would say, well, that's ten, 100 times better than ten to the -21. 194 00:21:45,160 --> 00:21:54,010 So much better. We worded because when we say ten to the -21, what we are thinking is a rude, mean square. 195 00:21:54,030 --> 00:21:56,040 If you have a noisy time series. 196 00:21:57,480 --> 00:22:05,070 But when you have a time series, what you have to think is you have a time series that is filtered through some frequency band. 197 00:22:05,640 --> 00:22:15,299 So the rudiments of that time series, let's say filter between 102 hundred hertz would be the integral of this keyword, 198 00:22:15,300 --> 00:22:19,500 of this good versus frequency, and then taking the square root. 199 00:22:19,830 --> 00:22:25,170 And then you see here that this is two or 310 to the -23. 200 00:22:25,170 --> 00:22:29,040 You multiply by ten because you're integrating over frequency. 201 00:22:29,040 --> 00:22:35,640 So that gives you ten to the -20. But it would have been better than that, but only near these frequencies. 202 00:22:36,580 --> 00:22:40,090 That was initial logo. It took a while to get it done. 203 00:22:40,270 --> 00:22:50,440 This was what we had in 2010. That was enough to see this binary neutron start coalescence, this individual cluster at 1520 megabytes six. 204 00:22:50,890 --> 00:22:56,530 We didn't see any. But they told you that we expect those to happen once every 50 years. 205 00:22:56,920 --> 00:23:00,310 So it wasn't a surprise. We would hoping to be lucky, but we didn't. 206 00:23:01,330 --> 00:23:11,560 But we had this plan. We had this plan for advanced legal for that we do better seismic isolation so that we have lower seismic noise. 207 00:23:12,010 --> 00:23:18,879 We have a different way. We have lowered sources of energy dissipation. 208 00:23:18,880 --> 00:23:24,280 And in the question period, you can ask me why that reduces the brown noise, but that's what it does. 209 00:23:24,940 --> 00:23:34,810 Let me see bezel less brown and noise at these frequencies and we can use more for those higher laser power and that reduces the quantum noise. 210 00:23:35,530 --> 00:23:45,160 So that was a plan for the advanced liger. Of course, that plan takes a lot of technology, a lot of time, a lot of money. 211 00:23:46,000 --> 00:23:49,569 Let me just show you some pictures. It's all in a vacuum. 212 00:23:49,570 --> 00:23:57,670 So these lasers are travelling for kilometres in a vacuum tube that has to be big so that you don't have scads of light bouncing on the walls. 213 00:23:57,700 --> 00:24:07,030 These are emitted and a half in diameter. You have the laser, which is an amplified laser of 200 watts, and we had only 24 now. 214 00:24:07,600 --> 00:24:14,140 The mirrors are state of the art mirrors, both for the polishing and the coding of the mirrors. 215 00:24:14,170 --> 00:24:23,799 The laser is an infrared, so our mirrors are made reflective or transmission when they need to be in the infrared and also at twice the frequency, 216 00:24:23,800 --> 00:24:32,800 because we also use three lasers to align. So they are two frequencies and we have them for seismic isolation. 217 00:24:32,800 --> 00:24:36,690 We have them hanging from a quadruple pendulum. 218 00:24:36,710 --> 00:24:41,200 So this mass, which is a full silica mass, is hanging from this mass. 219 00:24:41,560 --> 00:24:47,230 It's hugging from blades is coming from blades. We have four horizontal pendulums. 220 00:24:47,230 --> 00:24:54,430 We have three a couple system of blades for vertical seismic isolation. 221 00:24:54,670 --> 00:25:00,160 And we have all of that hanging from a seismic isolation platform, 222 00:25:00,610 --> 00:25:07,540 which is a triple platform in which we measure the noise, the relative noise between the stages and we can select. 223 00:25:08,050 --> 00:25:20,770 And all of that is standing on hydraulic actuators that also have ways to measure this is that helps the hundred. 224 00:25:22,240 --> 00:25:25,440 Oh, 00i see. I've been talking for too long. 225 00:25:25,450 --> 00:25:30,130 I think that's what it's telling me that is all hanging from. 226 00:25:32,950 --> 00:25:38,950 Yes. Okay. This is the seismic isolation platform. 227 00:25:41,050 --> 00:25:48,670 I said this is a hydraulic actuators. These can be moved by hundreds of microns to compensate for earth tides. 228 00:25:49,750 --> 00:25:53,470 And all of that is in vacuum chambers. Yes. 229 00:25:53,500 --> 00:26:00,340 All of that all of these hanging from the are hanging from that is is inside vacuum chambers. 230 00:26:00,790 --> 00:26:07,600 And then it all has to work together. The topology has these quadruple pendulums. 231 00:26:07,840 --> 00:26:11,710 We have two we need to arm because we make the line go back and forth. 232 00:26:12,010 --> 00:26:18,940 Many times what we actually have is an optical resonant cavity in each arm, fabricated cavity. 233 00:26:19,330 --> 00:26:27,610 We keep the interference of the output destructive, which means that the interference going back to the input is constructive. 234 00:26:27,610 --> 00:26:33,950 So this is acting like a mirror. So we have another mirror in here that we call a powered resulting metre that makes 235 00:26:33,950 --> 00:26:38,980 the light go back and forth that it has more photons circulating in the system. 236 00:26:39,130 --> 00:26:47,320 The quantum noise. We have a signal recycling mirror to the output and with all of that we have a noisy time series 237 00:26:47,320 --> 00:26:55,600 and the end that has about anonymous of ten to the -22 foot advance like Putin to the -22. 238 00:26:56,730 --> 00:27:04,830 Now with that, I'm serious. We need to search for gravitational waves and we search for several kinds of gravitational waves. 239 00:27:05,220 --> 00:27:12,720 We look for cross correlation between the two, nor is it time series in the ligo's detectors. 240 00:27:12,720 --> 00:27:18,330 And that's because then we can we can measure the unknown. 241 00:27:18,390 --> 00:27:21,870 We don't need to have a template. We don't need to know what we are looking for. 242 00:27:21,870 --> 00:27:24,990 We just cross correlate Hanford with Livingston. 243 00:27:25,230 --> 00:27:31,110 We see whether there's something there. If there is, then we can find anything in there that very sensitive. 244 00:27:31,110 --> 00:27:37,980 That would be the case, for example, for the supernova explosion. What wave forms are very complicated and not very well known. 245 00:27:38,640 --> 00:27:47,160 We look for periodic signals. Those would be produced by neutron stars rotating if they are not perfectly symmetric. 246 00:27:47,310 --> 00:27:56,700 That's another amazing thing about neutron stars. They have the mass of the sun concentrated in the size of a small city, about ten kilometres. 247 00:27:57,270 --> 00:28:04,440 We, as Felicity said, that the highest mountain is a couple of centimetres. 248 00:28:05,460 --> 00:28:09,330 It's a pothole. I know it's in Baton Rouge. 249 00:28:09,330 --> 00:28:15,120 Isn't that larger than that? So but they cannot be perfect. 250 00:28:15,120 --> 00:28:20,429 If they are not perfect spheres, then they will produce gravitational waves and those will be periodic. 251 00:28:20,430 --> 00:28:21,990 So we know how to look for those. 252 00:28:22,410 --> 00:28:31,920 We have way for four binary systems, whether they are black holes or do we have the full waveform or neutron stars where it gets messy at the end. 253 00:28:32,190 --> 00:28:36,239 But then we know that at the end neutron stars would produce a black hole. 254 00:28:36,240 --> 00:28:42,510 So we would see a black hole getting born, just produce perhaps with the supernova too. 255 00:28:42,780 --> 00:28:49,109 And we could even look for cross correlations of stochastic background between 256 00:28:49,110 --> 00:28:53,160 the detectors we looked at using cross correlation between the detectors. 257 00:28:53,520 --> 00:29:01,770 This would be the case for gravitational waves from the early universe which exist in our frequency band, but they're very, very small. 258 00:29:02,430 --> 00:29:05,940 So so far we've put the best upper limits in those frequencies, 259 00:29:06,240 --> 00:29:11,370 but we don't expect to see those because they yet they are predicted to be way, way below the noise. 260 00:29:11,700 --> 00:29:18,150 But we could see many of these signals or many of these signals if they are small and all mixed together. 261 00:29:18,510 --> 00:29:20,430 We call that the astrophysical bug. 262 00:29:21,210 --> 00:29:27,660 So we need to build the detectors, develop the technology, build the detectors, operate the detectors, look for the data. 263 00:29:28,050 --> 00:29:32,970 That's what we've been doing for years and that takes a lot of people. 264 00:29:32,970 --> 00:29:39,280 So when I say a thousand people, it's really true. The League of Collaboration was formed in 97. 265 00:29:39,300 --> 00:29:47,430 It has now about 1100 members. We are in 15 countries and since 2007 we actually take the data and jointly 266 00:29:47,430 --> 00:29:53,100 analyse data with a video collaboration that has about 250 people in Europe. 267 00:29:53,520 --> 00:29:56,940 So it takes a lot of people a lot of time to do that. 268 00:29:57,330 --> 00:30:06,899 But we are very proud. And I just want to show you that there's about 235 people in the UK doing this too in several different universities. 269 00:30:06,900 --> 00:30:17,250 And there's everything from hardware development, data analysis, future detectors, upgrades, optics, technology for future detectors. 270 00:30:17,820 --> 00:30:24,389 We have a very nice Life magazine. One of the main editors and magazine is in Birmingham. 271 00:30:24,390 --> 00:30:33,870 There's a lot of outreach, so a lot of this is the main design for the these quadruple suspensions was done in the University of Glasgow. 272 00:30:35,170 --> 00:30:37,150 So we are very good friends of the UK. 273 00:30:39,530 --> 00:30:49,240 In 2015, we had the Liger, the advance liger detector's working at a sensitivity that we thought it was quite good. 274 00:30:49,250 --> 00:30:52,950 It was about three times better to be better, three times better. 275 00:30:52,970 --> 00:31:03,050 Initial Liger. In initial liger we had this reached the binary neutral stars of 26 with this noise level that we had in September 2015, 276 00:31:03,290 --> 00:31:07,430 we had to reach to neutron stars of about 70, make up us six. 277 00:31:07,880 --> 00:31:12,680 So we had decided in advance of this that at that level it was good to take data 278 00:31:12,680 --> 00:31:18,350 for a few months and then stop and keep tweaking the detector to get to the 246, 279 00:31:18,680 --> 00:31:23,089 which is the potential of the advanced liger detectors. So that's what we had. 280 00:31:23,090 --> 00:31:28,490 We would using 20 volts. That's why one of the reasons we are not so good everywhere. 281 00:31:28,820 --> 00:31:33,740 But we had a sensitivity that was very decent. Look at this, ten to the -23 there. 282 00:31:34,940 --> 00:31:46,099 And before we started taking data into 24, seven before we started officially this observing run while we were running tests, diagnosing, 283 00:31:46,100 --> 00:31:53,960 calibrating the data in the detectors, analysing the data whenever that was time for the two detectors to take the data jointly. 284 00:31:54,590 --> 00:31:58,940 We saw this on September 14 of 2058. 285 00:31:59,690 --> 00:32:04,429 That must have been the busiest Monday well I've had in my life. 286 00:32:04,430 --> 00:32:12,590 But most of the people in collaboration have had probably some people went on vacation where he was, for example, was in vacation. 287 00:32:12,980 --> 00:32:20,480 And he knew he suspected about this not because he was reading it, but because he was reading the electronic logs. 288 00:32:20,910 --> 00:32:32,660 To use this, we have something in space and he saw that we had cancelled maintenance and say so he could that say what happened this attack. 289 00:32:35,240 --> 00:32:45,080 So I'm telling you, you probably know this already, but in blue is the data with some minimal filtering in living still in red, 290 00:32:45,260 --> 00:32:56,600 there's the data again with minimal filtering in a frequency band and taking away some some power lines in Hanford 3000 kilometres away. 291 00:32:56,810 --> 00:33:04,340 And here we have shifted the Hanford data 7 milliseconds so that it lines up with Livingstone 292 00:33:04,880 --> 00:33:12,680 and we saw something that looked larger than the noise in both detectors the same. 293 00:33:13,160 --> 00:33:20,450 So that was a gravitational wave. It was larger than the noise, it was the same in both detectors. 294 00:33:20,450 --> 00:33:24,890 We couldn't find any other explanation. It was the gravitational wave. 295 00:33:25,250 --> 00:33:32,840 And on top of that you could see the signature of a sine wave that got higher in frequency, 296 00:33:32,840 --> 00:33:37,610 higher in amplitude, and then smoothly settled down to that noise. 297 00:33:38,090 --> 00:33:46,250 That's what you expect from black holes. And then when you calculated the black hole masses, these were so distorted it must black holes. 298 00:33:46,910 --> 00:33:51,799 I mean, this is all incredible. We were not expecting gravitational waves. 299 00:33:51,800 --> 00:33:55,250 And here we had a gravitational wave. We were not expecting black holes. 300 00:33:55,250 --> 00:33:59,270 And had we had black holes, we would not expect them through the black hole. 301 00:33:59,270 --> 00:34:03,380 Nobody was. And here they were. We couldn't believe it. 302 00:34:04,160 --> 00:34:10,010 It took us months to convince ourselves that this was real and there was no other explanation for this. 303 00:34:10,280 --> 00:34:17,900 But you know that now we did discover gravitational waves on September 14, 2050, but that wasn't the only one. 304 00:34:18,500 --> 00:34:23,360 We had planned to take data for three months. We decided to take that for four months. 305 00:34:24,080 --> 00:34:30,260 We had a lot to do on the detector, so we had decided to stop in January and on December, 306 00:34:30,290 --> 00:34:36,110 December 26 in Europe, but it was still December 25, it was still Christmas in the US. 307 00:34:37,250 --> 00:34:40,520 We saw another signal. It wasn't as big. 308 00:34:40,520 --> 00:34:44,839 I cannot. I'll show you some drawings. It wasn't as clear as that. 309 00:34:44,840 --> 00:34:48,829 First detection. That's because these are smaller black holes. 310 00:34:48,830 --> 00:34:54,380 These are 30 solar mass black holes each 29 and 36, this at eight than 14. 311 00:34:54,710 --> 00:34:57,800 So they're quite a bit smaller even though they are different in mass. 312 00:34:58,040 --> 00:35:05,510 So the signal was smaller in amplitude, but it was longer and that's why we could detect it with very, very high confidence. 313 00:35:05,990 --> 00:35:14,360 And we even had a third candidate. We call this a candidate not a detection because it's much lower amplitude. 314 00:35:14,570 --> 00:35:20,390 It has it is consistent with binary black hole system with masses in between these two. 315 00:35:20,690 --> 00:35:23,900 It depends who you ask whether this is a detection or not. 316 00:35:24,290 --> 00:35:31,399 If you ask me, it has a 15% chance of not being a detection of not being an astrophysical signal. 317 00:35:31,400 --> 00:35:38,860 So for me that's not quite enough. But as a member of a family, it's a very reasonable candidate if we didn't have this. 318 00:35:39,530 --> 00:35:48,500 We wouldn't pay any attention to that. But now we have this debate, and this is the beginning of gravitational wave astronomy. 319 00:35:49,160 --> 00:35:54,410 We do know from X-ray astronomy that there are black holes of stellar masses. 320 00:35:54,440 --> 00:36:02,680 We know that that's how they start from supernova explosions with a few solid masses, but hot ten solar masses. 321 00:36:02,690 --> 00:36:09,290 These are big stars, but they need to get compressed. We do know that when two neutron stars merge, they form a black hole. 322 00:36:09,290 --> 00:36:18,920 But those would be two black holes. We have seen black holes from X-ray studies, from X-rays emitted from either the stellar companion, 323 00:36:19,730 --> 00:36:23,360 which is not a black hole, or from the jets of the black hole itself. 324 00:36:23,660 --> 00:36:25,760 But they are all relatively low mass. 325 00:36:26,270 --> 00:36:35,840 And here it was our first detection and the two largest black holes known in this range, forming another black hole. 326 00:36:36,170 --> 00:36:40,610 This is 29. This was 36 solar masses and this was 62. 327 00:36:41,630 --> 00:36:48,890 And for those people who can do algebra in their heads, which is not me, I always make mistakes. 328 00:36:48,900 --> 00:36:53,060 But 29 to 36, it's not 62. 329 00:36:54,060 --> 00:37:02,220 There are three solid masses missing, though. Those are three solid masses that went away in energy in the gravitational waves. 330 00:37:02,760 --> 00:37:06,120 Energy and mass are the same. Equal, empty square. 331 00:37:06,150 --> 00:37:12,030 That's Einstein's formula. That would be three solid masses in energy of gravitational waves. 332 00:37:12,330 --> 00:37:19,220 If that had been visible energy in a fraction of a second, that's us. 333 00:37:19,290 --> 00:37:25,050 All the galaxies in the visible together, exploding in a fraction of a second. 334 00:37:25,830 --> 00:37:30,480 That's how much power that was indeed. It was incredible. 335 00:37:30,870 --> 00:37:38,880 No wonder it took us months to convince ourselves the December the December event was more normal, we say. 336 00:37:39,180 --> 00:37:43,660 It was similar, consistent with these masses. The candidate is in between. 337 00:37:43,920 --> 00:37:53,180 Again, we said we sent a member of the family. This is actually a few more details about that detection. 338 00:37:53,210 --> 00:37:57,740 This is what they showed you before. Just the raw data minimally filtered. 339 00:37:58,130 --> 00:38:04,430 This is the numerical simulation I showed you before medical relativity simulation for 340 00:38:04,430 --> 00:38:09,409 two black holes compared with the template that we used to find it an approximate, 341 00:38:09,410 --> 00:38:17,930 approximate template and with a reconstructed template that we can use just by cross correlating the date that without using any assumption. 342 00:38:18,350 --> 00:38:23,780 So you can recover the template if even if you don't know anything about the wafer. 343 00:38:24,320 --> 00:38:28,580 That only happened because this is such a strong signal in general, you can not do that. 344 00:38:29,240 --> 00:38:33,680 If you subtract the signal from the data, then you get noise. 345 00:38:34,070 --> 00:38:41,870 So this is the Time series I was telling you about before that you can reconstruct an amplitude from their spectral density. 346 00:38:42,320 --> 00:38:49,250 This is another way we like to plot it. This is the colour plot that shows the normalised amplitude of the wave. 347 00:38:49,730 --> 00:38:53,420 So blue in here is normal noise. And this is time. 348 00:38:53,570 --> 00:38:57,200 This is frequency. So this is a time frequency plot. 349 00:38:57,200 --> 00:39:02,870 And this is showing you in Hanford that the signal gets brighter, larger amplitude, 350 00:39:03,200 --> 00:39:09,410 higher in frequency, and then it gets the node at the highest frequency and it disappears. 351 00:39:10,040 --> 00:39:14,749 This is what we call a took and this is in Hanford, this is in Livingston. 352 00:39:14,750 --> 00:39:18,700 You see that it's demoed in Livingston and Hanford. 353 00:39:18,710 --> 00:39:22,970 So you would say you saw it. It was Lanzarote in Hanford and in Livingston. 354 00:39:22,970 --> 00:39:29,150 But here you show the the same amplitude, they are the same calibrated amplitude. 355 00:39:29,720 --> 00:39:34,910 But Livingston had a bit higher noise at these frequencies than Hanford did. 356 00:39:35,210 --> 00:39:40,400 So the signal to noise ratio was higher in Hanford, the Livingston. 357 00:39:40,550 --> 00:39:45,620 That's why this is brighter than that. But they have the same calibrated amplitude. 358 00:39:46,280 --> 00:39:50,510 So there's a lot of information in these plots thing more than people usually assume. 359 00:39:51,290 --> 00:39:58,759 But this is not how we found the signal. We are not plotting this time series on the screen to see all of this. 360 00:39:58,760 --> 00:40:03,140 The gravitational body would have to be too many graduate students to do this in parallel. 361 00:40:03,950 --> 00:40:12,110 So what we actually do is we use something called smart filtering, which is a cross correlation, 362 00:40:13,100 --> 00:40:21,650 but it's a cross correlation of a template that we suspect this in the data with the data itself. 363 00:40:22,280 --> 00:40:30,769 So in this example, the red is the data out of the detector that may or may not have a gravitational wave in this simulation. 364 00:40:30,770 --> 00:40:35,839 We have injected the gravitational wave into there, but you cannot see it's hiding in the data. 365 00:40:35,840 --> 00:40:40,550 So we show it to you in blue. This is a gravitational wave hiding in the data. 366 00:40:41,120 --> 00:40:48,689 So if you know you are looking for this kind of wave form, then what you do is you take the wave form of any amplitude. 367 00:40:48,690 --> 00:40:56,420 It doesn't have to be the same when you take a white form of any amplitude and then you cross correlated with the data as a function of time. 368 00:40:57,080 --> 00:41:00,800 And that is what this movie is showing. 369 00:41:01,640 --> 00:41:07,220 So that cross correlation gives you a random number that has some distribution. 370 00:41:08,460 --> 00:41:12,200 Until it finds something in the data. 371 00:41:12,210 --> 00:41:15,870 So if there is something similar to the way from your looking forward, 372 00:41:16,140 --> 00:41:21,750 then that will be a peak in this time series, which is a signal to noise ratio time series. 373 00:41:22,140 --> 00:41:29,160 And then if that peak is larger than what you expect from the statistics in the noise, you call this a candidate. 374 00:41:30,120 --> 00:41:38,910 Now this is a candidate in one detector. You do this, you do this for tens of thousands of templates in one detector. 375 00:41:39,450 --> 00:41:44,400 And then you do the same thing with tens of thousands of templates in the other detector. 376 00:41:44,640 --> 00:41:53,460 And you have a list of candidates. And now you look for candidates that within a ten millisecond window that slide through that window, 377 00:41:53,670 --> 00:41:58,080 we actually use a bit wide that window to account for possible errors, 20 milliseconds. 378 00:41:58,950 --> 00:42:02,400 So you look for coincidences within a time window. 379 00:42:02,880 --> 00:42:11,070 And if there are coincidences, then you check that the template that produced this big is consistent with a template that produced this big. 380 00:42:11,520 --> 00:42:15,060 And only then you call that a coincidence. 381 00:42:16,200 --> 00:42:23,640 Now. How many coincidences do you think we found in that in those four months of data which weighed about 50 times of coincident data? 382 00:42:25,850 --> 00:42:29,510 We found about 2000 of those concerns as. 383 00:42:30,950 --> 00:42:35,840 But we know that most of those are not astrophysical. 384 00:42:35,850 --> 00:42:40,370 Sometimes you find that coincidences off a peak just by chance. 385 00:42:40,910 --> 00:42:46,710 And you would think that these things are easy to pick. But our data is not as clean as this at all. 386 00:42:46,970 --> 00:42:51,230 Our data has a lot of glitches that even in the raw data you can pick up. 387 00:42:51,560 --> 00:42:57,930 So this picks up even more. So you have lots of spikes and sometimes they had a quality to by chance. 388 00:42:58,280 --> 00:43:05,240 So we have to be careful and differentiate a real astrophysical coincidence from one that is not. 389 00:43:05,700 --> 00:43:14,180 Tell you a bit more about that statistic. But let me show you now what this looks like for the second detection, the December detection. 390 00:43:14,750 --> 00:43:19,489 So I am plotting here in red. The Hanford data is in blue. 391 00:43:19,490 --> 00:43:29,030 The Livingston data. I have mistakenly deleted the time scale, but this is a second in here, one full second or the fraction of a second. 392 00:43:29,610 --> 00:43:35,090 And you don't see that way for just don't see it in the filter data. 393 00:43:35,100 --> 00:43:43,040 There's no way to filter the data, but in a reasonable way that shows it as obvious as for the first one. 394 00:43:43,340 --> 00:43:47,540 So I have to superimpose in the template that we used to find the data. 395 00:43:48,050 --> 00:43:55,640 So that's just like in the example I showed you before, I have to plot it in there to show you that it said, but that's not how we found it. 396 00:43:55,640 --> 00:44:03,890 We found it in this signal, the Noise Time series, because it produced a big, big in Hanford, a bit smaller peak in Livingston. 397 00:44:04,310 --> 00:44:12,710 And it was because the same template views these large amplitude peaks within one millisecond in this time. 398 00:44:12,980 --> 00:44:20,000 In this case, it wasn't seven millisecond. It was just one millisecond time difference that we knew that this was a detection to. 399 00:44:22,090 --> 00:44:28,299 This is a statistic that we found. This is how we say this coincidence is not just any coincidence. 400 00:44:28,300 --> 00:44:36,000 It's very significant. I told you that we calculate the signal to noise ratio in one detector, a signal to noise ratio and the other detector. 401 00:44:36,000 --> 00:44:40,709 We make some weighting in those with chi squared and then we combine those two numbers. 402 00:44:40,710 --> 00:44:45,960 So we have a number that we call a detection statistic that measures the loudness of the signal. 403 00:44:46,800 --> 00:44:55,080 This is a number of events that we found, a histogram of the number of events we found number of coincidences versus this loudness. 404 00:44:55,770 --> 00:44:59,220 So this was one our loudest event. 405 00:44:59,640 --> 00:45:04,770 This is one we found one event that had the loudness of about 23. 406 00:45:05,430 --> 00:45:08,820 We had another event that had the loudness of 13. 407 00:45:09,480 --> 00:45:21,030 We had another one that had 11 and something, but we had like a thousand coincidences all together of smaller loudness. 408 00:45:22,260 --> 00:45:27,510 The statistic of a we had like a thousand in these 50 days of data. 409 00:45:28,140 --> 00:45:32,670 So there are lots and lots of coincidences, but we know those are not all events, 410 00:45:32,670 --> 00:45:38,730 because if we calculate how many coincidences we expect as a function of loudness, 411 00:45:39,150 --> 00:45:47,010 we can say that some number is, is, is significant and some other not some other number is just consistent with not. 412 00:45:47,070 --> 00:45:49,110 It's the way we do that. 413 00:45:49,110 --> 00:45:57,180 The way we would like to do that is see our experiment and say there are no gravitational waves in the experiment, but we could not do that. 414 00:45:57,780 --> 00:46:04,050 The nice thing about gravitational waves is that they bring very beauty information because they go through everything, 415 00:46:04,740 --> 00:46:06,570 but that's why we cannot see them. 416 00:46:07,320 --> 00:46:21,030 So what we actually do is we fake the data one detector, we add to the clock, one detector to 10.32 one 6 seconds, some arbitrary number of seconds, 417 00:46:21,510 --> 00:46:31,500 such that when we do the analysis and find the coincidence, we know it's a coincidence in a non physical time window. 418 00:46:31,680 --> 00:46:38,819 So we know it's not that it's a coincidence, but it's Northern Astrophysical signal and we can do this time. 419 00:46:38,820 --> 00:46:44,250 Shifting a million times. 10 million times depends on how much that we have. 420 00:46:44,250 --> 00:46:47,940 The more that we have, the more time shifts we can have in both directions. 421 00:46:48,540 --> 00:46:51,869 So that's what we do. For this signal. 422 00:46:51,870 --> 00:46:54,030 It was so loud that we did. 423 00:46:54,040 --> 00:47:02,190 This time shifts more than 10 million times and we didn't find anything that got the loudest things we find were out of about 20. 424 00:47:02,520 --> 00:47:12,420 And it turns out that most of those non astrophysical coincidences was the signal in one detector with something else in the other detector. 425 00:47:13,050 --> 00:47:20,190 So if we cut out the fraction of a second that had this astrophysical signal in both detectors, 426 00:47:20,550 --> 00:47:26,690 then the number of events with this this blue thing, we didn't find anything louder than 13 up. 427 00:47:26,700 --> 00:47:31,840 So. So this is huge. This was more than Five Sigma. 428 00:47:33,450 --> 00:47:36,690 This one is also heute after. 429 00:47:36,990 --> 00:47:41,760 Even if you don't take this one out, you have to do about 200,000 shifts. 430 00:47:41,760 --> 00:47:48,059 So that gives you the probability, 100,000 of these being an artificial noise. 431 00:47:48,060 --> 00:47:54,300 Coincidence, a false alarm. But if you take it away again, you don't see anything like that. 432 00:47:54,730 --> 00:47:59,400 And the ones you see are more lost. But not all the coins. 433 00:47:59,560 --> 00:48:03,270 Is this signalling one detected coincidence with something else? 434 00:48:03,820 --> 00:48:11,430 So this one was also a very strong detection. This one not so much for this one. 435 00:48:11,730 --> 00:48:19,110 You only have to do the equivalent of two years of data to see coincidences are just as loud. 436 00:48:19,650 --> 00:48:25,559 So that means that once every two years you expect a coincidence by chance. 437 00:48:25,560 --> 00:48:34,120 Like this one, we saw one in 50 days. To me, that's not the detection, but that's how you do the statistic. 438 00:48:34,720 --> 00:48:39,640 The templates that that found these signals are shown in here. 439 00:48:40,030 --> 00:48:44,620 The one for the largest one is a short template, large black holes. 440 00:48:44,920 --> 00:48:52,570 The one for the December one is no one has a longer signal, one second signal there than one in between. 441 00:48:53,440 --> 00:49:02,940 So this is all we found for binary black holes. Now my favourite part and I'm running late so I begin to go fast. 442 00:49:03,630 --> 00:49:11,670 But I want to do this. My favourite part and I hope this works, is to play what they call gravity music. 443 00:49:12,540 --> 00:49:17,129 This I showed you the frequencies they had near 100 hertz. 444 00:49:17,130 --> 00:49:23,100 They start at 40 hertz. They they coalesce at about 220 hertz. 445 00:49:23,100 --> 00:49:27,720 They bring down at 250 hertz. Those are frequencies we can hear. 446 00:49:28,650 --> 00:49:33,360 They're not very musical, though. So you might remember having heard these. 447 00:49:33,360 --> 00:49:38,430 But let me let me see if I'm going to have you here. 448 00:49:38,880 --> 00:49:45,600 If the sound works first, what we call the natural beat, that is the minimally filtered signal. 449 00:49:47,980 --> 00:49:51,130 Put in a speaker. I said wipe us waves sound. 450 00:49:55,420 --> 00:50:00,450 That's a December one. This is the September glut. 451 00:50:00,520 --> 00:50:06,370 One is long. The other is short, but they're not very musical. 452 00:50:07,240 --> 00:50:13,630 So we added 400 hertz, and that's what you probably heard on YouTube. 453 00:50:18,550 --> 00:50:22,570 We like that. We call it gravity's music. 454 00:50:24,550 --> 00:50:29,290 Okay. Let me quickly give you some more details and then you can ask me a lot of questions. 455 00:50:29,890 --> 00:50:39,129 First of all, the first one was so loud that the way we sounded is not looking at the Time series, but with an online system. 456 00:50:39,130 --> 00:50:45,310 We had testing. We were testing on that day looking for bursts. 457 00:50:45,700 --> 00:50:53,770 This is cross correlating the signal between the two detectors and looking for cross cross correlations without templates. 458 00:50:54,100 --> 00:51:00,129 Actually, they divide their their coincidences, their cross correlations in three classes. 459 00:51:00,130 --> 00:51:06,040 One that we know it's mostly noise, one that has increasing frequency. 460 00:51:06,040 --> 00:51:13,690 So it has some pattern to the frequency and another that it has increasing frequency, but also a bit of noise. 461 00:51:14,260 --> 00:51:21,790 So on the key on what we call the clean sector, this one that has increasing signal, then there is this very significant detection. 462 00:51:22,120 --> 00:51:28,929 Notice that the second one did not appear. The second one and the candidate were not found with this method. 463 00:51:28,930 --> 00:51:34,780 But this could be found if you bought this second class of sources in there. 464 00:51:34,810 --> 00:51:42,970 We do find some other coincidences. So the probability of this one is not as high as with much filtering. 465 00:51:43,180 --> 00:51:47,800 And that's because we do have transients in the signals we call these little glitches. 466 00:51:48,670 --> 00:51:51,819 We don't know where they come from. There are glitches in the detectors. 467 00:51:51,820 --> 00:51:56,680 In both detectors. About once an hour, they're not coincident. 468 00:51:56,680 --> 00:52:00,100 So we know they're not gravitational waves, but we don't know where they come from. 469 00:52:00,100 --> 00:52:06,580 We don't know how to eliminate them. And we can calculate that there is by chance they could be coincident. 470 00:52:06,910 --> 00:52:12,190 And that's points of which sensitivity to these transients, if they are of that form. 471 00:52:12,910 --> 00:52:18,639 But this is another powerful method. We have to find gravitational waves using templates. 472 00:52:18,640 --> 00:52:26,920 However you can find parameters. We have tens of thousands of templates, each of which has several parameters. 473 00:52:26,920 --> 00:52:32,680 It has the two masses, the mass and the spin of the final black hole. 474 00:52:33,550 --> 00:52:40,840 We know that the final black hole will be spinning because it has the system has angular momentum from the beginning, 475 00:52:40,840 --> 00:52:46,600 from the to it from the two black holes. Even if the black holes are not spinning, there is angular momentum in the system. 476 00:52:47,170 --> 00:52:56,770 We can get the distance, but the distances coupled with the all the inclination of the orbit, we are more sensitive if this signal, 477 00:52:56,770 --> 00:53:03,040 if they orbit is parallel to the plane of the detector, we are almost insensitive if it is at all. 478 00:53:03,490 --> 00:53:09,550 So we have a scatterplot of the match of all these templates in all these planes. 479 00:53:10,000 --> 00:53:15,250 This is the mass must play. This is the largest mass, the smaller mass. 480 00:53:15,250 --> 00:53:20,889 And then we draw a 50% contour and then 95% goes to this line. 481 00:53:20,890 --> 00:53:26,230 And here is a diagonal line because these we this is always the heavier mass. 482 00:53:26,560 --> 00:53:32,520 So when I say 29 and 36 solid masses, well, it's 36 plus minus the feel. 483 00:53:32,570 --> 00:53:40,870 This is a 90% confidence protection, 36 and 29 plus minus a few more. 484 00:53:41,230 --> 00:53:44,560 So that's how we measure the error in. I would put out metres too. 485 00:53:45,700 --> 00:53:50,760 This is the 62 solid mass, the final mass of the black hole on the speed. 486 00:53:50,790 --> 00:53:58,719 The speed is all of this being calculated as a fraction of the maximum spin is the black hole cannot be more than 487 00:53:58,720 --> 00:54:04,540 a certain amount and that certain amount is when the points in the horizon are moving at the speed of light. 488 00:54:05,230 --> 00:54:09,670 The black hole cannot move, cannot spin any faster than that. 489 00:54:10,270 --> 00:54:15,310 So we call that speed one, and then any other speed is going to be less than one. 490 00:54:15,730 --> 00:54:21,730 This final huge black hole was spinning at about 70% of that maximum spin. 491 00:54:22,120 --> 00:54:25,240 This is the distance is about 400 megabytes, six. 492 00:54:25,510 --> 00:54:29,890 That's 1.3 billion light years away. 493 00:54:30,130 --> 00:54:40,780 This event happened 1.3 billion years ago when on earth that would just multicellular organisms getting organised. 494 00:54:41,330 --> 00:54:48,160 Imagine everything that happened when this gravitational wave was travelling and that's the way I like to think about it. 495 00:54:48,730 --> 00:54:53,620 So that's how far it is, but that has about the 50% error in there. 496 00:54:53,920 --> 00:54:58,480 And notice that this is this has a strange shape. 497 00:54:58,870 --> 00:55:02,740 It's most likely that the system was face off meaning. 498 00:55:02,790 --> 00:55:08,820 Then the angular momentum was pointing away from the from the earth, from the line of sight. 499 00:55:09,480 --> 00:55:13,680 This would be face on and this would be edge on. 500 00:55:13,690 --> 00:55:17,910 So it is probably because it was face on that was so strong. 501 00:55:18,540 --> 00:55:23,940 And this is a localisation. Localisation is mostly from these seven millisecond time difference. 502 00:55:24,270 --> 00:55:27,960 The time difference between two detectors gives you a circle in the sky. 503 00:55:29,040 --> 00:55:37,090 If it is zero, if you have no time difference, then you know it's halfway between the between the detectors, and that is simply in the sky. 504 00:55:37,110 --> 00:55:42,960 In this case, it came first to Livingston, then to Hanford. So it was in the South Hemisphere, but it's still a circle. 505 00:55:43,470 --> 00:55:49,800 And you can tell that some points are more likely than others because of the amplitude and phase comparison between the two. 506 00:55:50,070 --> 00:55:59,670 The thing that's similar to detectors, but that's not very good localisation because these are the same plots for the three three years I 507 00:55:59,670 --> 00:56:04,969 was telling you about the first detection and getting the mass must play in the December detection. 508 00:56:04,970 --> 00:56:12,150 Notice they said this is longer, so it's that much thinner release that doesn't give you better mass, especially in this direction, 509 00:56:12,930 --> 00:56:20,489 but it gives you a better but a better combination of parameters that is called the two mass of the system. 510 00:56:20,490 --> 00:56:28,060 That is that we can measure the best. Again, these are two final buzzes and final spins around similar. 511 00:56:29,230 --> 00:56:38,560 This is the spin, the initial spin of the black holes projected over on the angular momentum, the direction of the angular momentum system. 512 00:56:38,920 --> 00:56:42,100 So it's a next being so good. It's zero. Is it two? 513 00:56:42,310 --> 00:56:47,469 If the black holes are not spinning or one is spinning up and the other is spinning down on, 514 00:56:47,470 --> 00:56:54,490 we can tell only for the longer signal, for the December signal that at least one of the black holes was spinning. 515 00:56:54,520 --> 00:56:57,880 The nets being projected in that direction is about point two. 516 00:56:58,840 --> 00:57:04,690 So we can tell quite a bit from this thing that we can also test general relativity. 517 00:57:05,230 --> 00:57:13,809 This is a test on different pulse, Newtonian parameters, and this is the fractional accuracy, 518 00:57:13,810 --> 00:57:20,500 the the limit that you can put in that fractional accuracy of general relativity. 519 00:57:21,580 --> 00:57:31,840 So that's a fractional deviation from g r on for the theorem for the lowest Newtonian, but you get the best and that's about the 10% test. 520 00:57:32,230 --> 00:57:39,840 But that's not very good. 10% test. We have much, much better tests of general relativity from binary pole sets, for example, 521 00:57:40,420 --> 00:57:50,379 we have like a thousand now, so this is not as good, but we have this off order of magnitude in here. 522 00:57:50,380 --> 00:57:59,020 So that's about the fact that of two or three on these higher Newtonian orders and these are the first kind of tests 523 00:57:59,020 --> 00:58:06,490 and these are strong field effects and that's because we are shooting a very different with them than other systems. 524 00:58:08,340 --> 00:58:12,690 We looked for Biden and it was suddenly close. We didn't find any. 525 00:58:12,780 --> 00:58:16,770 I can go back to this to tell you that we will find this later on. 526 00:58:19,110 --> 00:58:27,690 This is showing that even in Initial Largo, it took us years to get to that sensitivity I showed you in the beginning, 527 00:58:28,560 --> 00:58:31,920 saying we think it's going to happen with Advanced Largo. 528 00:58:32,340 --> 00:58:40,600 A few years ago, we had made this plan that was just a guess on what sensitivity we would have in 2015. 529 00:58:40,620 --> 00:58:45,270 It was somebody with Bloomberg in 2016 and 2017. 530 00:58:45,600 --> 00:58:52,590 We would actually spot on. We would in the lower part of this, but 2015 and that's when we got our first detection. 531 00:58:53,100 --> 00:58:58,380 But we haven't made a lot much of a dent in this green band yet. 532 00:58:58,710 --> 00:59:03,170 So this is next science, the next upsetting one. 533 00:59:03,600 --> 00:59:07,979 But we think that we will be progressing. So this will be will start to run. 534 00:59:07,980 --> 00:59:12,020 That will be somewhat on the upper side of this green band this year. 535 00:59:12,030 --> 00:59:15,089 We start that in December would take about six months. 536 00:59:15,090 --> 00:59:27,180 Ron Virgo will join early next year. Then we will be here and then sometime, maybe 2000, 1919 will be on the design sensitivity. 537 00:59:27,690 --> 00:59:35,670 And all this time now we know we will be detecting black holes, but we are going to be detecting other signals too. 538 00:59:36,360 --> 00:59:45,770 So that's very exciting. This is how in the press, this is the sensitivity that was with very preliminary calibration measured at Livingstone. 539 00:59:46,740 --> 00:59:52,049 Oh, this blue was what we had in in February, just after the run. 540 00:59:52,050 --> 00:59:57,270 That was 80 megaparsec until the green is what we had just after the run. 541 00:59:57,270 --> 01:00:02,800 The blue is what we had October 11th and that was 95 make up out of six. 542 01:00:02,820 --> 01:00:06,510 So we had to be better than we were. We wanted to be at the hundred. 543 01:00:07,170 --> 01:00:12,829 Not quite good. But we you get good. And life is not the only story. 544 01:00:12,830 --> 01:00:19,910 I told you that we will be joining next year with some lower sensitivity, but then they will catch up. 545 01:00:20,390 --> 01:00:27,920 Khadra is a Japanese detector's underground cryogenic that would probably begin operating 2000, 18, 19 or so. 546 01:00:29,780 --> 01:00:35,810 India is going to be building the new side, the new observatory next year. 547 01:00:36,110 --> 01:00:42,800 So we'll probably have a detector, reliable detector there, 2024 or so. 548 01:00:43,250 --> 01:00:50,690 So there's going to be a network. You all are going to be hearing about more and more black holes and then neutron 549 01:00:50,690 --> 01:00:56,540 stars and then other gravitational waves from this network and from future upgrades. 550 01:00:57,170 --> 01:01:08,100 It's a very bright future and we are also hoping to see gravitational waves in coincidence with other kinds of ways with electromagnetic waves. 551 01:01:08,120 --> 01:01:12,829 That's what we expect from merging of mutual stuff. But even with black holes, we look. 552 01:01:12,830 --> 01:01:17,400 I mean, people, astronomers, not we, but astronomers look for those neutrinos. 553 01:01:17,420 --> 01:01:21,650 We also look for causes as we extend those if these are supernova. 554 01:01:21,980 --> 01:01:25,100 So this is going to be multi messenger astronomy. 555 01:01:25,520 --> 01:01:29,660 And this is coming. This is going to happen. It's going to take time. 556 01:01:29,960 --> 01:01:38,350 But this is happening. And this is just one window that we have opened in the gravitational wave spectrum. 557 01:01:38,620 --> 01:01:41,440 But there's a whole spectrum of gravitational waves. 558 01:01:41,920 --> 01:01:48,730 We know that small, smallest black holes and neutron stars produce gravitational waves in this frequency band. 559 01:01:49,240 --> 01:01:52,030 Massive black holes and white dwarf binaries. 560 01:01:52,030 --> 01:02:02,260 And the galaxy will produce gravitational waves in the frequency band of a space detector, which has been approved by a European space agency. 561 01:02:02,680 --> 01:02:06,999 We know that supermassive black holes and perhaps the inflationary background 562 01:02:07,000 --> 01:02:12,760 produces gravitational waves of wavelengths that are appropriate for bolster timing. 563 01:02:12,850 --> 01:02:17,980 This is like having an interferometer using the radio beams from neutron stars. 564 01:02:18,160 --> 01:02:22,710 It's like having a galactic interferometer. That's the way I like to think about this. 565 01:02:23,450 --> 01:02:31,360 And early universe. Gravitational waves can be measured through the polarisation of the cosmic microwave background. 566 01:02:31,690 --> 01:02:35,290 And we saw this as thing that it wasn't quite. 567 01:02:35,650 --> 01:02:39,640 But it's coming. These detection textures are getting better and better. 568 01:02:40,300 --> 01:02:48,400 So this is just the beginning. We're going to have more and more of these and all of these in the next decade. 569 01:02:48,640 --> 01:02:51,310 This is a great beginning. Thank you.