WEBVTT 00:00.160 --> 00:04.160 Modern physics is weird. In quantum mechanics we  have the Schrodinger’s cat thought experiment, 00:04.160 --> 00:08.320 a cat in a quantum superposition of  being simultaneously alive and dead. 00:08.320 --> 00:13.760 And in relativity the twin paradox thought  experiment, a space-faring twin ages less 00:13.760 --> 00:19.760 than her stay-at-home sibling. But what if we  could combine these and send our quantum twin 00:19.760 --> 00:24.640 around the galaxy. Could we be simultaneously  old and young? What would that tell us about 00:24.640 --> 00:29.280 the quantum nature of time? And would  we need schrodinger birthday candles? 00:24.640 --> 00:29.520 much as it's inspired us, check it out at the  Space Time merch store. Now on to the episode. 00:29.096 --> 00:32.776 Time is both the most intuitive and most  mysterious feature of the universe. It 00:32.776 --> 00:35.896 seems fundamental and unavoidable,  but physicists and philosophers have 00:35.896 --> 00:40.296 argued over its true nature forever. Even our  best modern theories of physics don’t agree. 00:40.296 --> 00:45.496 In our strangest modern theory, quantum mechanics,  time is surprisingly straightforward. It appears 00:45.496 --> 00:50.136 the Schrodinger equation as a well defined  parameter that moves at the same rate for all 00:50.136 --> 00:56.856 particles—even if those particles have all the  other sorts of quantum weirdness. That’s just 00:56.856 --> 01:02.136 as it is in good ol’ Newtonian physics, time is a  global parameter that ticks the same for everyone 01:02.136 --> 01:07.896 and everyone agrees on a singular “now”. But in relativity, time isn’t global, 01:07.896 --> 01:13.016 it’s local. The rate of its passage depends  on the relative location and speed of a clock 01:13.016 --> 01:18.936 compared to whoever’s watching the clock. Even the  definitions of past, present and future get warped 01:18.936 --> 01:25.016 in relativity. That leads to strange scenarios,  like the twin paradox, where one twin goes on 01:25.016 --> 01:30.216 a quick space adventure at close to the speed  of light and returns to find her stay-at-home 01:30.216 --> 01:36.216 sibling has been worried sick for decades. An essential step in our quest for a unified 01:36.216 --> 01:40.216 picture of reality is to bring together  quantum mechanics and the full relativity 01:40.216 --> 01:46.936 theory—general relativity. At a bare minimum  that means they need to agree on what time is. 01:46.936 --> 01:50.936 In relativity, there’s a beautiful symmetry  between the nature of time and the nature 01:50.936 --> 01:56.936 of space. But in the quantum, time remains ..  special somehow. For example, a quantum object 01:56.936 --> 02:03.576 can occupy two places or travel two paths at  once—like in the double slit experiment. But 02:03.576 --> 02:10.536 can an quantum object occupy two times at once,  or have two different ages at the same time. We 02:10.536 --> 02:16.056 really don’t know how to “make time quantum”. Well, one thing we can try is to explore the 02:16.056 --> 02:21.256 strange relativistic effects of time on  truly quantum systems. Maybe something 02:21.256 --> 02:25.496 like a quantum twin paradox. But what does that even mean? 02:25.496 --> 02:30.456 One approach might be to take a two different  quantum particles and send them on different 02:30.456 --> 02:34.776 routes—with different speeds and/or different  positions in a gravitational field—and then 02:34.776 --> 02:39.896 see how they age differently. In fact, this has  been done already and with incredible precision. 02:39.896 --> 02:44.056 The quantum particles in question are the  cores of atomic clocks. I’ll come back to 02:44.056 --> 02:48.456 exactly how atomic clocks work later, because  they may be the key to this whole thing. 02:48.456 --> 02:53.656 The first twin paradox experiment—the 1971  Hafele–Keating experiment—involved sending 02:53.656 --> 02:59.176 an atomic clock on a plane ride around the world  and comparing to one that stayed at home. Time 02:59.176 --> 03:04.856 on the traveling clock slowed due to relative  speed—we’ll call that motional time dilation—but 03:04.856 --> 03:11.336 time sped up due to its increased distance from  Earth’s center—that’s gravitational time dilation. 03:11.336 --> 03:16.776 The resulting difference was exactly as predicted  by Einstein’s relativity theory. By the way, 03:16.776 --> 03:21.976 relativistic time dilation had already been  measured decades earlier using muons and other 03:21.976 --> 03:27.976 unstable particles, but the Hafele-Keating  showed it directly with atomic clocks. 03:27.976 --> 03:32.856 Modern atomic clocks are so precise that  motional time dilation can in principle be 03:32.856 --> 03:39.176 observed taking an atomic clock for a walk,  and gravitational version is measurable by 03:39.176 --> 03:44.296 moving an atomic clock up by centimeters. As fascinating as these experiments are, 03:44.296 --> 03:49.496 they aren’t measuring anything quantum  about time, even if the mechanism at the 03:49.496 --> 03:54.296 heart of the clock is quantum. Really there  are just precision measurements of a purely 03:54.296 --> 03:58.776 relativistic version of the twin paradox. What we really want to do is some sort of 03:58.776 --> 04:04.296 quantum twin paradox. The quintessential weirdness  of quantum mechanics is superposition—the fact 04:04.296 --> 04:10.376 that quantum systems can occupy multiple states  simultaneously. So what about doing a twin paradox 04:10.376 --> 04:15.176 experiment where, instead of sending identical  twins on paths that have different time flows, 04:15.176 --> 04:21.576 we send the same particle on two separate paths  at the same time by using quantum superposition. 04:21.576 --> 04:25.656 Well, that sounds like the double-slit  experiment but with time dilation. 04:25.656 --> 04:31.656 In the double slit experiment, the “wavefunction”  of a single particle travels through two slits, 04:31.656 --> 04:37.576 recombining on the other side to be detected as a  single particle again at a screen. Although each 04:37.576 --> 04:42.696 particle is detected at a single point, successive  particles build up an interference pattern that 04:42.696 --> 04:47.976 reflects a wave-like passage though both slits.  The interpretation is that each particle travels 04:47.976 --> 04:52.136 through both slits as a probability wave,  and is more likely to be detected where the 04:52.136 --> 04:58.696 two components of this wavefunction stack up or  are “in phase”. In general, clean, double-slit 04:58.696 --> 05:03.976 interference is observed when there are strong  phase correlations between the two paths. A 05:03.976 --> 05:08.296 disruption of that phase correlation blurs and  ultimately erases the interference pattern. 05:08.296 --> 05:12.456 Curiously, this elimination of double-slit  interference seems to precisely track how 05:12.456 --> 05:17.816 much knowledge we can gain about which path  the particle took. If a measurement is strong 05:17.816 --> 05:22.376 enough to tell you with certainty which path was  traveled, it also scrambles phase information 05:22.376 --> 05:27.336 and eliminates the interference pattern.  If the measurement is weaker and leaves 05:27.336 --> 05:33.096 uncertainty as to the path, then the interference  pattern will be blurred rather than destroyed. 05:33.096 --> 05:39.656 So, what happens if one of the double-slit paths  experiences a different amount of time? For 05:39.656 --> 05:43.816 example, we could orient the experiment  so the paths have different altitudes 05:43.816 --> 05:49.416 so that gravitational time dilation slows the  clock of the lower path relative to the upper. 05:49.416 --> 05:55.416 The easiest way to do this is to use a  Mach-Zehnder interferometer—an MZI—rather 05:55.416 --> 05:59.016 than a classic double slit experiment.  In this version, a source of particles 05:59.016 --> 06:03.496 is split into two paths by a beamsplitter and  then the two paths are brought back together 06:03.496 --> 06:07.976 and scrambled by a second beamsplitter.  The “interference pattern” is observed 06:07.976 --> 06:13.336 in the relative probability of the particle  being seen in each of the final detectors. 06:13.336 --> 06:17.256 Those probabilities shift with any  phase shift of the particles along the 06:17.256 --> 06:23.336 paths. And if we do something along those  paths to determine which path was taken, 06:23.336 --> 06:27.816 interference fades or vanishes just like  with the regular double-slit experiment. 06:27.816 --> 06:34.056 So what if the MZI paths are at different  altitudes? Particles should accrue more 06:34.056 --> 06:38.216 age on the upper path compared to the lower path.  Measuring the age of the particle should tell us 06:38.216 --> 06:40.296 which path it took. Or, if a particle travels both  paths simulaneously it should be simultaneously 06:40.296 --> 06:44.856 older and younger. At the very least, checking Seems simple, right? Well, this requires 06:44.856 --> 06:49.416 incredible precision, but it’s “simple”  enough as an experiment that we actually 06:49.416 --> 06:56.056 did it half a century ago. Back in the  70s the Colella–Overhauser–Werner​ or COW 06:56.056 --> 07:00.776 ​experiment sent neutrons through a  MZI-style interferometer and found 07:00.776 --> 07:06.376 that the neutrons taking the elevated path  showed a phase shift you’d expect from the 07:06.376 --> 07:12.136 tiny speedup of time at their higher altitude. On the surface this feels like it can only be 07:12.136 --> 07:16.616 explained with both quantum mechanics and  general relativity, bringing us closer to 07:16.616 --> 07:21.816 a union between the two. And maybe that’s  right. But this isn’t a slam dunk. It turns 07:21.816 --> 07:26.616 out that this effect can be perfectly well  described with quantum mechanics plus good 07:26.616 --> 07:32.696 old Newtonian gravity. If we treat gravity as a  simple force rather than some timey-wimey thing, 07:32.696 --> 07:38.136 then we find that the force induces exactly the  same phase shift in the neutrons. This effect 07:38.136 --> 07:43.576 has been called the “Gravitational Aharonov-Bohm​  Effect,” where the regular Aharonov-Bohm​ effect 07:43.576 --> 07:49.816 is the same thing but for an electric field. So, yeah, relative particle phase shift can 07:49.816 --> 07:54.936 track relative time, and the COW experiment  is consistent with seeing a superposition of 07:54.936 --> 08:00.056 gravitational time dilation this way. But  the interpretation is a bit muddy. But to 08:00.056 --> 08:04.696 really say for sure that we’ve observed  a superposition of different time flows, 08:04.696 --> 08:09.656 we need something closer to an internal clock  for our particle rather than a simple phase. 08:09.656 --> 08:17.256 And that’s exactly what Magdalena Zych and company  proposed in a paper 15 years ago. Actually they 08:17.256 --> 08:22.296 proposed something a bit simpler—an internal  quantum pendulum rather than the full quantum 08:22.296 --> 08:28.216 clock. More precisely, a superposition of two  internal energy states whose relative quantum 08:28.216 --> 08:34.776 phase evolves predictably. They propose that if  gravitational time dilation changes timeflow of 08:34.776 --> 08:39.816 our two paths at different heights, then  the tick-tocking of this quantum pendulum 08:39.816 --> 08:43.976 will get out of sync for the superposition  components of a particle traveling the two 08:43.976 --> 08:48.776 paths. Each component should build up  an internal memory of the path it took, 08:48.776 --> 08:55.016 recorded in the flow of time that it measured. So  when those components are brought back together, 08:55.016 --> 08:59.496 they would “know” which path they took. Now if you remember from the double-slit 08:59.496 --> 09:03.976 experiment, path information disrupts the  appearance of an interference pattern. 09:03.976 --> 09:08.136 Such a disruption in this case could be taken  as evidence of a superposition of different 09:08.136 --> 09:15.336 particle ages—the two superposed clocks got out  of sync, so they have a record of the path taken, 09:15.336 --> 09:19.576 so can no longer interfere perfectly. On  the other hand, if no change is seen in 09:19.576 --> 09:24.856 the interference pattern, then there’s no path  information and it must be the clocks remained in 09:24.856 --> 09:31.096 sync. That might indicate that both wavefunction  components were kicking to the same global clock. 09:31.096 --> 09:37.176 In practice, this quantum pendulum gives an  imperfect measure of the time and so an imperfect 09:37.176 --> 09:42.776 determination of the path traveled. The result is  a blurred interference pattern, but it turns out 09:42.776 --> 09:48.616 that the amount of this blurring can be precisely  determined by the difference in time flow between 09:48.616 --> 09:54.376 the paths. So, time dilation should in principle  be measurable this way and would be less ambiguous 09:54.376 --> 09:59.416 in its interpretation than the COW experiment  which depends on simple phase difference along the 09:59.416 --> 10:05.816 paths. On the other hand, some have argued that  even the internal quantum clock of a Zych-type 10:05.816 --> 10:11.896 experiment can be interpreted in terms of the  phase of those internal degrees of freedom. So, 10:11.896 --> 10:16.216 I dunno, maybe there’s no way to disentangle  the interpretations of phase decoherance vs. 10:16.216 --> 10:22.856 time dilation as a which-path marker. So maybe  the COW experiment got it right 50 years ago. 10:22.856 --> 10:28.296 The Zych et al paper was 15 years ago. And no  one has managed to actually pull it off yet. To 10:28.296 --> 10:33.176 be fair it’s a very hard experiment to do, and  although there are many clever ideas of how, 10:33.176 --> 10:39.736 we haven’t managed to main quantum coherence  on paths separated by enough height and with 10:39.736 --> 10:45.096 enough length for a measurable relative time  difference to build up. Hopefully that experiment 10:45.096 --> 10:51.656 will come along. But actually we may not need it.  There’s a much “easier” way to measure a quantum 10:51.656 --> 10:57.096 superposition of time in an actual quantum clock. In September last year, a new paper came 10:57.096 --> 11:03.416 out by Gabriel Sorci and colleagues in Igor​ ​Pikovski's group, together with the experimental 11:03.416 --> 11:08.776 teams at the national institute for standards  and technology--NIST-- and Colorado State. 11:08.776 --> 11:13.736 This idea brings us back full circle to  the atomic clock that we started with. The 11:13.736 --> 11:20.216 experiment proposed by the Zych team used only  the heart of an atomic clock—a quantum pendulum, 11:20.216 --> 11:25.576 more technically a quantum oscillator. But every clock in history has​ ​the same 11:25.576 --> 11:31.096 basic anatomy: an oscillator and something to  count the oscillations. Like the pendulum of a 11:31.096 --> 11:36.456 grandfather clock and the gear system to tick  over on each swing. Now nature gives us a very 11:36.456 --> 11:42.056 precise pendulum: the electron transition inside  an atom, which in some cases can oscillate between 11:42.056 --> 11:47.496 two states with near perfect regularity. To turn  such a quantum pendulum into an atomic clock, 11:47.496 --> 11:52.936 you take an ion and hit it with a laser pulse that  places it​ ​in a superposition of its ground state 11:52.936 --> 11:57.416 and its excited state. This triggers the quantum  system​ ​to begin evolving in a predictable 11:57.416 --> 12:03.496 fashion, with the phase of the system evolving at  precisely​ ​the transition frequency. By locking 12:03.496 --> 12:10.776 the frequency of a laser to that oscillator, and  then “gearing down” to lower frequency laser, the 12:10.776 --> 12:16.776 oscillations can be counted. The resulting clock  can be so precise that it loses less than a second 12:16.776 --> 12:22.376 in precision over the entire age of the universe. The reason that no one has sent a full atomic 12:22.376 --> 12:27.336 clock through a full double-slit or MZI-like  experiment is that we haven’t figured out 12:27.336 --> 12:32.296 how to create a quantum superposition of  different spatial paths for such complex 12:32.296 --> 12:36.776 systems. But there’s another way to create a  superposition of different rates of time flow 12:36.776 --> 12:42.536 without the spatial separation. Now we’ve been  talking about gravitational time dilation—time 12:42.536 --> 12:47.576 ticks slower when you’re deeper in gravitational  fields. But remember there’s also the motional 12:47.576 --> 12:52.536 time dilation due to relative speed. Now that doesn’t sound immediately useful. 12:52.536 --> 12:57.656 If you try to put an object in a superposition  of two different speeds moving in a straight 12:57.656 --> 13:03.176 line then one will outpace the other and end up  on different spatial trajectories … which we just 13:03.176 --> 13:08.136 figured out was hard to do for an atomic clock. But the idea of Sorci et al. is to put the quantum 13:08.136 --> 13:14.456 oscillator in a superposition of different  internal motional states. Take an atom with an 13:14.456 --> 13:19.336 electron that can move between energy levels  in a way that makes it a good atomic clock. 13:19.336 --> 13:22.456 The atom can have a net charge—it’s an 13:22.456 --> 13:26.536 ion—and that means it can be trapped  in an electric or magnetic field. 13:26.536 --> 13:33.176 Then it can be made to vibrate back and forth  in that EM trap. Its motional states are also 13:33.176 --> 13:38.136 quantized—it can be moving back and forth with  specific frequencies. The higher the frequency, 13:38.136 --> 13:42.776 the faster the motion, and so, in principle,  the more relativistic time dilation. 13:42.776 --> 13:46.296 The critical thing here is that if these  motional states are really quantum, then the 13:46.296 --> 13:52.616 ion can be in a quantum superposition of moving  slower and faster at the same time. Without any 13:52.616 --> 13:58.296 time dilation, this superposition state would be  stable and lead to a regular oscillation in the 13:58.296 --> 14:03.416 interference between the states, which is measured  by the laser. But with motional time dilation, 14:03.416 --> 14:08.456 the superposition of speeds means also a  superposition of rates of time flow. The 14:08.456 --> 14:12.776 “faster” motional state accrues time slower  than the slower motional state, and so they 14:12.776 --> 14:18.296 become increasingly distinguished from each other.  The regular interference pattern is disrupted, 14:18.296 --> 14:23.656 and this is picked up by the coupled laser. In this proposal, the superposition of speeds 14:23.656 --> 14:27.896 of the trapped ion takes the place  of the superposition of paths in the 14:27.896 --> 14:33.016 various interferometer experiments. But the  powerful thing here is that no large spatial 14:33.016 --> 14:38.056 separations are needed. A downside is that  this is a purely special-relativistic effect, 14:38.056 --> 14:42.936 and gravity is irrelevant. So it’s not the  quantum-general relativity link everyone is 14:42.936 --> 14:48.456 desperate for. There’s also debate about how to  interpret the math of this proposal—is the flow 14:48.456 --> 14:54.456 of time changing due to motion, or is the motion  changing the internal mass of the system, which 14:54.456 --> 15:00.296 in turn affects the evolution of that motion. Whatever the interpretational limitations, 15:00.296 --> 15:07.016 being able to actually do an experiment counts for  a lot. And according to the Sorci paper we do have 15:07.016 --> 15:11.976 the measurement precision to do this experiment  now. ​For example, for an Aluminum-ion clock, the 15:11.976 --> 15:19.736 time shift​ ​is of order few parts in ten billion  billion. But the best NIST ion clocks do now have 15:19.736 --> 15:25.176 the precision to measure that discrepancy. This means soon quantum mechanics​ ​may 15:25.176 --> 15:30.376 have to give up its single clock ticking  in the background and instead treat time 15:30.376 --> 15:34.696 as a​ ​quantum property of each branch of  the wavefunction, just as the double-slit 15:34.696 --> 16:00.296 experiment does with space. One step closer  to a unified quantum description of spacetime.