177 | Monika Schleier-Smith on Cold Atoms and Emergent Spacetime
What this covers
Experimental physicist Monika Schleier-Smith and host Sean Carroll discuss how cold-atom systems—collections of atoms cooled to near absolute zero and manipulated with lasers—can serve as physical laboratories for studying quantum mechanics and quantum gravity. The conversation bridges three domains: the practical use of entanglement for precision timekeeping and sensing, the construction of controllable quantum systems with non-local interactions, and the possibility of building desktop toy models whose structure mirrors gravitational physics through holographic duality. Schleier-Smith's research centers on the claim that entanglement—not just the Schrödinger equation—is the most distinctive feature of quantum mechanics, and that information can reside in correlations between distant systems rather than in any single location.
The discussion covers how photons can mediate effectively non-local interactions between atoms separated by millimeters (vast distances at atomic scales), enabling researchers to engineer specific entanglement patterns. A key technical achievement is Schleier-Smith's lab construction of a toy model using power-of-two interaction distances that exhibits properties linked to anti-de Sitter space and the fast scrambling conjecture—the idea that information scattered through a quantum system becomes inaccessible at a fundamental rate possibly saturated by black holes. The conversation also addresses why such systems matter practically: entanglement reduces measurement noise below classical limits, improving atomic clocks to the point where millimeter height changes affect their ticking through gravitational redshift. Throughout, Schleier-Smith emphasizes how the [exponential complexity of quantum states](concept:Exponential Complexity of Quantum States) makes a gravitational dual potentially useful—if a geometric description can efficiently predict dynamics where no classical method can.
Schleier-Smith argues that cold-atom systems with programmable, non-local entanglement let experimentalists physically build toy models whose entanglement structure has a dual gravitational (holographic) description, turning abstract quantum-gravity ideas into testable laboratory systems while also serving practical ends like precision metrology.
- Entanglement, not just the Schrödinger equation, is the central and most special feature of quantum mechanics, and information can be stored non-locally in correlations.
- Photons can mediate effectively non-local interactions between distant atoms, allowing engineered entanglement graphs that resemble holographic models.
- Because a full quantum description grows exponentially, a holographic/geometric dual could provide an efficient way to predict and understand strongly entangled systems.
Entanglement structures obey trade-offs and can be engineered for specific tasks like precision timekeeping, sensing, and optimization.
- A superposition of a group of atoms all being in their ground states or all in their excited states is not just a curiosity but a useful resource that, if it can be made, improves precision measurements of time.
“That actually turns out to be a state that is potentially very useful, if you can make it. It's a resource for improving precision measurements of time.”
- Entanglement obeys a rule called monogamy: a system can be maximally entangled with one partner or only weakly entangled with many partners, so there are inherent trade-offs—one can have all-to-all collective entanglement (useful for atomic clocks) or pairwise nearest-neighbor entanglement, and the research goal is to control the structure in between.
“there is sort of a rule called monogamy of entanglement which says, you know, if I can have two things that are kind of maximally entangled or I can have many things that are all sort of weakly entangled. So there are some trade-offs there.”
- A whole class of optimization problems—travelling-salesman-type and certain scheduling problems—can be mapped to minimizing the energy of an interacting spin system, which generically requires spins with programmable non-local interactions whose graph and sign (whether spins align or anti-align, and how strongly) must be controllable; this makes cold-atom systems with programmable non-local interactions a natural platform to explore whether they can efficiently solve classically hard problems.
“there's a whole class of optimization problems that can be mapped to essentially minimizing the energy of an interacting spin system”
- Entanglement helps timekeeping and sensing by introducing correlations so that if one atom's measurement lands one way, another is more likely to land the opposite way, reducing fluctuations in the total count below the standard quantum limit; squeezed states with all-to-all collective entanglement are the simplest well-understood example, and an open question is whether richer entanglement structures provide further benefits for tasks like imaging fields or tracking time-varying signals.
“entanglement can help you introduce correlations that if one coin lands heads, another is more likely to land tails, so to speak, and so the fluctuations in that total count are reduced”
- Atoms are our best timekeepers (the second is defined by an oscillation frequency in cesium), and the best laser-cooled atomic clocks are so precise that quantum projection noise limits them and that a height change as small as a millimeter changes the clock's tick rate via gravitational redshift, letting one resolve effects of general relativity directly in the clock.
“atoms are our best measurements of time, and the second is defined in terms of an oscillation frequency in the cesium atom”
- For unentangled atoms, measurement fluctuations scale with the square root of the number of atoms (the binomial-statistics limit on, e.g., measuring a magnetic field); if a measurement does better than this limit, that operationally demonstrates the presence of entanglement in the system, which is the kind of measurement Schleier-Smith's lab uses to prove entanglement.
“those fluctuations scale roughly with the square root of the number of coins you tossed or in our case, the square root of the number of atoms”
Cold atoms can be scaled to hundreds of individually addressable particles with tunable long-range interactions enabling quantum algorithms and entanglement.
- A recent major advance in cold atoms is scalably creating ordered arrays of hundreds of individually addressable atoms; what is missing from that toolbox is the longer-range connectivity valuable for quantum algorithms and entanglement structures, so an exciting near-term direction is merging scalable single-particle control with control over the graph of interactions.
“the ability to really scalably create arrays of individually tracked atoms, where one can have an ordered array of hundreds of atoms that you can look at one by one”
- Lasers can cool atoms to millionths of a degree above absolute zero by tuning counter-propagating beams slightly below the atomic resonance frequency, so that the Doppler effect makes a moving atom preferentially absorb photons opposing its motion (giving a slowing momentum kick) and then re-emit a slightly higher-energy photon in a random direction—extracting kinetic energy and pumping entropy into the light field while ordering the atoms.
“we tune them so that they're like a little bit too low in energy, or too low in frequency to be on resonance with the atoms”
- Schleier-Smith's toy model replaces nearest-neighbor lattice interactions with interactions at distances of one, two, four, eight sites (any power of two), which allows information to spread exponentially fast across the system: whereas with local interactions the time for information to traverse the system scales linearly with system size, here it scales only with the logarithm of system size.
“you have interactions at a distance of one site, two sites, four sites, eight sites, any power of 2”
- Although physical interactions are always local, the lab can create effectively non-local interactions by using photons to carry information at the speed of light between atoms that are far apart (e.g., a millimeter, which is vast on atomic scales); afterward the photons can be ignored, and the result looks like distant atoms having directly interacted, which is the key mechanism for efficiently building non-local entanglement.
“we can have a photon convey information from one atom to another atom that's a millimeter away or from one little cloud to another cloud that's a millimeter away”
- Generating or manipulating entanglement between distant spins via photon-mediated interactions is distinct from Einstein's 'spooky action at a distance': the latter refers to measurement correlations of an already-entangled pair, whereas creating the entanglement requires information to physically travel from point A to point B, even though the eventual verifying measurements display the spooky correlations.
“this has nothing to do with the spooky action at a distance that Einstein worried about”
A toy model of power-of-two interactions exhibits emergent geometry and fast scrambling, connecting experimentally to holographic duality concepts.
- As a starting point, Schleier-Smith's lab has built a toy model demonstrating an emergent geometry that describes the structure of correlations in the quantum system, with a tunable parameter: at one point it can (to the best of understanding) exhibit fast scrambling, and at another point it shows features resembling an emergent curved-space geometry—though she does not claim it is the holographic dual of a black hole.
“one thing we've done is kind of build what I would say is kind of a toy model for this idea of an emergent geometry that describes something about the structure of the correlations in the quantum system”
- Through a serendipitous chain of contacts, Schleier-Smith's power-of-two toy model turned out, when tweaked slightly, to connect to a version of AdS/CFT holographic duality formulated by Steve Gubser that builds on deep ideas in number theory, allowing the lab to build a system exhibiting an emergent geometry resembling anti-de Sitter space.
“it turned out that this toy model we've been thinking about, if you tweak it a little bit, connects to a version of this AdS/CFT correspondence of holographic duality that Steve formulated, that is, actually really kind of builds on some very deep ideas in number theory”
- Information that falls into a black hole is not lost but very quickly scrambled—hidden in complex quantum correlations and entanglement so that initially local information becomes rapidly delocalized; the fast scrambling conjecture proposes a fundamental limit to how fast this can occur, expected to be saturated by black holes and by systems holographically dual to them.
“our understanding is, no, it's not lost, but it is very quickly scrambled, which is to say it kind of gets hidden in complex quantum correlations and entanglement”
- Building a fast-scrambling system is a necessary but not a sufficient condition for having the holographic dual of a black hole; researchers are developing increasingly refined measurement tests—beyond just how fast information scrambles—to probe whether a given system might have a gravity dual.
“if one can build a system that's a fast scrambler, that's not a sufficient condition for having the holographic dual of a black hole, but it is a necessary condition”
Holographic duality's generality beyond solvable models is testable through experiments on quantum systems suspected to be fast scramblers.
- Although Maldacena's original duality used a very specific theory (super Yang-Mills with much supersymmetry), the key open question is how generic holographic duality is; theorists are often restricted to models they have the tools to analyze, so the experimental contribution is to build systems suspected to be fast scramblers or interesting and learn in regimes that can't easily be analyzed on paper, testing whether the concepts generalize beyond solvable models.
“one of the questions is like how generic are these ideas about holographic duality. And I believe that theorists are often restricted to thinking about models that they have the right tools to analyze. It would be great if the concepts generalize beyond those models.”
- Because a full classical description of a quantum system grows exponentially with particle number (even ~50 particles exceed the largest supercomputer), the value of a holographic geometric dual is that it may provide an efficient, predictive description—using the bulk geometry to predict the quantum system's dynamics or the most efficient way to transfer information between qubits—where no efficient classical representation is currently known.
“that description grows exponentially with the number of particles, and even for 50-some particles that just won't fit on the world's largest super computer”
- A fascinating theoretical feature of holographic duality is a direct connection between entanglement entropy—a measure of entanglement in the boundary quantum system—and geometry in the extra dimension: the entanglement entropy of a region on the boundary is connected to the area of a particular surface in the higher-dimensional bulk space.
“there's this idea that the amount of entanglement entropy in some region on the quantum system that lives on the boundary of this higher dimensional space is connected to the area of a particular surface in the higher dimensional space in the bulk”
Detecting entanglement requires complementary measurements across many copies because individual particles reveal nothing about their entanglement state.
- There is literally no measurement that can be made on a single atom that reveals whether it is entangled with other atoms; in a Bell state each part looks completely random in isolation, so detecting entanglement fundamentally requires preparing the same state many times and performing complementary measurements (including spin measurements in different directions) on multiple parts of the system.
“in the experiment, you always need to basically prepare the same state many times and do a set of measurements that... So that allows you to do kind of complementary measurements”
- Being forced to work only with measurable quantities led the lab to discover that, although they could not measure entanglement entropy directly, there were simpler experimental observables that captured some of the same physics—an insight they had not anticipated from their theoretical plots and that illustrates how confronting the lab clarifies the physics.
“we were sort of surprised that actually by being forced to work with what we could, there were simpler observables that kind of actually showed some of the same physics”
Experimental autonomy and confrontation with measurement realities reveal physics that theoretical analysis alone would not anticipate.
- Quantum mechanics teaching should put entanglement front and center from day one rather than repeatedly solving the Schrödinger equation; in practice students can grasp entanglement and its implications in a couple of weeks and reach the Schrödinger equation only around week seven, which lets them get to the cutting-edge features of quantum mechanics faster.
“you can sort of start talking about entanglement from day one, and certainly really start to get an understanding of its implications in a couple of weeks. And actually at week seven, we get to the Schrödinger equation, but we sort of flip it around”
- Whether quantum systems can efficiently solve classically hard optimization problems is partly settled and partly wide open: there are known cases where certain quantum approaches will not help, cases where they look like they can help, and a huge open space where progress requires playing with the systems experimentally rather than only theorizing.
“There are certain cases where it's known that they won't help, or certain approaches won't help, and certain cases where it looks like the quantum systems can help, and then a huge wide open space where I really think you need to kind of play with the systems and learn from experiments.”
- For the frontier problems Schleier-Smith works on, progress depends on time in individual ~500-square-foot labs and the ingenuity of grad students and postdocs devising clever solutions rather than large funding; she finds problems that money alone could solve less exciting than ones where the path to the answer is genuinely unknown.
“we need time in individual 500 square foot labs and the ingenuity of grad students and post-docs coming up with clever solutions”
- A core advantage of being an experimentalist is autonomy: if you have an idea that merits experimentation, you can do it in your own lab without convincing anyone else, and the data often speak for themselves so you don't have to argue whose idea it was.
“if I can have an idea... and it's an idea that merits experimentation, I don't need to convince somebody else to do the experiment, it's on me”
- The most revolutionary aspect of quantum mechanics is that information does not have to exist locally: through superposition and entanglement, information can be hidden in correlations between systems such that each part looks completely random, yet comparing measurements reveals shared information that exists nowhere in either location individually.
“to me, the most kind of remarkable and revolutionary aspect of quantum mechanics is the concept that information is not something that has to exist locally”
Macroscopic quantum superpositions become exponentially harder to maintain as system size grows because uncontrolled decoherence collapses the state.
- The current state of the art for superposing many spins in a controlled macroscopic-superposition state is around 20 spins, going beyond a single qubit superposition to roughly 20, with the expectation that this number will keep being pushed.
“There are a couple of physical systems where people have done things like this for let's say around 20 spins, right? So that's kind of the state of the art.”
- The difficulty of building macroscopic quantum superpositions (Schrödinger-cat-like states of many atoms) grows with system size because the larger the system, the higher the probability that one atom does something uncontrolled—such as decaying and emitting a photon that, if observable, leaks information about that atom's state and collapses the entire superposition; thus the bigger the system the harder it is to control what you don't know about it.
“the larger I try to make my system, the higher the probability that one atom does something it shouldn't do”