arXiv subjects
Publications and source records attributed to Leonard Susskind.
This is the written version of a lecture given at KITP in Oct 2014 on Black Holes and quantum complexity. I've included (in boldface) various questions that came up during the lecture and discussions the following day, as well as the quantitative calculations that form the basis of the arguments.
This paper is in three parts: Part 1 explains the relevance of Einstein-Rosen bridges for one-sided black holes. Like their two-sided counterparts, one-sided black holes are connected to ERBs whose growth tracks the increasing complexity of the quantum state. Quantitative solutions for one-sided ERBs are presented in the appendix. Part 2 calls attention to the work of Nielsen and collaborators on the geometry of quantum complexity. This geometric formulation of complexity provides a valuable tool for studying the evolution of complexity for systems such as black holes. Part 3 applies the Nielsen approach to geometrize two related black hole quantum phenomena: the rapid mixing of information through fast-scrambling; and the time dependence of the complexity of precursors, in particular the switchback effect.
We argue that the deconfinement transition of large-N Yang-Mills theory is the condensation of very long and self-intersecting chromo-electric flux strings (QCD string), which is analogous to the formation of a black hole in string theory. We do this by using lattice gauge theory and matrix models. As evidence, we derive an analytic formula for the deconfinement temperature in the strong coupling limit and confirm it numerically. Dual gravity descriptions interpreted in this manner should make it possible to understand the details of the formation of black holes in terms of fundamental strings. We argue that very simple matrix models capture the essence of the formation of black holes.
In this paper we refine a conjecture relating the time-dependent size of an Einstein-Rosen bridge to the computational complexity of the of the dual quantum state. Our refinement states that the complexity is proportional to the spatial volume of the ERB. More precisely, up to an ambiguous numerical coefficient, we propose that the complexity is the regularized volume of the largest codimension one surface crossing the bridge, divided by $G_N l_{AdS}$. We test this conjecture against a wide variety of spherically symmetric shock wave geometries in different dimensions. We find detailed agreement.
In this follow-up to our paper from 2005, we briefly discuss the implications of the apparent detection of B modes in the Cosmic Microwave Background for the issues raised in that paper. We argue that under the assumptions of eternal inflation, there is now stronger support for the detectability of a Coleman-De Luccia bubble nucleation event in our past. In particular, the odds that the spatial curvature of the universe is large enough to be detectable by near future experiments are increased.
In this addendum to [arXiv:1402.5674] two points are discussed. In the first additional evidence is provided for a dual connection between the geometric length of an Einstein-Rosen bridge and the computational complexity of the quantum state of the dual CFT's. The relation between growth of complexity and Page's ``Extreme Cosmic Censorship" principle is also remarked on. The second point involves a gedanken experiment in which Alice measures a complete set of commuting observables at her end of an Einstein-Rosen bridge is discussed. An apparent paradox is resolved by appealing to the properties of GHZ tripartite entanglement.
Computational complexity is essential to understanding the properties of black hole horizons. The problem of Alice creating a firewall behind the horizon of Bob's black hole is a problem of computational complexity. In general we find that while creating firewalls is possible, it is extremely difficult and probably impossible for black holes that form in sudden collapse, and then evaporate. On the other hand if the radiation is bottled up then after an exponentially long period of time firewalls may be common. It is possible that gravity will provide tools to study problems of complexity; especially the range of complexity between scrambling and exponential complexity.
It has been argued that the AMPS paradox implies catastrophic breakdown of the equivalence principle in the neighborhood of a black hole horizon, or even the non-existence of any spacetime at all behind the horizon. Maldacena and the author suggested a different resolution of the paradox based on the close relationship between Einstein-Rosen bridges and Einstein-Podolsky-Rosen entanglement. In this paper the new mechanisms required by the proposal are reviewed: the ER=EPR connection: precursors: timefolds: and the black hole interior as a fault-tolerant, negative information message. Along the way a model of an ADS black hole as a single long-string is explained, and used to clarify the relation between Wilson loops and precursors.
In this paper I return to the question of what kind of perturbations on Alice's side of an Einstein-Rosen bridge can send messages to Bob as he enters the horizon at the other end. By definition "easy" operators do not activate messages and "hard" operators do, but there are no clear criteria to identify the difference between easy and hard. In this paper I argue that the difference is related to the time evolution of a certain measure of computational complexity, associated with the stretched horizon of Alice's black hole. The arguments suggest that the AMPSS commutator argument is more connected with butterflies than with firewalls.
General relativity contains solutions in which two distant black holes are connected through the interior via a wormhole, or Einstein-Rosen bridge. These solutions can be interpreted as maximally entangled states of two black holes that form a complex EPR pair. We suggest that similar bridges might be present for more general entangled states. In the case of entangled black holes one can formulate versions of the AMPS(S) paradoxes and resolve them. This suggests possible resolutions of the firewall paradoxes for more general situations.
Black hole complementarity, as originally formulated in the 1990's by Preskill, 't Hooft, and myself is now being challenged by the Almheiri-Marolf-Polchinski-Sully firewall argument. The AMPS argument relies on an implicit assumption---the ``proximity postulate---which says that the interior of a black hole must be constructed from degrees of freedom that are physically near the black hole. The proximity postulate manifestly contradicts the idea that interior information is redundant with information in Hawking radiation, which is very far from the black hole. AMPS argue that a violation of the proximity postulate would lead to a contradiction in a thought-experiment in which Alice distills the Hawking radiation and brings a bit back to the black hole. According to AMPS the only way to protect against the contradiction is for a firewall to form at the Page time. But the measurement that Alice must make, is of such a fine-grained nature that carrying it out before the black hole evaporates may be impossible. Harlow and Hayden have found evidence that the limits of quantum computation do in fact prevent Alice from carrying out her experiment in less than exponential time. If their conjecture is correct then black hole complementarity may be alive and well. My aim here is to give an overview of the firewall argument, and its basis in the proximity postulate; as well as the counterargument based on computational complexity, as conjectured by Harlow and Hayden.
Black hole complementarity requires that the interior of a black hole be represented by the same degrees of freedom that describe its exterior. Entanglement plays a crucial role in the reconstruction of the interior degrees of freedom. This connection is manifest in "two-sided" eternal black holes. But for real black holes which are formed from collapse there are no second sides. The sense in which horizon entropy is entanglement entropy is much more subtle for one-sided black holes. It involves entanglement between different parts of the near-horizon system. As a one-sided black hole evaporates the entanglement that accounts for its interior degrees of freedom disappears, and is gradually replaced by entanglement with the outgoing Hawking radiation. A principle of "transfer of entanglement" can be formulated. According to the argument of Almheiri, Marolf, Polchinski and Sully, it is when the transfer of entanglement is completed at the Page time, that a firewall replaces the horizon. Alternatives to firewalls may suffer contradictions which are similar to those of time travel. The firewall hypothesis would be similar to Hawking's chronology protection conjecture.
Almheiri, Marolf, Polchinski, and Sully, recently claimed that once a black hole has radiated more than half its initial entropy (the Page time), the horizon is replaced by a "firewall" at which infalling observers burn up, in apparent violation of the equivalence principle and the postulates of black hole complementarity. In this paper I review the arguments for firewalls, and give a slightly different interpretation of them. According to this interpretation the horizon has standard properties, but the singularity is non-standard. The growing entanglement of the black hole with Hawking radiation causes the singularity to migrate toward the horizon, and eventually intersect it at the page time. The resulting collision of the singularity with the horizon leads to the firewall. Complementarity applies to the horizon and not to the singular firewall. Almheiri, Marolf, Polchinski, and Sully conjecture that firewalls form much earlier then the Page time; namely at the scrambling time. I argue that there is no reason to believe this generalization, and good reason to think it is wrong. For most of this paper I will assume that the firewall argument is correct. In the last section before the conclusion I will describe reasons for having reservations.
This paper has been withdrawn because the author no longer believes the firewall argument is correct.
As a result of discussions with Bousso and Vilenkin I want to return to the question of whether the multiverse is past-eternal or if there was a beginning. Not surprisingly, given three people, there were three answers. However, the discussions have led to some common ground. The multiverse being past-eternal, or at least extremely old has content and potential phenomenological implications. I will discuss how the oldness of the multiverse is connected with recent speculations of Douglas.
In this note I respond to Vilenkin's claim that there must have been a beginning.
This is the written version of a lecture at the KITP workshop on Bits, Branes, and Black Holes. In it I describe work with D. Harlow, S. Shenker, D. Stanford which explains how the tree-like structure of eternal inflation, together with the existence of terminal vacua, leads to an arrow-of-time. Conformal symmetry of the dS/CFT type is inconsistent with an arrow-of-time and must be broken. The presence in the landscape of terminal vacua leads to a new kind of attractor called a fractal-flow, which both breaks conformal symmetry, and creates a directional time-asymmetry. This can be seen from both the local or causal-patch viewpoint, and also from the global or multiversal viewpoint. The resulting picture is consistent with the view recently expressed by Bousso. In the last part of the lecture I illustrate how the tree-model can be useful in explaining the value of the cosmological constant and the cosmic coincidence problem. The mechanisms are not new but the description is.
It has been argued that the only mathematically precise quantum descriptions of gravitating systems are from vantage points which allow an unbounded amount of information to be gathered. For an eternally inflating universe that means a hat, i.e., the asymptotic future of a flat FRW universe. The boundary of the hat (the place where it enters the bulk geometry) is the seat of the FRW/CFT duality. In this paper we discuss the perturbative and non-perturbative fixed points of FRW/CFT as seen from the three regions which share this boundary. Perturbatively, there is nothing universal about the FRW duality; there is a separate construction for each possible initial vacuum. We explain how bubble collisions induce a non-perturbative flow to a universal fixed point which contains information about the entire landscape. We also argue that there is a duality between the landscape of de Sitter vacua and a discretuum of very low-dimension relevant operators in the FRW/CFT spectrum. In principle this provides us with a precise definition of unstable de Sitter vacua. Along the way we show that BPS domain walls play a special role in reinforcing the "persistence of memory" and breaking the symmetries of the hat.