Blog

With Chandramouli and Olga: “A physical protocol for observers near the boundary to obtain bulk information in quantum gravity”.

https://arxiv.org/abs/2008.01740

Think of observers who live near the boundary of global AdS and can only explore the spacetime on a single Cauchy slice at large values of the radial coordinate. We put the theory in a low-energy state, and ask these observers to glean as much information they can about the state, without ever directly entering the bulk. How much information can they get?

In a local quantum field theory, the answer is obviously “nothing”. The observers are spatially separated from the region in the middle of AdS, and so operators there commute with all observables accessible to the observers.

It turns out that when gravity is turned on, the surprising answer is “everything”! We develop a protocol, which involves only simple manipulations of the state, and simple measurements from the near-boundary region, and allows the observers to completely determine the state of the bulk.

This is a perturbative check of the idea that signatures of holography are already implicit in semiclassical gravity. It is also closely related to one of the lessons that we learn from a study of black-hole evaporation: information about the interior is also available in a scrambled form in the exterior. This has been discussed for several years — most recently in all the work on “islands” — and so what we have here is a controlled example, where a very similar effect can be studied precisely, and directly in a Lorentzian setting.

Another implication is for quantum information measures. If the von Neumann entropy is understood physically as a measure of lack-of-information, then our result suggests that the von Neumann entropy of an “annular region” (Fig 1b) between r = r0 and r = \infty is zero in a theory of gravity. This is very different from the local QFT result, which would be proportional to r0^(d-1). (Remember, this region is not the entanglement wedge of any subregion on the boundary.)

There has been a lot of recent discussion of the entanglement entropy of regions in theories of gravity, but I think adequate thought isn’t being given to what this quantity means. The idea seems to be do a calculation where gravity switches off exactly at some point, and then just claim that the same result should hold in the presence of dynamical gravity because the corrections must be “small” when gravity is weak. But the fine-grained entropy is not a perturbative quantity, and so the idea that “weak gravity = no gravity” is just wrong in some cases, as this example shows!

If you have questions, comments or, especially, criticism, please email.

I will be part of some panel discussions tomorrow (6 August).

1. The first is organized by the All India Peace and Solidarity Organization (MP). The topic is “75वें हिरोशिमा दिवस पर वेब-चर्चा” and I will speak about the “Iran nuclear conflict and U.S. imperialism”. Achin Vanaik will speak about the situation in South Asia. This event is from 3 pm – 5 pm at https://meet.google.com/rgo-bytk-key. This meeting will largely be in Hindi.

2. The second is organized by the India Peace Centre. Achin will be a panelist here as well, and we will also be joined by Vaishali Patil. The title is “No Nukes: No More Hiroshima”. This event is from 5 pm – 7 pm at https://zoom.us/j/7122556952?pwd=QjYvbFZVT3U3d3ZCOHRac1FlQUpSQT09 I think the language will be a mix of English and Hindi. I am planning to speak about “Challenges and opportunities for the peace movement.” [update: recorded video at: https://www.facebook.com/indiapeacecentre.nagpur/videos/3181889788559884/]

3. The third is organized by India March for Science (Karnataka). Here the discussion is about issues of nuclear energy and its relevance (or lack thereof?) in India. The other panelist is George Joseph, and the event will be moderated by Anil Jagalur. This event is from 7 pm – 8:30 pm. You can join at https://www.facebook.com/watch/?v=1007791319677011 The language will be English.

Please join if you can and spread the word to others who might be interested.

A talk that I gave at the Strings 2020 conference: https://www.youtube.com/watch?v=u73rsTpnPrY Since the Strings-conference audience is very broad, it was intended for people who are not directly thinking about the same questions, and provides a quick summary of some of our recent work. So the video may be useful.

We had a long panel discussion on Black Hole information later, which capped off the conference. But, I should mention that we also discussed similar issues at the BH microstructure conference last month with some of the same panelists. (Bena, de Boer, Maldacena, Mathur and me.) The emphasis was slightly different because the question there was “Do typical states have horizons?” but I think that discussion was also interesting. It is available here: https://youtu.be/31f_2G5DLaw?t=11367

About a month ago, the government conducted a press-conference reporting the results of various “studies”, each claiming that the lockdown had saved thousands of lives. A few days later, some independent scientists also decided to chime in with their own predictions that broadly supported this idea. Even though a few weeks have passed, I think these claims remain relevant, as another round of lockdowns is contemplated in various parts of the country.

So I decided to examine these claims in a little more detail and found that they were very misleading. In fact if the compartmental models used in these studies teach us anything, it is that lockdowns of the kind witnessed in India — which are imposed early in the epidemic but fail to quash it — may delay fatalities but do not avert them. So, within these models, the efficacy of a lockdown cannot be gauged by simply comparing figures for fatalities at the end of the lockdown with the projected figure for fatalities at the same date without the lockdown.

These models do not account for several important real-world effects: the lockdown has caused a humanitarian crisis, it has restricted access to essential healthcare for other conditions and, to a limited extent, it has also been used to ramp up testing and tracing. Moreover, the lockdown has engendered economic insecurity, and in the longer-run, this makes it harder for people to maintain physical-distancing norms. All these effects are more important than the transient direct effect of the lockdown on fatalities. But given the paucity of currently available data, none of them can be quantified. So epidemiological models are simply not in a position to accurately quantify the impact of the lockdown on fatalities.

If this is the case, what motivates these “studies”? Perhaps the independent groups inadvertently fell into a trap set by the government. But, as for the government-backed studies themselves, the answer seems to be clear. There is widespread concern that the lockdown has failed. So, rather than address these concerns, it is “useful to point to “advanced scientific studies” that “show” that the lockdown has saved thousands of lives. Since people tend to get intimidated by differential equations, this helps to shut down critics.”

I discussed some of this in an explainer for the Scroll today:

https://scroll.in/article/965814/what-do-mathematical-models-tell-us-about-the-impact-of-lockdown-on-covid-19-fatalities-in-india

I also discuss what these “studies” do in practice, how they get these numbers, why it doesn’t make sense, and some broader concerns about models.

This is a summary of a more technical note (written for internal discussions in the Indian Scientists Response to COVID-19 group) that is available publicly here: https://zenodo.org/record/3901516

Both the technical note and the explainer make very simple points. I only felt compelled to write them in light of the claims above, which have received wide publicity.

The government would like to claim that the lockdown has “saved lives”. But in this article for the Wire, I tried to explain why it makes no sense to gauge the efficacy of the lockdown by blindly comparing the current number of cases to the number obtained by extrapolating the early rates of growth. The final toll of the pandemic is almost entirely controlled by its rate of spread after the lockdown ends, not its rate of spread during the lockdown. So the correct question is whether the lockdown has put the country in a position where it can keep this rate of spread low after the lockdown ends. The data suggests otherwise. A simple calculation comparing the number of recorded deaths to recorded cases suggests that a large number of active cases remain undetected, and so contact-tracing cannot succeed. Most seriously, the long-term rate of spread is controlled by whether people can sustain physical-distancing measures and other precautions. But, because the welfare of people was relegated to an afterthought, the lockdown has made it harder for people to implement precautionary measures, which might otherwise have been manageable. https://thewire.in/government/indias-lockdown-has-failed-heres-what-we-can-learn-from-it

Abolition 2000, which is a global network of nuclear disarmament groups asked 9 disarmament-campaigners across the world to record short videos in advance of its AGM that takes place on the weekend. The question we were asked to address was a broad one about how we saw the world changing by the time of the NPT Review Conference next year and how we felt this would affect our work. So rather than addressing topical issues, which we do often, I took the opportunity to talk about some of the broader issues that frame our work and spoke about “The Pandemic, Hegemony and Disarmament.” Here is the video. https://www.youtube.com/watch?v=jGOnPFLJz5A and here is a link to the other videos: https://www.youtube.com/playlist?list=PLNLSpPdpFraDiSCjFejKiVk7WNgIQRImq

An op-ed I wrote for the Hindu on the importance of accountability during a crisis. https://www.thehindu.com/opinion/op-ed/lacking-in-transparency/article31447809.ece

A virtual seminar I gave at the Albert Einstein Institute at Potsdam a few days ago https://www.youtube.com/watch?v=RlZlj6_Y92s

A few comments on the topic of virtual seminars, which have become the norm over the past few weeks. Michal Heller invited me to give this seminar as a virtual seminar last year, well before COVID surfaced. The AEI group was ahead of the times and they were organizing virtual seminars as standard practice. I must admit that, at the time, I was slightly skeptical :-)

This is because, virtual seminars I’ve organized and attended in pre-lockdown days have usually involved a virtual speaker but a physical audience. This certainly doesn’t work as well as a physical speaker. For instance, it is not easy to ask a question from the audience. And there is information asymmetry. The speaker can’t see the audience too clearly but the audience-members can see each other. This matters a surprising amount because the physical audience has collective dynamics that the speaker can’t respond to. For instance, if one person looks skeptical or bored, others in the audience notice but the speaker can’t take note and modulate the talk.

But, things are completely different if everyone is in front of an individual screen. In many respects this works as well as a physical talk and, in some ways, even better: the slides are certainly clearer. And, of course, a significant advantage is that people from all over the globe can attend on an equal footing. This talk, for instance, had people from Japan, India, Europe, U.S, Argentina… There is even scope for informal discussion at the end of the talk if a few people remain behind. (And yet another advantage is that if the talk is boring, you can switch to doing other things without being rude and “leaving” the virtual room :-) )

I think there is something to be said, quite seriously, for sticking to this mode of organizing seminars and conferences even when COVID is gone.