You're reading the public-facing archive of the Category Theory Zulip server.
To join the server you need an invite. Anybody can get an invite by contacting Matteo Capucci at name dot surname at gmail dot com.
For all things related to this archive refer to the same person.
Paolo Perrone (Jun 04 2020 at 19:10):Hey all,
This is the discussion thread of Rob Spekkens' talk, "Disentangling inference and influence in classical and quantum theories".
The talk, besides being on Zoom, is livestreamed here: https://youtu.be/zn7k15sq_TE
Paolo Perrone (Jun 04 2020 at 19:24):Date and time: Saturday, 6 Jun, 12h UTC.
Paolo Perrone (Jun 06 2020 at 11:40):Hi all! We start in 20 minutes.
Arthur Parzygnat (Jun 06 2020 at 12:59):That was very interesting, thank you! I was very curious about something from your earlier slides but I felt it was a bit too technical to ask over zoom. With regards to the EPR experiment you mentioned, you had a slide with analogues of Bayesian inference in terms of density matrices and certain star products. So in EPR, if one observer sees spin up, is the update density matrix based on your formalism on the right side of that slide the density matrix for spin down for the other observer? If so, could you please point me to which paper discusses this (and more precisely where)?
Joshua Meyers (Jun 06 2020 at 13:00):Great talk @Rob Spekkens! Do you have any idea how a quantum C-I formalism could describe the Bell experiment which you started the talk with? You said that you wanted a description which was causally conservative and inferentially radical, but you also said that there was experimental evidence contradicting the particles having a common cause: doesn't this mean it couldn't be causally conservative?
Joshua Meyers (Jun 06 2020 at 13:00):(I tried to ask this question at the end but my microphone wasn't working :()
Javier Prieto (Jun 06 2020 at 13:32):You said that you wanted a description which was causally conservative and inferentially radical, but you also said that there was experimental evidence contradicting the particles having a common cause: doesn't this mean it couldn't be causally conservative?
I was thinking the same during the talk. My hunch is that "causal conservatism" will be something very similar to local hidden variables.
Arthur Parzygnat (Jun 06 2020 at 13:32):@Joshua Meyers (btw, hi Joshua!) I was also confused about that statement. If I understood what Rob said, then I think he said a classical model that was conservative and inferential (as opposed to a quantum one). If one uses a quantum model with a common cause (such as the EPR state), then you can reproduce statistics that do not coincide with the classical statistics. Right?
Rob Spekkens (Jun 06 2020 at 13:33):Arthur Parzygnat said:
With regards to the EPR experiment you mentioned, you had a slide with analogues of Bayesian inference in terms of density matrices and certain star products. So in EPR, if one observer sees spin up, is the update density matrix based on your formalism on the right side of that slide the density matrix for spin down for the other observer? If so, could you please point me to which paper discusses this (and more precisely where)?
Hi Arthur, thanks for your question. The answer to your question is yes. The paper that presents this is the one I mentioned near the beginning of my talk:
M. S. Leifer and Robert W. Spekkens
Towards a formulation of quantum theory as a causally neutral theory of Bayesian inference
Phys. Rev. A 88, 052130 (2013)
[arXiv:1107.5849 (quant-ph)]
See Sec. V.3 for an analysis of the EPR experiment. In particular, see proposition V.1 and the discussion that follows it for the proof that our formalism reproduces the standard quantum remote state-update rule.
Rob Spekkens (Jun 06 2020 at 13:47):Joshua Meyers said:
Do you have any idea how a quantum C-I formalism could describe the Bell experiment which you started the talk with? You said that you wanted a description which was causally conservative and inferentially radical, but you also said that there was experimental evidence contradicting the particles having a common cause: doesn't this mean it couldn't be causally conservative?
Hi Joshua, sorry that I wasn't clear about that. What Bell's theorem shows is that one cannot be both causally conservative and inferentially conservative. That is, one cannot explain Bell inequality violations using a common cause structure if one presumes that a classical theory of inference dictates how one ought to propagate beliefs (that is, how one uses what one learns about the outcome at one side of the experiment to update what one knows about the unobserved common cause and thereby update what one knows about the outcome at the other side) . The research program I described aims to salvage a common-cause explanation (i.e., a causally conservative explanation) by presuming that the common cause is not a classical variable but a more exotic object, and that propagating beliefs through such a common cause is dictated by a novel calculus of inference.
You can read more about the "causal gloss" on Bell's theorem in this article:
Christopher J. Wood and Robert W. Spekkens
The lesson of causal discovery algorithms for quantum correlations: Causal explanations of Bell-inequality violations require fine-tuning
New J. Phys. 17, 033002 (2015)
[arXiv:1208.4119 (quant-ph)]
Javier Prieto (Jun 06 2020 at 14:04):Rob Spekkens said:
The research program I described aims to salvage a common-cause explanation (i.e., a causally conservative explanation) by presuming that the common cause is not a classical variable but a more exotic object, and that propagating beliefs through such a common cause is dictated by a novel calculus of inference.
I wonder whether this novel calculus of inference will be related to the (admittedly strange) update rule in the qplex and if so how.
Simon Burton (Jun 06 2020 at 15:08):I highly recommend watching this lecture https://www.youtube.com/watch?v=n8NRSPCekmI , which covers much of the same material presented today.
Paolo Perrone (Jun 06 2020 at 21:44):Hey all! Here's the video.
https://youtu.be/OkeAQAqx5MY
Arthur Parzygnat (Jun 07 2020 at 19:20):Rob Spekkens said:
Arthur Parzygnat said:
With regards to the EPR experiment you mentioned, you had a slide with analogues of Bayesian inference in terms of density matrices and certain star products. So in EPR, if one observer sees spin up, is the update density matrix based on your formalism on the right side of that slide the density matrix for spin down for the other observer? If so, could you please point me to which paper discusses this (and more precisely where)?
Hi Arthur, thanks for your question. The answer to your question is yes. The paper that presents this is the one I mentioned near the beginning of my talk:
M. S. Leifer and Robert W. Spekkens
Towards a formulation of quantum theory as a causally neutral theory of Bayesian inference
Phys. Rev. A 88, 052130 (2013)
[arXiv:1107.5849 (quant-ph)]See Sec. V.3 for an analysis of the EPR experiment. In particular, see proposition V.1 and the discussion that follows it for the proof that our formalism reproduces the standard quantum remote state-update rule.
Thanks! I finally got around to checking the definitions and verifying that for EPR it indeed reproduces the expected correlations. I read your paper with Leifer about a year ago, but I did not quite understand how the belief propagation worked because I (incorrectly) thought you were only using the distortion map (I think I already forgot what you call it, but I'm referring to the recovery map), and I wasn't able to reproduce the correlations with that method.
Rob Spekkens (Sep 08 2020 at 16:54):An article describing the research reported in this talk is finally out:
https://arxiv.org/pdf/2009.03297.pdf
"Unscrambling the omelette of causation and inference: The framework of causal-inferential theories" with David Schmid and John Selby