A conversation between Bob Coecke and Stephen Wolfram at the Wolfram Summer School 2021

10,128
2
2021-07-30に共有
Stephen Wolfram plays the role of Salonnière in this new, on-going series of intellectual explorations with special guests. Watch all of the conversations here: wolfr.am/youtube-sw-conversations

Follow us on our official social media channels.

Twitter: twitter.com/WolframResearch/
Facebook: www.facebook.com/wolframresearch/
Instagram: www.instagram.com/wolframresearch/
LinkedIn: www.linkedin.com/company/wolfram-research/

Contribute to the official Wolfram Community: community.wolfram.com/
Stay up-to-date on the latest interest at Wolfram Research through our blog: blog.wolfram.com/
Follow Stephen Wolfram's life, interests, and what makes him tick on his blog: writings.stephenwolfram.com/

コメント (21)
  • @pjmoran42
    There's nothing like this - thanks for setting this up and sharing.
  • To answer your question... quantum mechanics is taught to electrical engineering majors as an introduction into the physics of semiconductors. Chemical engineers also take this course, but from a chemistry view.
  • For reference, the guy who wrote the category theory paper 1945 discussed: https://en.wikipedia.org/wiki/Saunders_Mac_Lane
  • Wonderful conversation, fascinating to listen to. More please! 🤗
  • The question at 2:20:25 and on regarding the programing of distributed computation needs to look at how the concept of Bisimilarity can be used to define the interaction Interface between separate computations. Think of this as one computation generating a Bayesian simulation of itself and other computations simultaneously and then apply ideas such as Aumann's Agreement theorem.
  • Bob didn't mention the work of Aerts & Gabora, which embeds natural language concepts in a Hilbert Space, with some direct analogies to 'collapse' and other QM operations beyond simple word2vec linearity. I am pretty sure Bob studied in their group in Brussels.
  • Ya'll might wish to look at the work of Peter Wegner (Brown University) on Interactive Computation!
  • @47:00 that is what quantum computers are for, in this context. Compositional (tensor) algebra combinatorial explosions Bob is talking about will require QC to handle. You can simulate any quantum system with classical computers, the problem is the run time becomes insanely large very quickly with exponential scale. You need the quantum to compute the quantum.
  • @TEKRific
    That Einstürzende Neubauten pin made me happy. Very interesting conversation. Thanks for sharing!
  • Lakoff & Johnson "Metaphors We Live By" is a crucial source and metaphors are really just categorical diagrams. Someone should do a categorical treatment of their work. It is relatively easy to bootstrap (or ground ) natural language in a small number of physical cases (basis) and process frames. The basis turns out to be the obvious things: space, time, cause, effect, up, down... Then it's metaphors all the way up, such that complex real usage is just an n-category theory over the ground metaphors. A dead metaphor is a layer in the strata that has lost its active metaphorical associations, and become fossilized into the language as a new grounded piece of the bedrock language, but the historical origins can be reconstructed. At the top you get some unique metaphor in Shakespeare with multiple hidden layers of meaning, as the n ..(n-1).. (n-2).. metaphors branch/merge/conflict/reinforce down the stack to basic feelings. An interesting extra overlay for poetic language is provided by the alliteration/homophone/rhyme equivalence relations, which are like wormholes that teleport the listener to different entry points of the meaning space, creating new associations and resonances, e.g. "Made glorious summer by this sun of York."
  • @2:38:00 nice response from Bob. But what Bob's Pictorial QM is all about is basically Quantum Logic, not mechanics. You can easily use it do do some mechanics, but then, as he explains, it is just a different formalism, perhaps one that has more clarity (entanglement is a cup and cap, or in a classical picture wormhole bridges). I love Bob's dismissal of field theorist arrogance. I think it's true, field theory is a complicated way to compute amplitudes, but you can compute amplitudes also using operator algebras. You'd never use PQM do compute scattering amplitudes though (unless Bob tells me otherwise that's my take) unless you are a masochist, the best use-case is in QC, not QD/QM.
  • Interesting to hear the history of how the diagrammatic axiomatisation of composition came about. I very much like how Stephen tries to learn about the history of subject, this is the kind of stuff one rarely finds in papers ... and just tracking back citations never reveals the actual flow of ideas.
  • @1:35:00 one useful way to think about the complications posed for your knowledge answering systems is decision heuristics. Once you get beyond the trivialities ;-) of "satisficing" Diophantine equations solvers and general knowledge stuff, and you want to do more emotional and intentional computation simulation then you might at a minimum need to incorporate hueristics and even irrationality: the way a human might think is, "Oh, I can't do this, let's try and do the exact opposite..." that sort of "irrationality" or "madness" is how people come up with reductio ad absurdem (which probably gets reinvented in different guises every year somewhere).
  • At the 1 hour 11 minute mark in the video, they speak of sequential thinking...My husband can think timelessly, and that can be described as infinite parallel thinking...That kind of thinking is definitely not sequential...
  • @1:50:00 the need for top down is obvious if you broaden your outlook on physics enough. In the Newtonian/computationalist paradigms you need initial/boundary conditions, they are global and top-down but only forward in time, which allows the reductionistic bottom-up (Jaegwon Kim's supervenience/physicalism prejudice[^]) paradigm to hold sway, but it's a false paradigm. With both GR and QM you need two boundaries to form a cobordism, interpolation between is a nondeterministic process in both theories (in QM obviously, in GR because of CTC's when there is nontrivial topology in the block universe picture). I am talking here about completions of the cobordisms, clearly you do not need the BC's at both ends to do field theory or orthodox QM, in fact the whole point of QFT and orthodox QM is they are just calculational tools to deal with only the past boundary data even when assuming that data is complete (maximal). That means virtually any moderately complicated process has to allow for top-down feedback effects, the whole is more than the sum of the parts, you cannot solve a 4D PDE without knowing the (t₁, t₂) boundary conditions, to put it bluntly. It is what delayed-choice experiments and the like are telling you, listen to Bob more carefully and you will notice this confluence of ideas. In GR and QM information is always "leaking across" the Cauchy boundaries so-to-speak (whether they be virtual in the case of George Ellis' weird ideas on a "crystallizing block universe," or whether they be just plain old GR and QM Cauchy boundaries) that is, a single snapshot in time does not determine the cobordism like Newton and Leibniz supposed. Feynman and Hibbs basically said the same, you have to include histories that go way outside the lightcone, if not you cannot get Pauli exclusion and spin-statistics nor compute the amplitudes accurately. This is all so basic (not the computations, the ideas). [^] That is to say physicalism regarding free will. Obviously if we have both bottom-up and top-down causality (at least I believe it is obvious, and I "get it" that most others might not grok this, yet) at play in our universe ten we have a variety of free will that is "nonphysical," by which I just mean there is a sort of Knightian uncertain resource which creatures with enough complexity can exploit in the Cauchy boundary conditions. The determinism of our universe is thus pushed all the way back to the boundaries/IVC's, and must_thus be beyond spacetime: whether spacetime is infinite in extent or not. It is possible for us to obtain some small knowledge about the BC/IVT's of our universe, but impossible based on everything we know about physics to have any physical process set the BC's. This applies to cyclical universes, CCC universe, Big Bang singularity tunnel inflation origins, the whole lot. At least it applies if our ideas about preservation of information are correct, and as Susskind has noted, they are the most fundamental thing we know. If you abandon those principles you do not really have science any more, not even if you are Paul Feyerbend.
  • @1:17:00 I think you've got that completely backwards Stephen. Our ability to find/see/imagine meaning is what gives us capacity to understand and write down computations, not the other way around. I'm a bit old fashioned I suppose, I side with Gödel, our thought capacity/consciousness/soul whatever you care to name it, is surely beyond computation, it is something else, I do not know what, but I'd guess if science ever figures out consciousness it will turn out to be non-computational in some way. The orthodox view that consciousness is "nothing but computation" is a prejudice of our times, understandably so, but nevertheless clearly a prejudice (to a dude with a hammer everything is a nail type of thing). Penrose's whole argument on this might not be correct, but formally it is not wrong either (to deny a logical argument you only need reject a single premise, and you only need reject it because you don't like it, no one else will then be able to convince you the argument was in fact correct and you choose to reject a true premise). I am almost sure I cannot convince a person who believes "everything is computation" that they are wrong, they'll just shift goalposts and include as "computation" anything new that falls outside their previous definition of computation. That's the way Seth Lloyd seems to operate. But then I'd argue you have no meaningful theory. You have to define what computation is and what it is not. Then you can say meaningful things. As well as Gödel, I tend to have sympathy for David Chalmers too, at least regarding some of the problems of ineffability of mental qualia. No one on Earth has a proper computational explanation for mental qualia. All those Psi-Theory folks are pure research grant grifters who do not understand category errors! (Prove me wrong!)