Logs for the Monero Research Lab Meeting Held on 2019-08-26

Postato da: el00ruobuob / sarang


<sarang> All righty!
<sarang> Let's begin our meeting
<sarang> Agenda: https://github.com/monero-project/meta/issues/385
<sarang> Logs will be posted there after the meeting
<sarang> GREETINGS
<rehrar> hi
<parasew[m]> hello
<suraeNoether> good morning
<sarang> Let's move to ROUNDTABLE
<sarang> suraeNoether: care to go first?
<suraeNoether> sure, last week I made a commit to my buttercup branch with some matching algorithm fixes, which is now operating as far as i can tell 100% smoothly
<sarang> Nice! Looking forward to reviewing it
<suraeNoether> i also have the framework for a markov chain simulator of a basic ledger almost ready to push later today, which we will be using to identify best practices while churning, etc
<suraeNoether> in addition to that, I discovered a discrete distribution over the weekend that could be very helpful in working with output selection
<suraeNoether> that's pretty much it. not tremendously eventful, except the matching algo unit tests
<sarang> Will be neat to see if we can adjust output selection to such a distribution efficiently and robustly
<sarang> Low priority in the grand scheme of things, but intriguing nonetheless
<sarang> I am finishing preparation of the CLSAG paper for submission to Financial Cryptography 2020
<hyc> definitely, will be nice to have some exploration of churning that we can document
<sarang> PR 5707 has been updated to simplify the hashing code
<suraeNoether> oh, sarang and i decided we should also submit the thring signature paper to FinCrypt2020 also, sort of doubling our chances (but not really, it's nonlinear :P)
<sarang> Note that this removes hashToPointSimple and moves its functionality directly into the relevant unit test
<sarang> hashToPointSimple is only used to produce a Pedersen generator, but it's brittle on arbitrary input and not useful elsewhere
<sarang> (unless we updated how we do hash-to-points to use an iterated index, but this is not backwards-compatible)
<sarang> Additional eyes on PR 5707 would be welcome
<sarang> And: Stanford announced the dates and CFP of their next blockchain conference
<sarang> TBH it's probably not worth our time to submit a paper/presentation, given their acceptance rate
<sarang> Unless anyone has an idea for something something zero knowledge ethereum =p
<sarang> I kid, I kid
<suraeNoether> *cough*
<suraeNoether> but not
<hyc> lol
<sarang> Finally, I continue analysis work on proving systems
<suraeNoether> 5707 is for how the second basepoint in our pedersen commitments are selected. if i recall from my conversation with sarang earlier, this PR will make the monero codebase slightly more robust, less brittle, more consistent with the way it works, and more importantly *more easily verified to be correct* by new users, even though it's a relatively small change
<sarang> RingCT3.0 is exploitable in its current form, but I'm told a forthcoming fix likely solves the problem (unclear if provably at this point) with little impact on efficiency
<sarang> suraeNoether: the change only makes it harder to use the wrong hashing method :)
<sarang> and enables slightly more efficient hash-to-point operations under some constructions
<sarang> Anyway, a new preprint (IACR/944) discusses a new proving system for more general constraint systems
<sarang> And there will be a forthcoming preprint with a proving system on pairing groups, which simplifies cleanly to non-pairing groups too
<sarang> I've been looking over that one, since the authors graciously provided an advance copy
<sarang> What will be interesting is whether the Omniring prover relations can be efficiently moved into a more general proving system
<sarang> Making it easier (in theory) to do future changes
<kenshamir[m]> Could you comment on the computational complexity when you simplify the proving system to a non-pairing group?
<sarang> Eventually =p
<sarang> I've only just begun working through that paper
<sarang> FWIW I'm told the preprint is quite imminent
<kenshamir[m]> Ahh alright, any idea when it will be released?
<sarang> ^^
<sarang> Oh and suraeNoether: the primary purpose of 5707 is to speed up MLSAG by removing redundant operations… this led to simplifications of the available hashing operations as a side effect
<suraeNoether> thanks for correcting that! i didn't realize the larger scope (just started going through it)
<sarang> and at that point it seemed like a good idea to remove the brittle hashToPointSimple and put it directly into the unit test that enables verification of how H was derived
<sarang> There are two commits: the second commit moves the hash stuff around
<sarang> the first commit alters the MLSAG code
<suraeNoether> gotcha
<sarang> All righty
<sarang> Does anyone else wish to share interesting research work?
<rehrar> :/ if I were that smart, sure
<sarang> Don't underestimate yourself!
<sarang> While we're at it, we can also do GENERAL QUESTIONS for anyone
<rehrar> well, I did read a recent paper on hormesis as it relates to beetles.
<sarang> lol
<rehrar> my skills are not going to be helpful here :P
<hyc> oh, you just reminded me of that stream of tweets from Sarah Jamie Lewis yesterday, about ants and alarm pheromones
<sarang> Regarding CLSAG, the plan is to submit to several relevant conferences/journals (in sequence) and hope for acceptance
<hyc> the relevance was to illustrating how localized decision making can be gamed / subverted
<sarang> It falls into an awkward spot: it's a neat improvement on earlier work with good security model/proofs, but isn't profoundly new
<sarang> And unfortunately you can't do simultaneous submissions :/
<hyc> (and thus an argument why you need global consensus)
<sarang> The FC deadline is in September, with notification by November 15
<hyc> lots of stuff isn't profoundly new, but the imporvement still is worth talking about
<sarang> Yeah, but that's tough for publication
<hyc> maybe different forums then. heck, chip manufacturers give big presentations on ~5-10% gains
<sarang> Hence doing a fair bit of rewriting for better context
<hyc> cool. yes, expanding the scope of applicability is also a good angle
<sarang> and page limits (grumble grumble stupid wide margins)
<sarang> Aaaaanyway
<sarang> Perhaps on to ACTION ITEMS
<sarang> I have several, and I don't expect to complete them all this week
<sarang> CLSAG revisions, in preparation for submission
<sarang> My doctoral adviser always recommended putting down a paper for a week after your revisions, so you can revisit it with fresh eyes before submission
<sarang> Second is a better understanding of the reduction of this forthcoming proving system to non-pairing groups
<sarang> Third is going over suraeNoether's completed matching code
<sarang> Since suraeNoether had to step away for a few minutes, I assume his action items are applications of his test-passing matching code, and preparation of the thring signature paper for its submission (he can correct me later)
<sarang> Any last comments or questions or information from anyone?
<sarang> OK, we can adjourn! Thanks to everyone for joining in
<sarang> Logs will be posted shortly to the agenda issue

Post tags : Dev Diaries, Cryptography, Monero Research Lab