Logs for the Monero Research Lab Meeting Held on 2019-04-15

Posted by: el00ruobuob / sarang


<sarang> Let's begin our meeting!
<sarang> 1. GREETINGS
<sarang> Hi
<[-mugatu-]> o/
<sarang> Quiet day today…
<sarang> But I suppose we can still move to 2. ROUNDTABLE
<sarang> suraeNoether: care to go first today?
<suraeNoether> sure, my cat is missing and i want to go look for her, so i'm going to make this quick
<sarang> :(
<sarang> Understood
<suraeNoether> CLSAG signatures are fast and small, they are so fast and small that my naive colored-coin approach could support two assets and still be faster and smaller than our present MLSAG scheme
<suraeNoether> i'm not recommending coloring monero, but commenting on overall speed, it's nuts
<suraeNoether> however, as sarang mentioned, there is a key image problem i'm looking into
<suraeNoether> it's possible rectifying them will cost us some of those gains
<sarang> Yeah, I don't think a straightforward LSAG reduction works here
<suraeNoether> in the meantime, i'm handing CLSAG off to sarang for at least 7 days so i can focus on MRL11
<sarang> But I wonder if a redefinition of the security requirements to accommodate the new linking will be sufficient
<suraeNoether> since we're rounding the corner on that, it's my top priority, and i want to get CLSAG out of sight for a few days
<sarang> righto
<suraeNoether> okay, i'll be back later today, hopefully with suraecat
<sarang> Thanks, and best of luck with your search
<sarang> I completed the building blocks for a simple Lelantus transaction flow (insecure example code in agenda)
<sarang> and am in contact with the paper's author to discuss some privacy aspects of the construction
<sarang> the CLSAG example code has been updated to reflect some changes
<sarang> and, as suraeNoether said, still working on proper formalization, which is trickier than expected
<sarang> The output selection algorithm discussed here still has an open PR from moneromooo that needs eyeballs
<sarang> PR 5389
<needmoney90> Hi
<sarang> yo
<needmoney90> Will lurk mostly.
<sarang> Does anyone else have research to share?
<needmoney90> Just announcing presence
<sarang> Otherwise we can keep waxing poetic about CLSAG definitions
<sgp_> here now
<sarang> :/
<sarang> Hi suraeNoether
<sarang> sgp_:
<sarang> bah, silly autocomplete
<moneromooo> I have these multi user txes going in the background, and I am wondering whether the 'a' values can be reuesd for multiple outputs.
<sarang> Remind me what these values are/
<needmoney90> What's the status on M-of-N multisig?
<moneromooo> The idea is to make 16 actual outs for the "same" logical output, so they get shuffled as new outputs are added.
<sarang> (our notation is often inconsistent)
<moneromooo> And I don't know whether it's safe to keep those. I assume sharing them with other usesr of the same tx is not good.
<dEBRUYNE> <suraeNoether> however, as sarang mentioned, there is a key image problem i'm looking into <= This is referring to CSLAG right?
<sarang> dEBRUYNE: yes
<sarang> there are no such issues with MLSAG
<dEBRUYNE> All right, thanks for clarifying
<sarang> The problem refers to the fact that trying to reduce CLSAG to LSAG with an aggregated key yields the wrong key image
<moneromooo> a is the random secret keys generated at proive time to create the pseudoOuts.
<sarang> Hmm ok
<sarang> You asked me to review this earlier, and it completely slipped my mind
<sarang> I'll look for the code snippet you sent in PM
<sarang> to ensure I don't get wrong the terms you're referring to
<moneromooo> ty
<sarang> sgp_: did you have something you wished to discuss too?
<sgp_> I don't believe so
<sarang> Well, this meeting is turning out to be quite short :D
<sarang> moneromooo: anything specific, aside from the reuse question you posed?
<sarang> (to discuss here, I mean)
<moneromooo> Not at the moment I think.
<sarang> OK, I suppose we can move right along then
<sarang> to 3. QUESTIONS and 4. ACTION ITEMS
<sarang> While suraeNoether continues working on matching/churn via MRL-0011, I have several things for the week
<sarang> Now that CLSAG reduction to LSAG is proving so problematic, I want to see if definition modifications for the LSAG proofs will suffice for our use case
<sarang> I'll be checking on moneromooo's question shortly (apologies for letting that slip by)
<sarang> as well as more work on Lelantus transaction flows
<dEBRUYNE> sarang: Have you consulted RandomRun regarding this problem btw?
<sarang> suraeNoether and I have been in contact with him throughout the development process
<sarang> I don't believe this problem has practical effects on CLSAG's security, only in the complexity of the formalization
<sarang> Any other questions or action items on people's minds?
<dEBRUYNE> I see, so it does not render the scheme infeasible?
<sarang> Heh, depends
<sarang> If we end up not being able to prove secure under proper definitions, that's a bit of a quandry
<sarang> but in the worst case, we decide not to adopt the scheme, and are right back to where we are now
<dEBRUYNE> True, better safe than sorry I guess :p
<sarang> However, the space and time savings are so compelling that it's worth the effort
<sarang> ~25% space savings and 15% time savings for a 2-2 transaction
<sarang> (in the signature portion)
<needmoney90> Not bad
<sarang> OK, any last questions before we formally adjourn?
<sarang> Righto, in that case, we are adjourned. Discussion can of course continue
<sarang> Logs will be posted shortly to the GitHub agenda issue

Post tags : Dev Diaries, Cryptography, Monero Research Lab