has this paper been mentioned here yet? eprint.iacr.org/2023/318

would be relevant if we're still using stealth addresses in the long run

hyc: not yet, thanks

tevador: Dalek bulletptoofs was forked into a curve trees impl

<Rucknium[m]> "A simple standard test for..." <- Fishtest uses something called the x^2 test, which is very similar in essence to the chi test I think

(fishtest being some testing framework that has to deal with abnormalities)

Oh, I think it is the exact same, just a different name

Well then, there's a reference implementation (in python). They also have improved upon sprt generalizing into gsprt (well generalized sprt) using mle (maximum likelihood estimation) to, well, estimate unknown variables of the null hypothesis. I think I have an implementation of that in rust sitting around, but not sure if it is finished (I vaguely remember being annoyed I wasn't able to get rid of a lot of clones, even though

it is only 5 doubles in this case lol)

(saying, if you need for some reason a gsprt reference...)

xmrack[m]: i wonder why they chose to define 'untraceability' and 'unlinkability', as opposed to leaning on existing definitions of traceability and linkability, and then measuring against those

ECIP calculations: agorism.dev/uploads/ecip_calculations.ipynb

i also posted it on the github

liam sent me this sage worksheet

midipoet: not sure

ghostway: x^2 is supposed to be ASCII for χ² probably. χ is the Greek letter chi.

Oh, indeed. Never looked at the x too closely lol

Confusing part: By convention, "the" chi-squared test tests whether the frequency of discrete values is equal. However, if we are being 100% correct, there are many different chi-squared tests. Chi-squared is a statistical distribution that results from computing the product of two independent normal random variables. If we have the sum of k such products, it is "chi-squared with k degrees of freedom".

Interesting, I think this was mentioned somewhere I read. Too much time has passed to remember heh

So you can have a statistical test whose _test statistic_ has a chi-squared distribution when the test's null hypothesis is true. That test is not necessarily the "canonical" chi-squared test. So maybe the test you mentioned is "the" chi-squared test. Or many the test statistic just is ch-squared distributed under the null hypothesis.

You also sometimes have "the" F statistic. F is a specific distribution. The same confusion applies.

I'm quite sure what I mentioned is what you call by "the" chi-squared test

If I understand correctly

But with k degrees of freedom

Well that's in the regular test anyway

<UkoeHB> "big new paper analyzing ringct..." <- I reached out, on twitter, to the authors of the security audit and asked if they would join the next MRL meeting on Wednesday to discuss their work and answer any questions. They accepted and will be joining!

Thank you, xmrack . So we will have tx_extra, MRL issue #100 (trustless zk-SNARKs), and RingCT rigorous security proofs next meeting. Good luck, koe

xmrack[m]: nice thanks

xmrack: thanks for reaching out to them! This is very thorough.

UkoeHB: if I may suggest, put tx_extra at the tail end of Wednesday's agenda so that we don't end up in a situation where all the time gets eaten up by it and the authors of the paper won't have the chance to answer questions.

yes I will do that

❤️

xmrack[m]: interesting