Founders

I've come to terms with never finding the perfect shot and by now just go with a pragmatic: it's good enough

Founder, cryptography engineer and stack architect.

https://twitter.com/m2magician

What's tricky is mapping the logic (i.e. the program) we want to prove to a relation with an efficient prover. For the above two examples we have "Lookup arguments" (aptly/confusingly named to prove lookup relations) and sumcheck.

We're entering a phase where provers for special relations are becoming commonplace. E.g. existence of a value in a table is a good example of such a relation. Evaluations of a polynomial over {0,1}^n is another.

Common misconception: ZKP is not always about 'circuits'. Actually, what we are proving are 'relations'. It just happens that most of the early deployed SNARKs were for relations that were best expressed by (arithmetic) circuits. So, the name caught on.

Wake up babe new @dberenzon article just dropped

(Mans known for his category primers)

https://archetype.mirror.xyz/Lov-dI8FOueUt4J4MXPH9gXLyS4VXfHCdEmSg67jzoM

ZKML is all about:
1. Correct Application - fixed model
2. Integrity of Parameters - not tampered with
3. Correct Execution - correct output, given some input and a fixed model

I like to think of bundling 1 & 2 as "Model integrity" (just semantics).

+ honored to have Hungry Cats featured in the landscape!

Marti

@mmagician