feat(ivc): impl cyclefold::step_folding_circuit #369

cyphersnake · 2024-10-21T20:40:33Z

This task integrates all previously implemented parts (e.g., SPS verification and delegated verification) into one cohesive circuit, completing the core logic for StepFoldingCircuit.

Integrates #367 and #368 into one folding circuit.

**Motivation** Part of #369 **Overview** This PR is intended to demonstrate the `Input` design for `cyclefold::StepFoldingCircuit` Note what types are used, what curves are presented and how this data is assumed to be reported inside the circuit

**Motivation** Within #369 and cyclfold in general we need to count consistency markers, for this we need to be able to absorb sfc::input **Overview** Just implementation of `AbsorbInRO` trait for all sfc::input types

**Motivaion** As part of #369 we need impl assigned version of input. Also we have to make sure that {on,off}-circuit assigned works the same way to match the consistency marker **Overview** WIP

**Motivaion** As part of #369 we need impl assigned version of input. Also we have to make sure that {on,off}-circuit assigned works the same way to match the consistency marker **Overview** Implemented versions of Input structures. A test to check the consistency of the {on,off}-circuit hash has been implemented. Also this test allows to double-check correctness of assignment

**Motivation** We need consistent types for #369 between nifs <-> ivc protogalaxy modules **Overview** I had to make many types independent of the curve. Since for the random oracle logic we are byte-by-by-byte distilling from all fields to the target field anyway, I got rid of Scalar -> Base in favor of any field. So now it can be both C::Base (for sangria) and C::Scalar (for cyclefold)

**Motivation** We need consistent types for #369 between nifs <-> ivc protogalaxy modules **Overview** I had to make many types independent of the curve. Since for the random oracle logic we are byte-by-by-byte distilling from all fields to the target field anyway, I got rid of Scalar -> Base in favor of any field. So now it can be both C::Base (for sangria) and C::Scalar (for cyclefold) There are no other functional changes, only generic refactoring

**Motvation** As part of the sfc implementation (#369), we call `protogalaxy::verify_chip` for its own trace, among other things **Overview** Since ivc::protogalaxy & ivc::cyclefold are independent types, we have structures that are independent of each other, yet similar one to one. We need to cast them to an equivalent form (`BigUintPoint`) - [ ] WIP: W_commitment to BigUintPoint in both structure

**Motivation** During the implementation of #369, the types for protogalaxy & cyclefold differed in form but not in substance **Overview** Simple refactoring with removing identical structures and using in ivc::cyclefold types from ivc::protogalaxy

**Motivation** In ramkaz #369 we have to minimize the paired circuit with sangria **Overview** I've reused almost all fold-fn (except `instance`) from the existing sangria implementation

**Motivation** In #369 we have to minimize the paired circuit with sangria **Overview** I've reused almost all fold-fn (except `instance`) from the existing sangria implementation