Circle starks #924

ColoCarletti · 2024-10-04T14:14:17Z

Circle Starks

This PR introduces the foundational components necessary for the Circle Starks protocol implementation.

Key Features:

Mersenne-31 field: Incorporates the Mersenne-31 field, its quadratic and quartic extensions, along with all required operations.
Circle group: Defines the circle group and implements operations between its elements.
Sets and cosets: Generates sets and cosets to construct the necessary domains within the circle group.
CFFT: Implements the Circle Fast Fourier Transform (CFFT) for both interpolation and evaluation of the bivariate polynomial used in the Circle Starks protocol.

Type of change

New feature

Checklist

Linked to Github Issue
Unit tests added
This change requires new documentation.
- Documentation has been added/updated.
This change is an Optimization
- Benchmarks added/run

…rks into optimize_mersenne31

This reverts commit 21d09c6.

ilitteri

The PR looks good, left some nits, comments, and suggestions. In general, I'd try to remove every change for the code to panic (even if we know for certain that indexing in some index will always be valid).

math/src/circle/polynomial.rs

ilitteri · 2024-10-17T17:04:08Z

math/src/circle/point.rs

+    pub fn zero() -> Self {
+        Self::new(FieldElement::one(), FieldElement::zero()).unwrap()
+    }


This suggestion is not formatted, I recommend not to commit this one.

Note
I'm not sure whether FieldElement::one() or FieldElement::zero() are const (they should though). If they're not, you'd need to hardcode the values to make this function const.

Suggested change

pub fn zero() -> Self {

Self::new(FieldElement::one(), FieldElement::zero()).unwrap()

}

pub const fn zero() -> Self {

CirclePoint { x: FieldElement::one(), y: FieldElement::zero() }

}

This should be a const

Good catch! Just updated my suggestion

FieldElement::one() and FieldElement::zero() aren't const. I understand they should be, but to fix it we have to change it in every field of Lambdaworks. Maybe we can do it in another PR. If we wanted to hardcode the values, we should add them as parameters for every field we wanted to implement the Circle, I don't know if that is what you are suggesting.

ilitteri · 2024-10-17T17:55:43Z

math/src/circle/point.rs

CirclePoint UX and code's readability could be very much improved if we implement:

For CirclePoint addition (this would replace add implementation):

Add<&CirclePoint<F>> for &CirclePoint<F>.

Add<CirclePoint<F>> for CirclePoint<F>.

Add<CirclePoint<F>> for &CirclePoint<F>.

Add<&CirclePoint<F>> for CirclePoint<F>.

AddAssign<&CirclePoint<F>> for &CirclePoint<F>.

AddAssign<CirclePoint<F>> for CirclePoint<F>.

AddAssign<CirclePoint<F>> for &CirclePoint<F>.

AddAssign<&CirclePoint<F>> for CirclePoint<F>.

For CirclePoint scalar multiplication (this would replace scalar_mul implementation):

Mul<u128> for CirclePoint<F>.

Mul<u128> for &CirclePoint<F>.

MulAssign<u128> for CirclePoint<F>.

MulAssign<u128> for &CirclePoint<F>.

Ok! We replaced add for Add for &CirclePoint<F> and scalar_mul for Mul<u128> for CirclePoint<F>. Should we implement the other ones you mentioned here so we can add non-referenced values, multiply referenced values, etc.?

math/src/circle/point.rs

math/src/circle/errors.rs

Co-authored-by: Ivan Litteri <67517699+ilitteri@users.noreply.github.com>

codecov-commenter · 2024-10-24T15:00:52Z

Codecov Report

Attention: Patch coverage is 97.95597% with 13 lines in your changes missing coverage. Please review.

Project coverage is 71.66%. Comparing base (3bb1fa1) to head (fc68bea).

Files with missing lines	Patch %	Lines
math/src/circle/cosets.rs	88.00%	6 Missing ⚠️
math/src/field/fields/mersenne31/extensions.rs	0.00%	6 Missing ⚠️
math/src/circle/twiddles.rs	98.18%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #924      +/-   ##
==========================================
+ Coverage   71.12%   71.66%   +0.53%     
==========================================
  Files         144      149       +5     
  Lines       31880    32516     +636     
==========================================
+ Hits        22676    23302     +626     
- Misses       9204     9214      +10

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

diegokingston · 2024-10-25T13:08:34Z

math/src/circle/polynomial.rs

+/// As a result we obtain the coefficients of the polynomial in the basis: {1, y, x, xy, 2xˆ2 -1, 2xˆ2y-y, 2xˆ3-x, 2xˆ3y-xy,...}
+/// Note that eval has to be a vector of length a power of two 2^n.
+#[cfg(feature = "alloc")]
+pub fn interpolate_cfft(


As an optimization, we could just send the twiddles as input to the function. This is useful when we have to interpolate multiple columns

diegokingston · 2024-10-25T13:09:00Z

math/src/circle/polynomial.rs

+/// returns the evaluation of the polynomial on the points of the standard coset of size 2^n.
+/// Note that coeff has to be a vector with length a power of two 2^n.
+#[cfg(feature = "alloc")]
+pub fn evaluate_cfft(


Similar comment to the one after this one. Twiddles could be given as inputs

diegokingston · 2024-10-25T13:19:04Z

math/src/circle/cfft.rs

+pub fn order_cfft_result_naive(
+    input: &mut [FieldElement<Mersenne31Field>],
+) -> Vec<FieldElement<Mersenne31Field>> {
+    let mut result = Vec::new();


maybe we can use interleave

Nicole and others added 30 commits September 18, 2024 15:37

optimize add

852f856

save changes. Add, sub and mul checked

0d68798

fix tests

5ebc30a

add new inv

81439ae

add mult by powers of two

47cf7da

Merge branch 'optimize_mersenne31' of github.com:lambdaclass/lambdawo…

2acfdd7

…rks into optimize_mersenne31

replace inverse

cfba8bc

test new inv

01eba0d

modify old algorithm for inv

60fd981

fix tests extension

996e224

Merge branch 'optimize_mersenne31' of github.com:lambdaclass/lambdawo…

c41b977

…rks into optimize_mersenne31

add mul for degree 4 extension

28d8b0e

add fp4 isField and isSubField operations and benchmarks

5e1f533

new version for fp4 mul based on the paper

fde7faa

add mul of a fp2e by non-residue

60c8197

change inv using mul_fp2_by_non_resiude

25b8869

save work

1187be8

wip fp2 test

ecff11e

Merge branch 'optimize_mersenne31' of github.com:lambdaclass/lambdawo…

a6169ea

…rks into optimize_mersenne31

add fp2 tests

ae1446a

add 2 * a^2 - 1 function

fa12fc5

use karatsuba in fp4 mul version 1

461485b

clean up

d4e3f40

fix Fp as subfield of Fp2. Tests Fp plus Fp4 is now correct

f0437c7

fix inv

2c5a301

fix comments

4a42dbc

create crate

21d09c6

Revert "create crate"

f7efd90

This reverts commit 21d09c6.

Merge branch 'optimize_mersenne31' into circle-starks

db3ab30

add circle point implementation

eddd9be

fmt

28ab71f

ColoCarletti requested review from igaray, pablodeymo and ilitteri October 17, 2024 14:37

ColoCarletti added 2 commits October 17, 2024 11:47

remove unused functions

e0c666d

add comment

3073470

ilitteri reviewed Oct 17, 2024

View reviewed changes

nicole-graus and others added 10 commits October 23, 2024 11:24

Update math/src/circle/polynomial.rs

37aae9e

Co-authored-by: Ivan Litteri <67517699+ilitteri@users.noreply.github.com>

change generator and order CirclePoint functions as constants

4436e28

impl eq as PartialEq

835a565

implement scalar_mul as Mul

2ea3171

Change error name to a more descriptive one and move it to point.rs

c9acb74

fix lint

93fd515

fix lint

7ae4990

fix some comments

8801317

Merge branch 'main' into circle-starks

7fe6a7d

add alloc::vec

8b2b3d2

ColoCarletti added 4 commits October 24, 2024 12:10

fix no-std

ebc5e9d

cargo fmt

f3028e5

fix no-std

b645a7b

remove macro

c7cfd8f

ColoCarletti requested review from diegokingston, nicole-graus, ilitteri and jrchatruc October 24, 2024 15:32

diegokingston reviewed Oct 25, 2024

View reviewed changes

add comment

fc68bea

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Circle starks #924

Circle starks #924

ColoCarletti commented Oct 4, 2024 •

edited

Loading

ilitteri left a comment

ilitteri Oct 17, 2024 •

edited

Loading

diegokingston Oct 18, 2024

ilitteri Oct 18, 2024

nicole-graus Oct 23, 2024

ilitteri Oct 17, 2024 •

edited

Loading

nicole-graus Oct 23, 2024

ilitteri Oct 24, 2024

codecov-commenter commented Oct 24, 2024 •

edited

Loading

diegokingston Oct 25, 2024

diegokingston Oct 25, 2024

diegokingston Oct 25, 2024

Circle starks #924

Are you sure you want to change the base?

Circle starks #924

Conversation

ColoCarletti commented Oct 4, 2024 • edited Loading

Circle Starks

Key Features:

Type of change

Checklist

ilitteri left a comment

Choose a reason for hiding this comment

ilitteri Oct 17, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ilitteri Oct 17, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov-commenter commented Oct 24, 2024 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ColoCarletti commented Oct 4, 2024 •

edited

Loading

ilitteri Oct 17, 2024 •

edited

Loading

ilitteri Oct 17, 2024 •

edited

Loading

codecov-commenter commented Oct 24, 2024 •

edited

Loading