(easy) Nova forward ports + limit unused visibility for unchecked Sparse Matrix functions (#362)

* chore: make unchecked methods private

* update benchmark (#313)

* cleanup unused visibility modifiers (#314)

update doc

* install protoco for lurk-rs

---------

Co-authored-by: Srinath Setty <[email protected]>

Author: huitseeker

.github/workflows/rust.yml

@@ -42,6 +42,7 @@ jobs:
     uses: lurk-lab/ci-workflows/.github/workflows/check-lurk-compiles.yml@main
     with:
       runner: "buildjet-8vcpu-ubuntu-2204"
+      packages: "pkg-config libssl-dev protobuf-compiler libprotobuf-dev"
 
   # Rustfmt, clippy, doctests
   code-quality:

benches/compressed-snark.rs

@@ -27,7 +27,7 @@ type S1 = arecibo::spartan::snark::RelaxedR1CSSNARK<E1, EE1>;
 type S2 = arecibo::spartan::snark::RelaxedR1CSSNARK<E2, EE2>;
 // SNARKs with computation commitmnets
 type SS1 = arecibo::spartan::ppsnark::RelaxedR1CSSNARK<E1, EE1>;
-type SS2 = arecibo::spartan::ppsnark::RelaxedR1CSSNARK<E2, EE2>;
+type SS2 = arecibo::spartan::snark::RelaxedR1CSSNARK<E2, EE2>; // the computation commitment is not used for the trivial circuit
 
 // To run these benchmarks, first download `criterion` with `cargo install cargo-criterion`.
 // Then `cargo criterion --bench compressed-snark`. The results are located in `target/criterion/data/<name-of-benchmark>`.

src/r1cs/sparse.rs

@@ -164,13 +164,13 @@ impl<F: PrimeField> SparseMatrix<F> {
     level = "trace",
     name = "SparseMatrix::multiply_vec_unchecked"
   )]
-  pub fn multiply_vec_unchecked(&self, vector: &[F]) -> Vec<F> {
+  fn multiply_vec_unchecked(&self, vector: &[F]) -> Vec<F> {
     let mut sink: Vec<F> = Vec::with_capacity(self.indptr.len() - 1);
     self.multiply_vec_into_unchecked(vector, &mut sink);
     sink
   }
 
-  pub fn multiply_vec_into_unchecked(&self, vector: &[F], sink: &mut Vec<F>) {
+  fn multiply_vec_into_unchecked(&self, vector: &[F], sink: &mut Vec<F>) {
     self
       .indptr
       .par_windows(2)
@@ -197,7 +197,7 @@ impl<F: PrimeField> SparseMatrix<F> {
     level = "trace",
     name = "SparseMatrix::multiply_vec_unchecked"
   )]
-  pub fn multiply_witness_unchecked(&self, W: &[F], u: &F, X: &[F]) -> Vec<F> {
+  fn multiply_witness_unchecked(&self, W: &[F], u: &F, X: &[F]) -> Vec<F> {
     // preallocate the result vector
     let mut sink = Vec::with_capacity(self.indptr.len() - 1);
     self.multiply_witness_into_unchecked(W, u, X, &mut sink);
@@ -213,7 +213,7 @@ impl<F: PrimeField> SparseMatrix<F> {
 
   /// Multiply by a witness representing a dense vector; uses rayon to parallelize.
   /// This does not check that the shape of the matrix/vector are compatible.
-  pub fn multiply_witness_into_unchecked(&self, W: &[F], u: &F, X: &[F], sink: &mut Vec<F>) {
+  fn multiply_witness_into_unchecked(&self, W: &[F], u: &F, X: &[F], sink: &mut Vec<F>) {
     let num_vars = W.len();
     self
       .indptr

src/spartan/batched.rs

@@ -17,7 +17,7 @@ use super::{
   math::Math,
   polys::{eq::EqPolynomial, multilinear::MultilinearPolynomial},
   powers,
-  snark::batch_eval_prove,
+  snark::batch_eval_reduce,
   sumcheck::SumcheckProof,
   PolyEvalInstance, PolyEvalWitness,
 };
@@ -348,7 +348,7 @@ impl<E: Engine, EE: EvaluationEngineTrait<E>> BatchedRelaxedR1CSSNARKTrait<E>
     };
 
     let (batched_u, batched_w, sc_proof_batch, claims_batch_left) =
-      batch_eval_prove(u_vec, &w_vec, &mut transcript)?;
+      batch_eval_reduce(u_vec, &w_vec, &mut transcript)?;
 
     let eval_arg = EE::prove(
       ck,

src/spartan/polys/power.rs

@@ -27,7 +27,7 @@ impl<Scalar: PrimeField> PowPolynomial<Scalar> {
 
   /// Create powers the following powers of `t`:
   /// [t^{2^0}, t^{2^1}, ..., t^{2^{ell-1}}]
-  pub(in crate::spartan) fn squares(t: &Scalar, ell: usize) -> Vec<Scalar> {
+  pub fn squares(t: &Scalar, ell: usize) -> Vec<Scalar> {
     successors(Some(*t), |p: &Scalar| Some(p.square()))
       .take(ell)
       .collect::<Vec<_>>()

src/spartan/ppsnark.rs

@@ -584,10 +584,11 @@ impl<E: Engine, EE: EvaluationEngineTrait<E>> RelaxedR1CSSNARKTrait<E> for Relax
     let u: PolyEvalInstance<E> =
       PolyEvalInstance::batch(&comm_vec, tau_coords.clone(), &eval_vec, &c);
 
-    // we now need to prove three claims
+    // we now need to prove four claims
     // (1) 0 = \sum_x poly_tau(x) * (poly_Az(x) * poly_Bz(x) - poly_uCz_E(x)), and eval_Az_at_tau + r * eval_Bz_at_tau + r^2 * eval_Cz_at_tau = (Az+r*Bz+r^2*Cz)(tau)
     // (2) eval_Az_at_tau + c * eval_Bz_at_tau + c^2 * eval_Cz_at_tau = \sum_y L_row(y) * (val_A(y) + c * val_B(y) + c^2 * val_C(y)) * L_col(y)
     // (3) L_row(i) = eq(tau, row(i)) and L_col(i) = z(col(i))
+    // (4) Check that the witness polynomial W is well-formed e.g., it is padded with only zeros
     let gamma = transcript.squeeze(b"g")?;
     let r = transcript.squeeze(b"r")?;
 

src/spartan/snark.rs

@@ -250,7 +250,7 @@ impl<E: Engine, EE: EvaluationEngineTrait<E>> RelaxedR1CSSNARKTrait<E> for Relax
     ];
 
     let (batched_u, batched_w, sc_proof_batch, claims_batch_left) =
-      batch_eval_prove(u_vec, &w_vec, &mut transcript)?;
+      batch_eval_reduce(u_vec, &w_vec, &mut transcript)?;
 
     let eval_arg = EE::prove(
       ck,
@@ -414,8 +414,8 @@ impl<E: Engine, EE: EvaluationEngineTrait<E>> RelaxedR1CSSNARKTrait<E> for Relax
   }
 }
 
-/// Proves a batch of polynomial evaluation claims using Sumcheck
-/// reducing them to a single claim at the same point.
+/// Reduces a batch of polynomial evaluation claims using Sumcheck
+/// to a single claim at the same point.
 ///
 /// # Details
 ///
@@ -428,7 +428,7 @@ impl<E: Engine, EE: EvaluationEngineTrait<E>> RelaxedR1CSSNARKTrait<E> for Relax
 ///
 /// We allow the polynomial Pᵢ to have different sizes, by appropriately scaling
 /// the claims and resulting evaluations from Sumcheck.
-pub(in crate::spartan) fn batch_eval_prove<E: Engine>(
+pub(in crate::spartan) fn batch_eval_reduce<E: Engine>(
   u_vec: Vec<PolyEvalInstance<E>>,
   w_vec: &[PolyEvalWitness<E>],
   transcript: &mut E::TE,