Add ChunkedPippenger to variable-base MSM

CHANGELOG.md

@@ -19,6 +19,7 @@
 - [\#301](https://github.com/arkworks-rs/algebra/pull/301) (ark-ec) Add `GLVParameters` trait definition.
 - [\#312](https://github.com/arkworks-rs/algebra/pull/312) (ark-ec) Add `is_in_correct_subgroup_assuming_on_curve` for all `SWModelParameters`.
 - [\#348](https://github.com/arkworks-rs/algebra/pull/348) (ark-ec) Add `msm:{Fixed,Variable}Base:msm_checked_len`.
+- [\#364](https://github.com/arkworks-rs/algebra/pull/364) (ark-ec) Add `ChunkedPippenger` to variable-base MSM.
 
 ### Improvements
 

README.md

@@ -1,6 +1,6 @@
-<h1 align="center">arkworks::algebra</h1>
+<h1 style="text-align: center;">arkworks::algebra</h1>
 
-<p align="center">
+<p style="text-align: center;">
     <img src="https://github.com/arkworks-rs/algebra/workflows/CI/badge.svg?branch=master">
     <a href="https://github.com/arkworks-rs/algebra/blob/master/LICENSE-APACHE"><img src="https://img.shields.io/badge/license-APACHE-blue.svg"></a>
     <a href="https://github.com/arkworks-rs/algebra/blob/master/LICENSE-MIT"><img src="https://img.shields.io/badge/license-MIT-blue.svg"></a>

ec/src/msm/variable_base.rs → ec/src/msm/variable_base/mod.rs

@@ -6,6 +6,9 @@ use crate::{AffineCurve, ProjectiveCurve};
 #[cfg(feature = "parallel")]
 use rayon::prelude::*;
 
+pub mod stream_pippenger;
+pub use stream_pippenger::*;
+
 pub struct VariableBase;
 
 impl VariableBase {

ec/src/msm/variable_base/stream_pippenger.rs

@@ -0,0 +1,67 @@
+//! A space-efficient implementation of Pippenger's algorithm.
+//!
+use crate::AffineCurve;
+use ark_ff::{PrimeField, Zero};
+
+use ark_std::{borrow::Borrow, ops::AddAssign, vec::Vec};
+
+/// Struct for the chunked Pippenger algorithm.
+pub struct ChunkedPippenger<G: AffineCurve> {
+    pub scalars_buffer: Vec<<G::ScalarField as PrimeField>::BigInt>,
+    pub bases_buffer: Vec<G>,
+    pub result: G::Projective,
+    pub buf_size: usize,
+}
+
+impl<G: AffineCurve> ChunkedPippenger<G> {
+    /// Initialize a chunked Pippenger instance with default parameters.
+    pub fn new(max_msm_buffer: usize) -> Self {
+        Self {
+            scalars_buffer: Vec::with_capacity(max_msm_buffer),
+            bases_buffer: Vec::with_capacity(max_msm_buffer),
+            result: G::Projective::zero(),
+            buf_size: max_msm_buffer,
+        }
+    }
+
+    /// Initialize a chunked Pippenger instance with the given buffer size.
+    pub fn with_size(buf_size: usize) -> Self {
+        Self {
+            scalars_buffer: Vec::with_capacity(buf_size),
+            bases_buffer: Vec::with_capacity(buf_size),
+            result: G::Projective::zero(),
+            buf_size,
+        }
+    }
+
+    /// Add a new (base, scalar) pair into the instance.
+    #[inline(always)]
+    pub fn add<B, S>(&mut self, base: B, scalar: S)
+    where
+        B: Borrow<G>,
+        S: Borrow<<G::ScalarField as PrimeField>::BigInt>,
+    {
+        self.scalars_buffer.push(*scalar.borrow());
+        self.bases_buffer.push(*base.borrow());
+        if self.scalars_buffer.len() == self.buf_size {
+            self.result.add_assign(crate::msm::VariableBase::msm(
+                self.bases_buffer.as_slice(),
+                self.scalars_buffer.as_slice(),
+            ));
+            self.scalars_buffer.clear();
+            self.bases_buffer.clear();
+        }
+    }
+
+    /// Output the final Pippenger algorithm result.
+    #[inline(always)]
+    pub fn finalize(mut self) -> G::Projective {
+        if !self.scalars_buffer.is_empty() {
+            self.result.add_assign(crate::msm::VariableBase::msm(
+                self.bases_buffer.as_slice(),
+                self.scalars_buffer.as_slice(),
+            ));
+        }
+        self.result
+    }
+}

test-templates/src/lib.rs

@@ -47,6 +47,7 @@ macro_rules! generate_g1_test {
         #[test]
         fn test_g1_affine_curve() {
             test_var_base_msm::<G1Affine>();
+            ark_algebra_test_templates::msm::test_chunked_pippenger::<G1Affine>();
         }
 
         #[test]

test-templates/src/msm.rs

@@ -1,4 +1,7 @@
-use ark_ec::{msm::VariableBase, AffineCurve, ProjectiveCurve};
+use ark_ec::{
+    msm::{ChunkedPippenger, VariableBase},
+    AffineCurve, ProjectiveCurve,
+};
 use ark_ff::{PrimeField, UniformRand, Zero};
 
 fn naive_var_base_msm<G: AffineCurve>(
@@ -31,3 +34,26 @@ pub fn test_var_base_msm<G: AffineCurve>() {
 
     assert_eq!(naive.into_affine(), fast.into_affine());
 }
+
+pub fn test_chunked_pippenger<G: AffineCurve>() {
+    const SAMPLES: usize = 1 << 10;
+
+    let mut rng = ark_std::test_rng();
+
+    let v = (0..SAMPLES)
+        .map(|_| G::ScalarField::rand(&mut rng).into_repr())
+        .collect::<Vec<_>>();
+    let g = (0..SAMPLES)
+        .map(|_| G::Projective::rand(&mut rng))
+        .collect::<Vec<_>>();
+    let g = <G::Projective as ProjectiveCurve>::batch_normalization_into_affine(&g);
+
+    let arkworks = VariableBase::msm(g.as_slice(), v.as_slice());
+
+    let mut p = ChunkedPippenger::<G>::new(1 << 20);
+    for (s, g) in v.iter().zip(g) {
+        p.add(g, s);
+    }
+    let mine = p.finalize();
+    assert_eq!(arkworks.into_affine(), mine.into_affine());
+}