-
Notifications
You must be signed in to change notification settings - Fork 301
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
chore: Shared Permutation+Lookup relation arithmetic (AztecProtocol/b…
…arretenberg#559) Co-authored-by: ledwards2225 <[email protected]>
- Loading branch information
1 parent
97c9132
commit 4638333
Showing
23 changed files
with
564 additions
and
510 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
166 changes: 166 additions & 0 deletions
166
barretenberg/cpp/src/barretenberg/honk/proof_system/grand_product_library.hpp
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,166 @@ | ||
| #pragma once | ||
| #include "barretenberg/honk/sumcheck/sumcheck.hpp" | ||
| #include "barretenberg/plonk/proof_system/proving_key/proving_key.hpp" | ||
| #include "barretenberg/polynomials/polynomial.hpp" | ||
| #include <typeinfo> | ||
|
|
||
| namespace proof_system::honk::grand_product_library { | ||
|
|
||
| // TODO(luke): This contains utilities for grand product computation and is not specific to the permutation grand | ||
| // product. Update comments accordingly. | ||
| /** | ||
| * @brief Compute a permutation grand product polynomial Z_perm(X) | ||
| * * | ||
| * @details | ||
| * Z_perm may be defined in terms of its values on X_i = 0,1,...,n-1 as Z_perm[0] = 1 and for i = 1:n-1 | ||
| * relation::numerator(j) | ||
| * Z_perm[i] = ∏ -------------------------------------------------------------------------------- | ||
| * relation::denominator(j) | ||
| * | ||
| * where ∏ := ∏_{j=0:i-1} | ||
| * | ||
| * The specific algebraic relation used by Z_perm is defined by Flavor::GrandProductRelations | ||
| * | ||
| * For example, in Flavor::Standard the relation describes: | ||
| * | ||
| * (w_1(j) + β⋅id_1(j) + γ) ⋅ (w_2(j) + β⋅id_2(j) + γ) ⋅ (w_3(j) + β⋅id_3(j) + γ) | ||
| * Z_perm[i] = ∏ -------------------------------------------------------------------------------- | ||
| * (w_1(j) + β⋅σ_1(j) + γ) ⋅ (w_2(j) + β⋅σ_2(j) + γ) ⋅ (w_3(j) + β⋅σ_3(j) + γ) | ||
| * where ∏ := ∏_{j=0:i-1} and id_i(X) = id(X) + n*(i-1) | ||
| * | ||
| * For Flavor::Ultra both the UltraPermutation and Lookup grand products are computed by this method. | ||
| * | ||
| * The grand product is constructed over the course of three steps. | ||
| * | ||
| * For expositional simplicity, write Z_perm[i] as | ||
| * | ||
| * A(j) | ||
| * Z_perm[i] = ∏ -------------------------- | ||
| * B(h) | ||
| * | ||
| * Step 1) Compute 2 length-n polynomials A, B | ||
| * Step 2) Compute 2 length-n polynomials numerator = ∏ A(j), nenominator = ∏ B(j) | ||
| * Step 3) Compute Z_perm[i + 1] = numerator[i] / denominator[i] (recall: Z_perm[0] = 1) | ||
| * | ||
| * Note: Step (3) utilizes Montgomery batch inversion to replace n-many inversions with | ||
| */ | ||
| template <typename Flavor, typename GrandProdRelation> | ||
| void compute_grand_product(const size_t circuit_size, | ||
| auto& full_polynomials, | ||
| sumcheck::RelationParameters<typename Flavor::FF>& relation_parameters) | ||
| { | ||
| using FF = typename Flavor::FF; | ||
| using Polynomial = typename Flavor::Polynomial; | ||
| using ValueAccumTypes = typename GrandProdRelation::ValueAccumTypes; | ||
|
|
||
| // Allocate numerator/denominator polynomials that will serve as scratch space | ||
| // TODO(zac) we can re-use the permutation polynomial as the numerator polynomial. Reduces readability | ||
| Polynomial numerator = Polynomial{ circuit_size }; | ||
| Polynomial denominator = Polynomial{ circuit_size }; | ||
|
|
||
| // Step (1) | ||
| // Populate `numerator` and `denominator` with the algebra described by Relation | ||
| const size_t num_threads = circuit_size >= get_num_cpus_pow2() ? get_num_cpus_pow2() : 1; | ||
| const size_t block_size = circuit_size / num_threads; | ||
| parallel_for(num_threads, [&](size_t thread_idx) { | ||
| const size_t start = thread_idx * block_size; | ||
| const size_t end = (thread_idx + 1) * block_size; | ||
| for (size_t i = start; i < end; ++i) { | ||
|
|
||
| typename Flavor::ClaimedEvaluations evaluations; | ||
| for (size_t k = 0; k < Flavor::NUM_ALL_ENTITIES; ++k) { | ||
| evaluations[k] = full_polynomials[k].size() > i ? full_polynomials[k][i] : 0; | ||
| } | ||
| numerator[i] = GrandProdRelation::template compute_grand_product_numerator<ValueAccumTypes>( | ||
| evaluations, relation_parameters, i); | ||
| denominator[i] = GrandProdRelation::template compute_grand_product_denominator<ValueAccumTypes>( | ||
| evaluations, relation_parameters, i); | ||
| } | ||
| }); | ||
|
|
||
| // Step (2) | ||
| // Compute the accumulating product of the numerator and denominator terms. | ||
| // This step is split into three parts for efficient multithreading: | ||
| // (i) compute ∏ A(j), ∏ B(j) subproducts for each thread | ||
| // (ii) compute scaling factor required to convert each subproduct into a single running product | ||
| // (ii) combine subproducts into a single running product | ||
| // | ||
| // For example, consider 4 threads and a size-8 numerator { a0, a1, a2, a3, a4, a5, a6, a7 } | ||
| // (i) Each thread computes 1 element of N = {{ a0, a0a1 }, { a2, a2a3 }, { a4, a4a5 }, { a6, a6a7 }} | ||
| // (ii) Take partial products P = { 1, a0a1, a2a3, a4a5 } | ||
| // (iii) Each thread j computes N[i][j]*P[j]= | ||
| // {{a0,a0a1},{a0a1a2,a0a1a2a3},{a0a1a2a3a4,a0a1a2a3a4a5},{a0a1a2a3a4a5a6,a0a1a2a3a4a5a6a7}} | ||
| std::vector<FF> partial_numerators(num_threads); | ||
| std::vector<FF> partial_denominators(num_threads); | ||
|
|
||
| parallel_for(num_threads, [&](size_t thread_idx) { | ||
| const size_t start = thread_idx * block_size; | ||
| const size_t end = (thread_idx + 1) * block_size; | ||
| for (size_t i = start; i < end - 1; ++i) { | ||
| numerator[i + 1] *= numerator[i]; | ||
| denominator[i + 1] *= denominator[i]; | ||
| } | ||
| partial_numerators[thread_idx] = numerator[end - 1]; | ||
| partial_denominators[thread_idx] = denominator[end - 1]; | ||
| }); | ||
|
|
||
| parallel_for(num_threads, [&](size_t thread_idx) { | ||
| const size_t start = thread_idx * block_size; | ||
| const size_t end = (thread_idx + 1) * block_size; | ||
| if (thread_idx > 0) { | ||
| FF numerator_scaling = 1; | ||
| FF denominator_scaling = 1; | ||
|
|
||
| for (size_t j = 0; j < thread_idx; ++j) { | ||
| numerator_scaling *= partial_numerators[j]; | ||
| denominator_scaling *= partial_denominators[j]; | ||
| } | ||
| for (size_t i = start; i < end; ++i) { | ||
| numerator[i] *= numerator_scaling; | ||
| denominator[i] *= denominator_scaling; | ||
| } | ||
| } | ||
|
|
||
| // Final step: invert denominator | ||
| FF::batch_invert(std::span{ &denominator[start], block_size }); | ||
| }); | ||
|
|
||
| // Step (3) Compute z_perm[i] = numerator[i] / denominator[i] | ||
| auto& grand_product_polynomial = GrandProdRelation::get_grand_product_polynomial(full_polynomials); | ||
| grand_product_polynomial[0] = 0; | ||
| parallel_for(num_threads, [&](size_t thread_idx) { | ||
| const size_t start = thread_idx * block_size; | ||
| const size_t end = (thread_idx == num_threads - 1) ? circuit_size - 1 : (thread_idx + 1) * block_size; | ||
| for (size_t i = start; i < end; ++i) { | ||
| grand_product_polynomial[i + 1] = numerator[i] * denominator[i]; | ||
| } | ||
| }); | ||
| } | ||
|
|
||
| template <typename Flavor> | ||
| void compute_grand_products(std::shared_ptr<typename Flavor::ProvingKey>& key, | ||
| typename Flavor::ProverPolynomials& full_polynomials, | ||
| sumcheck::RelationParameters<typename Flavor::FF>& relation_parameters) | ||
| { | ||
| using GrandProductRelations = typename Flavor::GrandProductRelations; | ||
| using FF = typename Flavor::FF; | ||
|
|
||
| constexpr size_t NUM_RELATIONS = std::tuple_size<GrandProductRelations>{}; | ||
| barretenberg::constexpr_for<0, NUM_RELATIONS, 1>([&]<size_t i>() { | ||
| using GrandProdRelation = typename std::tuple_element<i, GrandProductRelations>::type; | ||
|
|
||
| // Assign the grand product polynomial to the relevant std::span member of `full_polynomials` (and its shift) | ||
| // For example, for UltraPermutationRelation, this will be `full_polynomials.z_perm` | ||
| // For example, for LookupRelation, this will be `full_polynomials.z_lookup` | ||
| std::span<FF>& full_polynomial = GrandProdRelation::get_grand_product_polynomial(full_polynomials); | ||
| auto& key_polynomial = GrandProdRelation::get_grand_product_polynomial(*key); | ||
| full_polynomial = key_polynomial; | ||
|
|
||
| compute_grand_product<Flavor, GrandProdRelation>(key->circuit_size, full_polynomials, relation_parameters); | ||
| std::span<FF>& full_polynomial_shift = | ||
| GrandProdRelation::get_shifted_grand_product_polynomial(full_polynomials); | ||
| full_polynomial_shift = key_polynomial.shifted(); | ||
| }); | ||
| } | ||
|
|
||
| } // namespace proof_system::honk::grand_product_library |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.