Change the alpha used when folding columns into fri layers.

[alon-f]

No description provided.

[reviewable-StarkWare]

This change is 

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.28%. Comparing base (c5af676) to head (6cee79f).
Report is 2 commits behind head on dev.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #972      +/-   ##
==========================================
+ Coverage   92.26%   92.28%   +0.02%     
==========================================
  Files         105      105              
  Lines       14292    14330      +38     
  Branches    14292    14330      +38     
==========================================
+ Hits        13187    13225      +38     
  Misses       1032     1032              
  Partials       73       73              

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

[ohad-starkware]

Reviewed 1 of 1 files at r2, all commit messages.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on @alon-f)

---

crates/prover/src/core/fri.rs line 447 at r2 (raw file):

        let first_layer_sparse_evals =
            self.decommit_first_layer(queries, first_layer_query_evals)?;
        let inner_layer_queries = queries.fold(CIRCLE_TO_LINE_FOLD_STEP);

This change alone is fine and IMO deserves a separate PR.

---

crates/prover/src/core/fri.rs line 470 at r2 (raw file):

    fn decommit_inner_layers(
        &self,
        inner_layer_queries: &Queries,

inner layer is already mentioned in the function name

Suggestion:

queries: &Queries,

---

crates/prover/src/core/fri.rs line 489 at r2 (raw file):

            {
                // Use the previous layer's folding alpha to fold the column's circle into the
                // current layer.

Suggestion:

                // Use the previous layer's folding alpha to fold the relevant circle evaluations into the current layer. 

---

crates/prover/src/core/fri.rs line 495 at r2 (raw file):

                    .fold_circle(previous_folding_alpha, *column_domain);

                // Mix the folded column evaluations into the current layer's evaluations.

? or more accurate doc, but I feel code is already made clear by the extraction to a function

---

crates/prover/src/core/fri.rs line 506 at r2 (raw file):

            (layer_queries, layer_query_evals) =
                layer.verify_and_fold(layer_queries, layer_query_evals)?;
            previous_folding_alpha = layer.folding_alpha;

very nice!

---

crates/prover/src/core/fri.rs line 691 at r2 (raw file):

impl<H: MerkleHasher> FriFirstLayerVerifier<H> {
    /// Verifies the first layer's merkle decommitment, and returns the evaluations needed for
    /// folding the columns to their corresponding layer.

the latter part of the sentence sounds weird to me, mixing evals with column might be confusing?
non-blocking

[ilyalesokhin-starkware]

crates/prover/src/core/fri.rs line 233 at r2 (raw file):

            layer_evaluation = folded_layer_evaluation;
            layers.push(layer);
        }

wdyt?

Suggestion:

      // fold max size columns.
        while let Some(column) = columns.next_if(|c| folded_size(c) == layer_evaluation.len()) {
            B::fold_circle_into_line(&mut layer_evaluation, column, folding_alpha, twiddles);
        }
        
        while layer_evaluation.len() > config.last_layer_domain_size() {
            let layer = FriInnerLayerProver::new(layer_evaluation);
            MC::mix_root(channel, layer.merkle_tree.root());
            let folding_alpha = channel.draw_felt();
            let folded_layer_evaluation = B::fold_line(&layer.evaluation, folding_alpha, twiddles);
            layers.push(layer);
            
            layer_evaluation = folded_layer_evaluation;
            // Check for circle polys in the first layer that should be combined in this layer.
            while let Some(column) = columns.next_if(|c| folded_size(c) == layer_evaluation.len()) {
                B::fold_circle_into_line(&mut layer_evaluation, column, folding_alpha, twiddles);
            }
        }

[alon-f]

Reviewable status: 0 of 1 files reviewed, 5 unresolved discussions (waiting on @ohad-starkware)

---

crates/prover/src/core/fri.rs line 447 at r2 (raw file):

Previously, ohad-starkware (Ohad) wrote…

This change alone is fine and IMO deserves a separate PR.

kk

---

crates/prover/src/core/fri.rs line 470 at r2 (raw file):

Previously, ohad-starkware (Ohad) wrote…

inner layer is already mentioned in the function name

Done.

---

crates/prover/src/core/fri.rs line 495 at r2 (raw file):

Previously, ohad-starkware (Ohad) wrote…

? or more accurate doc, but I feel code is already made clear by the extraction to a function

Done.

---

crates/prover/src/core/fri.rs line 506 at r2 (raw file):

Previously, ohad-starkware (Ohad) wrote…

very nice!

:D

---

crates/prover/src/core/fri.rs line 691 at r2 (raw file):

Previously, ohad-starkware (Ohad) wrote…

the latter part of the sentence sounds weird to me, mixing evals with column might be confusing?
non-blocking

i can change it to circle evaluations like you wrote above

---

crates/prover/src/core/fri.rs line 489 at r2 (raw file):

            {
                // Use the previous layer's folding alpha to fold the column's circle into the
                // current layer.

i don't like relevant, so i'll add the infamous sparse

[ohad-starkware]

Reviewable status: 0 of 1 files reviewed, 1 unresolved discussion (waiting on @alon-f)

---

crates/prover/src/core/fri.rs line 691 at r2 (raw file):

Previously, alon-f wrote…

i can change it to circle evaluations like you wrote above

u decide, im not sure my confusion is justified

[ohad-starkware]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @alon-f)

---

-- commits line 1 at r3:
squash

[alon-f]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware and @ohad-starkware)

---

crates/prover/src/core/fri.rs line 233 at r2 (raw file):

Previously, ilyalesokhin-starkware wrote…

wdyt?

Done.

[alon-f]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware and @ohad-starkware)

---

-- commits line 1 at r3:

Previously, ohad-starkware (Ohad) wrote…

squash

Done.

[ohad-starkware]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ilyalesokhin-starkware)