constant column attempt 1

[ohad-starkware]

This change is 

[ohad-starkware]

• constant column attempt 1 #849 👈
• dev

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @ohad-starkware and the rest of your teammates on Graphite

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 57.64706% with 36 lines in your changes missing coverage. Please review.

Project coverage is 91.62%. Comparing base (cdf66f3) to head (b2ba749).
Report is 1 commits behind head on dev.

Files with missing lines	Patch %	Lines
...rover/src/constraint_framework/constant_columns.rs	0.00%	13 Missing and 1 partial ⚠️
crates/prover/src/core/pcs/utils.rs	29.41%	10 Missing and 2 partials ⚠️
crates/prover/src/constraint_framework/info.rs	0.00%	3 Missing ⚠️
crates/prover/src/constraint_framework/point.rs	0.00%	3 Missing ⚠️
...tes/prover/src/constraint_framework/simd_domain.rs	0.00%	3 Missing ⚠️
...rates/prover/src/constraint_framework/component.rs	97.05%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #849      +/-   ##
==========================================
- Coverage   91.86%   91.62%   -0.24%     
==========================================
  Files          89       89              
  Lines       12080    12169      +89     
  Branches    12080    12169      +89     
==========================================
+ Hits        11097    11150      +53     
- Misses        876      908      +32     
- Partials      107      111       +4     

Flag	Coverage Δ
	91.62% <57.64%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[ilyalesokhin-starkware]

crates/prover/src/core/air/components.rs line 49 at r4 (raw file):

        let mut static_column_masks: Vec<BTreeSet<CirclePoint<SecureField>>> = vec![];
        for component in &self.0 {
            let component_static_masks = &component.mask_points(point)[CONST_INTERACTION];

What about the other interactions?

Code quote:

      let component_static_masks = &component.mask_points(point)[CONST_INTERACTION];

[ohad-starkware]

crates/prover/src/core/air/components.rs line 49 at r4 (raw file):

Previously, ilyalesokhin-starkware wrote…

What about the other interactions?

look at the function above

[ilyalesokhin-starkware]

crates/prover/src/core/pcs/utils.rs line 130 at r4 (raw file):

        let mut res = TreeVec(vec![Vec::new(); max_tree_index + 1]);

        for &location in locations.iter().filter(|l| l.col_start != l.col_end) {

What is happening here?

Code quote:

.iter().filter(|l| l.col_start != l.col_end)

[ohad-starkware]

Reviewable status: 0 of 30 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware)

---

crates/prover/src/core/pcs/utils.rs line 130 at r4 (raw file):

Previously, ilyalesokhin-starkware wrote…

What is happening here?

sry this should have come with a TODO to handle it differently
it ignores 'empty' subspans , because the unwrap panics

[ilyalesokhin-starkware]

Reviewable status: 0 of 30 files reviewed, 2 unresolved discussions (waiting on @ohad-starkware)

---

crates/prover/src/constraint_framework/component.rs line 182 at r4 (raw file):

            [
                trace.polys.sub_tree(&mask_spans),
                trace.polys.sub_tree_sparse(&self.static_columns_locations),

Why is this at the end?

Code quote:

 trace.polys.sub_tree_sparse(&self.static_columns_locations),

[ohad-starkware]

Reviewable status: 0 of 30 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware)

---

crates/prover/src/constraint_framework/component.rs line 182 at r4 (raw file):

Previously, ilyalesokhin-starkware wrote…

Why is this at the end?

order wont matter here -
Treevec::concat_cols is concat-by-index, meaning res[i] = concat (tree_0[i]..tree_n[i])
here there are 2 trees, the first has an empty vec in its first entry, and the second is empty in the rest.

[ilyalesokhin-starkware]

Reviewable status: 0 of 30 files reviewed, 3 unresolved discussions (waiting on @ohad-starkware)

---

crates/prover/src/core/air/components.rs line 83 at r4 (raw file):

        mask_values: &[Vec<SecureField>],
    ) -> ColumnVec<Vec<SecureField>> {
        let mask_by_column = &self.mask_points_by_column(point)[CONST_INTERACTION];

Why not?

Suggestion:

&self.const_mask_points_by_column(point);

[ohad-starkware]

Reviewable status: 0 of 30 files reviewed, 3 unresolved discussions (waiting on @ilyalesokhin-starkware)

---

crates/prover/src/core/air/components.rs line 83 at r4 (raw file):

Previously, ilyalesokhin-starkware wrote…

Why not?

I guess it was a mistake
iirc I tried to allow 'global access' regardless of the interaction, and later gave up on that.