perf: use faster polynomial coset evaluation

Cargo.toml

@@ -113,4 +113,4 @@ default-features = false
 features = ["precommit-hook", "run-cargo-clippy", "run-cargo-fmt"]
 
 [patch.crates-io]
-twenty-first = { git = "https://github.com/Neptune-Crypto/twenty-first.git", rev = "1e065d68" }
+twenty-first = { git = "https://github.com/Neptune-Crypto/twenty-first.git", rev = "b57ed599" }

triton-vm/src/arithmetic_domain.rs

@@ -1,3 +1,4 @@
+use std::ops::Mul;
 use std::ops::MulAssign;
 
 use num_traits::One;
@@ -74,14 +75,21 @@ impl ArithmeticDomain {
 
     pub fn interpolate<FF>(&self, values: &[FF]) -> Polynomial<FF>
     where
-        FF: FiniteField + MulAssign<BFieldElement> + From<BFieldElement>,
+        FF: FiniteField
+            + MulAssign<BFieldElement>
+            + Mul<BFieldElement, Output = FF>
+            + From<BFieldElement>,
     {
-        Polynomial::fast_coset_interpolate(self.offset.into(), self.generator, values)
+        // generic type made explicit to avoid performance regressions due to auto-conversion
+        Polynomial::fast_coset_interpolate::<BFieldElement>(self.offset, self.generator, values)
     }
 
     pub fn low_degree_extension<FF>(&self, codeword: &[FF], target_domain: Self) -> Vec<FF>
     where
-        FF: FiniteField + MulAssign<BFieldElement> + From<BFieldElement>,
+        FF: FiniteField
+            + MulAssign<BFieldElement>
+            + Mul<BFieldElement, Output = FF>
+            + From<BFieldElement>,
     {
         target_domain.evaluate(&self.interpolate(codeword))
     }

triton-vm/src/table/master_table.rs

@@ -1,3 +1,4 @@
+use std::ops::Mul;
 use std::ops::MulAssign;
 use std::ops::Range;
 
@@ -207,7 +208,10 @@ pub enum TableId {
 /// [master_quot_table]: all_quotients
 pub trait MasterTable<FF>: Sync
 where
-    FF: FiniteField + MulAssign<BFieldElement> + From<BFieldElement>,
+    FF: FiniteField
+        + MulAssign<BFieldElement>
+        + Mul<BFieldElement, Output = FF>
+        + From<BFieldElement>,
     Standard: Distribution<FF>,
 {
     fn trace_domain(&self) -> ArithmeticDomain;