Add operand_chunks method to JoltInstruction; add chunk_operand funct…

…ion to instruction_utils

jolt-core/src/jolt/instruction/mod.rs

@@ -9,6 +9,7 @@ pub trait JoltInstruction {
   fn g_poly_degree(&self, C: usize) -> usize;
   fn subtables<F: PrimeField>(&self, C: usize) -> Vec<Box<dyn LassoSubtable<F>>>;
   fn to_indices(&self, C: usize, log_M: usize) -> Vec<usize>;
+  fn operand_chunks(&self, C: usize, log_M: usize) -> (Vec<u64>, Vec<u64>);
 }
 
 pub trait Opcode {

jolt-core/src/utils/instruction_utils.rs

@@ -14,6 +14,16 @@ pub fn concatenate_lookups<F: PrimeField>(vals: &[F], C: usize, shift_bits: usiz
   sum
 }
 
+pub fn chunk_operand(x: u64, num_chunks: usize, bits_per_chunk: usize) -> Vec<u64> {
+  let bit_mask = (1 << bits_per_chunk) - 1;
+  (0..num_chunks)
+    .map(|i| {
+      let shift = ((num_chunks - i - 1) * bits_per_chunk) as u32;
+      x.checked_shr(shift).unwrap_or(0) & bit_mask;
+    })
+    .collect()
+}
+
 pub fn chunk_and_concatenate_operands(x: u64, y: u64, C: usize, log_M: usize) -> Vec<usize> {
   let operand_bits: usize = log_M / 2;
   let operand_bit_mask: usize = (1 << operand_bits) - 1;