Returns a vec of evaluations: Val(k) = unexpanded_pc(k) + gamma * imm(k) + gamma^2 * circuit_flags[0](k) + gamma^3 * circuit_flags[1](k) + ... This pa

Returns a vec of evaluations: Val(k) = rd(k, r_register) + gamma * rs1(k, r_register) + gamma^2 * rs2(k, r_register) where rd(k, k') = 1 if the k'th i

Returns a vec of evaluations: Val(k) = rd(k, r_register) + gamma * unexpanded_pc(k) + gamma^2 * instr_raf_flag(k) + gamma^3 * lookup_table_flag[0](k)

Deserializes data from a byte vector

Returns the digest of the r1cs shape

Generic version for distributing tA to svo accumulators

Computes the table of coefficients like `evals`, but also caches the intermediate results In other words, computes `{eq(r[i..], x) for all x in {0, 1

Same as evals_cached but for high-to-low (reverse) binding order

Computes the table of coefficients: scaling_factor * eq(r, x) for all x in {0, 1}^n serially. More efficient for short `r`.

Computes the table of coefficients like `evals_serial`, but also caches the intermediate results. Returns a vector of vectors, where the `j`th vector

evals_serial_cached but for "high to low" ordering, used specifically in the Gruen x Dao Thaler optimization.

This MLE is 1 if y = x + 1 for x in the range [0... 2^l-2]. That is, it ignores the case where x is all 1s, outputting 0. Assumes x and y are

For r in the Boolean hypercube, this MLE should evaluate to 1 if r falls in the range [range_start, range_end) and 0 otherwise In other words, LT(r, r