Suboptimal Instruction-Specific Constraints

[aszepieniec]

Upon skimming the result of cargo test print_number_and_degrees_of_transition_constraints_for_all_instructions -- --nocapture, 8 instructions stand out as having max degree 10 or 11. I am skeptical that these constraints must be so expensive. With the fancy technique that enabled pick and place, i.e., using an inner product with potentially shifted weight vectors, I am sure we can shrink these instructions' polynomials' degrees' maximum.

Instruction	#polys	max deg	Degrees
pop 1	16	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 4, 1, 1, 1]
divine 1	16	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 4, 1, 1, 1]
read_mem 1	30	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 10, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 1, 1]
write_mem 1	30	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 10, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 1, 1]
sponge_absorb	10	11	[1, 1, 1, 1, 1, 1, 11, 1, 1, 1]
sponge_squeeze	10	11	[1, 1, 1, 1, 1, 1, 11, 1, 1, 1]
read_io 1	16	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 5, 4, 1, 1]
write_io 1	16	10	[1, 1, 1, 1, 2, 2, 2, 2, 1, 5, 5, 10, 5, 4, 1, 1]