Joseandro Luiz

This means changing the Columns structs to have only one element called “X” or whatever, and then it can have different aliases for different types of columns that give “X”...

Create selectors for dummy rows in final instances

1

Looks like very few changes are needed after speaking with Fabrizio.

Correct generation of challenges with sponge absorptions in folding

1

- Keep order by absorbing each instance as bulks instead of order of execution. - Might be a good idea to have some kind of iterator across the instances to...

- Some columns come from “constraint-world” columns, and additional ones correspond to expressions built from those that need to be constrained. - Maybe we want to create a “compound” witness...

- Different traits for the columns to extract the per-instance columns? - Selector type inside the Env to keep track of what type of row/instance was added? - Probably need...

We probably want to keep the witness interpreter code mostly unchanged, and create traits to “extract” the individual instances from the columns and the list of constraints and lookups.

Track the length with the desired number of steps needed until squeeze happens

- On MIPS side, mainly about memory checks: This is a type of under-constraining. Detect them and include them. - On MIPS side: check if range-check to check we are...

The one in optimism has a mode and a magnitude, whereas the one in msm has a numerator (which in optimism can be obtained from the mode and the magnitude).

- Right now, in the main.rs, everytime the step_index (keccak) or instruction_counter (mips) hits 2^15, the code calls the fold function. - This assumes we just have 2 instances: one...