Faster way to do LLT?
As I understand it, everytime the LLT is done to M(n), computation starts from S(1) mod M(n) up to S(n1) mod M(n).
Suppose, if the unMODed value of S(n1) is saved so that it can be used for the next possible candidates and MODed to M(n+k), wouldn't it be faster?
I know that S(n1) can be extremely large in comparison to M(n), but does using more hard disk space compensate for the time spent in recomputing from S(1)?
