>>5695

CA Here, like usual doing my own thing. Have we tried this algorithm?

Take your start value c. Then make values e,n,d,x,a,b. Make a new value (c-n). Repeat with (c-n). It works a decent amount. But the problem is dealing with the parity because to make a record you need (a,b) to have the same parity. I've so far just taken away the largest power of 2 that I can from the number and I make the next record b-n.

>>5690

Something like this?

>>5701

should be 11 instead of 13

>>5700

Start with e,n,d,x,a=1,b=c

Then I make a new value from H = b-n.

Then from here I need to factor H (recursion) and get m (a factor of H)

Then, we make a new record e,n,d,x,a,b = AB(1,H/m)

Then make new value H=b-n

Factor H to get m

rec = AB(1,H/m)

etc..

At each step, if you check the gcd of the original number with each of H, n, and b one of these will have the result. I tried with the first 3741 coprimes and it worked

>>5704

Only problem is I need to know when a record is prime.

>>5704

nvm you just need to check the gcd with the b value.

>>5704

nvm false alarm