Define extended GCD, combined GCD+LCM #19
Conversation
strake
force-pushed
the
gcd
branch
5 times, most recently
from
6accdbc
to
f6f06f8
Compare
|
Rebased, PTAL |
src/lib.rs
Outdated
| #[cfg_attr(array_clone, derive(Clone))] | ||
| pub struct ExtendedGcd<A> { | ||
| pub gcd: A, | ||
| pub coeffs: [A; 2], |
There was a problem hiding this comment.
Let's use separate x and y, so we can document it something like this:
Bézout's identity — Let a and b be integers with greatest common divisor d. Then, there exist integers x and y such that ax + by = d.
Then we also don't have to worry about array_clone.
src/lib.rs
Outdated
| let q = r.1.clone() / r.0.clone(); | ||
| let f = |r: &mut (Self, Self)| { | ||
| mem::swap(&mut r.0, &mut r.1); | ||
| r.0 -= q.clone() * r.1.clone(); |
There was a problem hiding this comment.
Hmm, it's a little annoying to have SubAssign in the API just for this, and AFAICT it's only needed because you're working through &mut. If the closure used values instead, or if you just flattened these three calls, then you could move and then reassign r.0 just fine.
There was a problem hiding this comment.
Maybe we should require Signed though -- I believe it's useless for unsigned integers, since one of the coefficients must be negative, except the degenerate case with zeros.
|
@cuviper All done, PTAL |
|
Thanks! bors r+ |
Build failed |
|
Sorry, that was my fault. bors r+ |
Build succeeded |
No description provided.