-
Notifications
You must be signed in to change notification settings - Fork 430
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add WeightedIndexTree to rand_distr #1372
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Incomplete (haven't fully reviewed code).
rand_distr/src/weighted_tree.rs
Outdated
impl<W: Clone + PartialEq + PartialOrd + SampleUniform + SubAssign<W> + Weight> | ||
Distribution<Result<usize, WeightedError>> for WeightedTreeIndex<W> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We don't usually implement Distribution<Result<..>>
. Do we need to?
I think I would prefer to panic on error, but guarantee no panic if self.is_valid()
(self.can_sample()
).
@vks?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I removed the Result<>, so sample now panics instead. I added the info about is_valid
to the doc string and introduced safe_sample
as an alternative that does not panic.
Thank you for the review. I addressed all the comments |
Thanks, that is much better, changed to |
// Otherwise we found the index with the target weight. | ||
break; | ||
} | ||
assert!(target_weight >= W::ZERO); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If I understand correctly, this must be strictly > 0?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As an intermediate state, this can be zero if all weights are zero or there are no elements. is_valid
would then return false
. Allowing this is useful, so that the user does not have to apply the updates in the right order to avoid intermediate zero states.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In this case the function will already have returned an error.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry I mixed up what this comment was referring to.
I think target_weight
can be 0
in this line, for example: if we have a tree with only one node and gen_range
samples 0
, then the loop
block is a no-op and after we hit this line with target_weight = 0
.
However, I realized the line right after was incorrect, it should be: assert!(target_weight < self.get(index))
. Should be correct now.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I agree
We can ignore the Thanks for the implementation, @xmakro. |
As discussed in #1053