vers
vers copied to clipboard
Cydhra
very efficient rank and select
The pointer-free nature of succinct data-structures makes them very amenable to (de)serialization by simply casting their memory to/from a bunch of bytes. Not only would this remove most (de)serialization costs,...
It would be nice if there was a [WaveletMatrix](http://www.dcc.uchile.cl/~gnavarro/ps/spire12.4.pdf) implementation, as it would enable the most common succinct-datastructure applications in text-processing and databases.
Listed here are breaking changes to implement if a major version release is planned - [x] Rename `FastRmq` to `SmallRmq` or something similar, expressing its higher memory efficiency - [x]...
This adds a zerocopy capabilities based on the [`anybytes`](https://docs.rs/anybytes/latest/anybytes/) crate.
Add support for partitioned Min-Max-Trees as described in Section 3 of [Cordova & Navarro](https://www.sciencedirect.com/science/article/pii/S0304397516300706). The partitions will reduce the size of the mMTs, especially because it allows the trees to...
Similarly to how we exploit the rank/select structure in the `iter1()` implementation of `RsVec`, we can exploit it to find a predecessor/successor bit to any given index quickly (in the...
Now that nightly supports portable_simd, we can attempt to replace the SIMD implementations with portable_simd. In the current state, this does not avoid `unsafe` entirely, because while the `BlockDescriptor` struct...
This is the draft for a future 2.0 update. Most plans for changes are listed in #9 and #35. If more come up, they will be tracked here.
While you create an `RsVec` from a bit vector using `from_bit_vec`, the `BpTree` is created using `from_bit_vector`. Perhaps `from_bit_vector` can be deprecated in favor of `from_bit_vec`?
