Releases: cavemanloverboy/FNNTW
FNNTW v0.4.1
This release brings even greater performance through a handful of avenues and new functionality. In particular,
- we introduce a query specialized for cosmology which returns the components of the vector pointing from the query to the neighbor. This allows for studying anisotropy and 2D kNNs (See Yuan, Zamora, Abel 2023). For now, we only support axis-aligned line-of-sight and transverse decomposition, but plan to support more general decomposition in the future in anticipation of mock catalogs from light cone simulations and observational data.
f32
is now supported
FNNTW v0.2.0
This release introduces breaking changes while adding functionality. FNNTW now supports
- k-Nearest Neighbor queries
- Periodic Boundary Conditions.
Additional Info:
In addition, internal brute force tests were added for all query functionalities, and new benchmarks were added for the new features.
FNNTW v0.1.3
This release brings a minor improvement: the vector holding the tree nodes is preallocated with Vec::with_capacity
. Some of the documentation is also corrected
FNNTW v0.1.2
The RwLock
used from parking_lot
did not provide any measurable difference in performance for parallel builds while adding a dependency with a handful of internal dependencies. This release switched back to std::sync::RwLock
, significantly reducing the number of dependencies and, consequently, the build time.
FNNTW v0.1.1
- Added readme to docs.
- Changed name from FNSTW to FNNTW
FNSTW v0.1.0-alpha
In this alpha release, we are proud to announce the following basic features:
- Support for both parallel and nonparallel kDtree builds.
- Support for a nearest neighbor query function that returns the distance to, the index of, and the position of a query point's nearest neighbor.