v0.4.21

NearestNeighbors v0.4.21

Diff since v0.4.20

Merged pull requests:

• Use the new MinkowskiMetric type from Distances.jl (#206) (@eliascarv)

v0.4.20

NearestNeighbors v0.4.20

Diff since v0.4.19

Merged pull requests:

• revert "vectorize the compute_bbox function" (#205) (@KristofferC)

Closed issues:

• hyperractangles' compute_bbox seems to break KDTree due to setindex() for non StaticArrays types (#204)

v0.4.19

NearestNeighbors v0.4.19

Diff since v0.4.18

Merged pull requests:

• Grab bag of fixups / improvements (#181) (@KristofferC)
• Add mutating version of knn and inrange (#185) (@KristofferC)
• improve docs w.r.t what metrics can be used (#186) (@KristofferC)
• relax some types to make things work with views (#187) (@KristofferC)
• add a blurb about package not being suitable for very high dimensional points (#188) (@KristofferC)
• fix min max hyperrec (#191) (@KristofferC)
• optimize some hypersphere checks for minkowski metrics (#195) (@KristofferC)
• avoid some square rooting to check distances (#197) (@KristofferC)
• bump the default leafsize from 10 to 25 (#198) (@KristofferC)
• bump version (#202) (@KristofferC)

Closed issues:

• Approximate Nearest Neighbors algorithms (#73)
• Allow @view Vector{SVector} in KDTree (#74)
• Trees for non-Metrics? (#75)
• In-place version of knn (#81)
• Not working with Adjoint type? (#82)
• Creating tree on views (#88)
• Optimized counting of inrange nearest neighbors (#108)
• Performant way to estimate distances for bulk inrange search? (#119)
• Bug: BoundsError when skipping all entries (#127)
• nn lacking docstring (#147)
• README.md Misleading Custom Metric Documentation (#152)
• Compilation time issues with very high dimensions (#156)
• Make datatypes of the KNN results selectable for potentially lower memory overhead (#160)
• KDTree with Matrix{ComplexF64} (#175)
• get_min_distance_sq seems weird (#189)
• In place queries not available (#201)

v0.4.18

NearestNeighbors v0.4.18

Diff since v0.4.17

Merged pull requests:

• chatgpt the README (#178) (@KristofferC)
• allow views as input to query functions (#183) (@KristofferC)

Closed issues:

• Is there any plan to introduce una-dinosauria/Rayuela.jl ?? (#91)
• Can't do knn on AbstractVector{SVector} (#176)
• Test benchmarks and have them run on CI (#179)

v0.4.17

NearestNeighbors v0.4.17

Diff since v0.4.16

Merged pull requests:

• KDTree: SoA the Node struct to avoid paying alignment cost to store the split dimension (#177) (@KristofferC)

v0.4.16

NearestNeighbors v0.4.16

Diff since v0.4.15

Merged pull requests:

• use a separate parameter for the hyper rectangle (#174) (@KristofferC)

Closed issues:

• Cannot build KDTree with Subarrays since v0.4.14 (#173)

v0.4.15

NearestNeighbors v0.4.15

Diff since v0.4.14

Merged pull requests:

• Fix tree construction for vectors without StaticArrays.setindex (#172) (@fredrikekre)

v0.4.14

NearestNeighbors v0.4.14

Diff since v0.4.13

Merged pull requests:

• Allow usage of semi and pre-metrics with BruteTree (#137) (@davnn)
• Make KDNode smaller (#166) (@KristofferC)
• avoid copying inputs when doing a knn or inrange for a matrix input (#168) (@KristofferC)
• some fixes for weighted metrics (#170) (@KristofferC)
• avoid unnecessarily overloading Distances.evaluate (#171) (@KristofferC)

Closed issues:

• BruteTree SemiMetric support (#99)
• Benchmarking parallel implementation leads to segfaults (#136)
• Reverse Cuthill-McKee ordering option (#158)
• Julia 1.10 is waiting on IO to finish during compilation (#164)
• It should be possible to make KDNode smaller (#165)
• KDTree: Wrong results for non-Euclidean metrics (#169)

v0.4.13

NearestNeighbors v0.4.13

Diff since v0.4.12

Closed issues:

• [Question] Can you insert new data into an existing KDTree object? (#155)

Merged pull requests:

• add some code to run during precompilation (#157) (@KristofferC)

v0.4.12

NearestNeighbors v0.4.12

Diff since v0.4.11

Closed issues:

• AssertionError in kmedoids alg (#148)
• bug: BoundsError when the skip function returns true for all points (#149)
• Is there a reason sqeuclidean distance is not supported? (#151)
• Document that inrangecount also counts the point itself (#153)

Merged pull requests:

• Add generic constructor to HyperSphere (#154) (@eliascarv)