interpn 0.11.2

N-dimensional interpolation/extrapolation methods, no-std and no-alloc compatible.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289

# Changelog

## 0.11.2 2026-07-02

### Changed

* Implement analytic gradients for multidimensional methods

## 0.11.1 2026-06-13

### Changed

* Extract Hermite node weights from eval loop for cubic methods
  * 3x asymptotic speedup for large data; modest improvement in latency
* Use inlined helper function for FMA feature switching

### Added

* Add B-spline methods for regular and rectilinear grids
  * Tridiagonal coeff solve using preallocated scratch space, optionally parallelized
  * Use same BCs as Hermite methods (zero third derivative) for reduced ringing and graceful transition to linearized extrapolation

## 0.11.0 2026-01-15

### Changed

* !Remove recursive implementations
  * All dimensionalities now use const-generic bounded implementations
  * For dimensions higher than const-unrolling limit, a run-time loop of the same form is used
  * No notable change in performance for high dimensions
  * No change to API for `::interpn` convenience functions
  * All methods are now compatible with static analysis of memory usage via cargo-call-stack
* Refactor internals to reduce duplication
* Improve form of early bounds checks to be interpreted correctly by the compiler
  * 30% speedup in some cases, no effect on most
  * Bounds checks done via iterator do not always result in optimizing out panic branches
* Test cubic methods in higher dimensions

## 0.10.0 2026-01-10

### Added

* Rust
  * Add top-level `interpn`, `interpn_alloc` and `interpn_serial` methods along with supporting enums for selecting methods
  * Add `par` feature, enabled by default, that enables parallelism with rayon in `interpn` function
  * Add lazy static for getting number of physical cores to advise thread chunk sizes

### Changed

* Rust
  * Update deps
* Python
  * !Refactor `interpn` function
    * Combine `check_bounds: bool` and `bounds_check_atol: float` to `check_bounds_with_atol: float | None`
    * Replace `assume_regular` input with optional `grid_kind` to allow assuming either regular or rectilinear, or making no assumption
    * Add `max_threads: int | None` input to allow manually limiting parallelism
    * Use rust top-level `interpn` function as backend for method selection, bounds checks, and parallelism

## 0.9.1 2025-12-31

### Changed

* Rust
  * Remove explicit const-unrolling for shallow loops (eliminates nested const-unrolling)
    * Improves compile time
    * Minimal effect on performance in Rust benchmarks
    * About 20% improvement for single-sample throughput from Python
* Python
  * Lock python package version to rust version
  * Improve handling of `__version__` when package is not present

## 0.9.0 2025-12-27

### Added

* Rust
  * Add `deep-unroll` feature that sets crunchy unroll depth to 256 and enables 4D unrolled cubic interpolation
    * This improves compile times in typical use-cases

### Changed

* Rust
  * Gate 4D unrolled cubic interpolation behind `deep-unroll` feature
  * Enable `deep-unroll` feature for Python builds
  * Update pyo3 deps

## 0.8.2 2025-11-12

### Changed

* Use `x86-64-v3` reference CPU as compilation target instead of manually specifying features
  * Includes AVX2 and vector FMA

## 0.8.1 2025-11-10

### Added

* Python
  * Implement PGO for mac intel/aarch64, linux aarch64, and windows x64
  * Add script for installing llvm on linux and mac to support PGO

### Changed

* Python
  * Remove stored pgo profiles
  * Update PGO scripts

## 0.8.0 2025-11-08

Support ndarray inputs for python interpn function, require kwargs for most inputs,
and use numpy's internal borrowchecking interface.

### Changed

* Rust
  * Use numpy borrowchecking system in python interface
  * Update deps
* Python
  * !Require kwargs for most inputs to interpn function
  * Support ndarray inputs and outputs by flattening and reshaping arrays internally
  * Use plotly instead of matplotlib in examples and benchmarks
  * Use PGO for linux builds only, as other platforms will need their own profiles
    * While PGO profiles _could_ be cross-platform portable because they use instrumentation inserted at the
      LLVM IR level, in practice, the function hashes come out different, causing the counts not to be mapped
      correctly
    * Later, we may be able to resolve this by using QEMU to run the PGO builds on each platform or by uploading
      pgo profiles for individual platform runners as build artifacts

## 0.7.0 2025-10-25

Implement `interpn` convenience function that defers directly
to `raw` functions without serializable wrapper classes.

This enables use without pydantic for dep-constrained environments.

Contributions
* [Clément Robert](https://github.com/neutrinoceros)
  * #40
  * #41
  * #42
  * #44
  * #45
  * #46

### Added

* Python
  * Add type stubs for `raw` module
  * Add `interpn` function that defers directly to `raw` functions without serializable wrapper classes

### Changed

* Python
  * !Move pydantic dep to `serde` optional group
  * !Move optional dev dep groups to dependency groups (no longer installable via `.[dep]` syntax)
  * !Import serializable wrapper classes into top level module only if pydantic is found
  * !Drop support for python 3.9, which is leaving long-term support soon
  * Update PGO profile data
* Rust
  * Use abi3-py310 for python bindings

## 0.6.4 2025-10-24

Implement N-dimensional nearest-neighbor interpolation on regular and rectilinear grids.

### Added

* Rust
  * Add `nearest` module with regular- and rectilinear- grid methods in up to 6 dimensions
  * Reduce code duplication for fixed-dim array indexing
  * Update pyo3 and numpy rust deps
* Python
  * Add bindings, tests, benchmarks, and quality-of-fit plots for `NearestRegular` and `NearestRectilinear`
  * Update PGO profile data to include new functions

## 0.6.3 2025-10-22

Unpin max supported python version due to use of stable ABI3,
along with a host of other improvements to packaging and actions workflows.

### New Contributors

* [Clément Robert](https://github.com/neutrinoceros) contributed PRs 30,32,33,36 making up all the substantial changes in this release. Thanks, Clément!

### Changed

* Python
  * Unpin max python version
  * Roll forward pydantic version for python 3.14 compatibility
  * Update package metadata
  * Add python 3.14 to test matrix
* Workflows
  * Reconfigure from single-contributor to commons-project by segmenting release workflows to be dispatched manually only
  * Update python release workflow to depend on test-python instead of release-rust to support separate releases

## 0.6.2 2025-10-20

Add optional use of fused multiply-add, enabled for python distributions.
This substantially improves floating-point roundoff; cubic method now shows
O(1e-14) peak roundoff error even under extrapolation of a quadratic function,
and 0-4 epsilon roundoff inside interpolating region.
Overall effect on throughput performance is neutral.

### Added

* Rust
  * Add `fma` feature

### Changed

* Rust
  * Use Horner's method for evaluating normalized cubic hermite spline
  * If `fma` feature is enabled, use FMA in cubic and linear methods where possible
* Python
  * Enable `fma` feature for python distribution
  * Update pgo profile data and benchmark plots

## 0.6.1 2025-10-10

### Changed

* Python
  * Pass PGO profile-use argument for rustc via maturin args instead of RUSTFLAGS to avoid overriding flags set in .cargo/config.toml
  * Split PGO scripts into native and distribution variants
    * Distribution variant tests the exact build configuration used for distribution
    * Native variant builds with target-cpu=native to enable all available instruction sets
  * Update baked PGO profile based on distribution build
  * Only run pypi distribution for single python version, because ABI3 build is portable to later python versions
* Rust
  * Add x64 to platforms where extra instruction sets are enabled in .cargo/config.toml to capture windows 64-bit x86

## 0.6.0 2025-10-05

Combine python bindings project into rust crate to streamline development process.
Implement PGO (profile-guided optimization) for python releases, giving a 1.2-2x speedup for
dimensions 1-4 at the expense of a ~2x slowdown for higher dimensions.

### Changed

* Python
  * Port python project from `interpnpy` repo
  * Implement PGO with hand-tuned profile workload
  * !Set `linearize_extrapolation=True` as default for cubic interpolators
  * Use pytest-cov for coverage testing
  * Update test deps & add linter/formatter configuration
  * Add uv lock and uv cache configuration
  * Use uv for actions
  * Add more vector instruction sets for x86_64 targets reflecting a ~2015-era CPU
* Rust
  * Eliminate some length check error handling that is no longer necessary for const-generic flattened methods
  * Add PyO3 bindings from python project as `python.rs` module behind `python` feature gate

## 0.2.6 2025-09-27

2-5x speedup and fully-analyzable call stack (no recursion) for lower dimensions
(1D-6D linear, 1D-4D cubic). Recursive methods still available for higher dimensions.

### Changed

* Update rust deps
* Reduce boilerplate in bindings
* Update perf plots
* Use abi3-py39 for extension module build

## 0.2.5 2025-03-14

### Changed

* Update interpn rust dep
  * ~2-6x speedup for lower-dimensional inputs
* Update bindings for latest pyo3 and numpy rust deps
* Add .cargo/config.toml with configuration to enable vector instruction sets for x86 targets
  * ~10-30% speedup for regular grid methods on x86 machines (compounded with earlier 2-6x speedup)
* Regenerate CI configuration
* Support python 3.13

## 0.2.3 2024-08-20

### Changed

* Update interpn rust dep to 0.4.3 to capture upgraded linear methods

## 0.2.2 2024-07-08

### Changed

* Update python deps incl. numpy >2
* Update rust deps
* Support python 3.12