rexafs for Python

Published user guide · Versioned API reference

Process X-ray absorption spectra with Rust and work with the results as NumPy arrays. Start with Spectrum(energy, mu).fft(); configure only what you need. Energy is in eV, k/q in Å⁻¹, and R in Å.

Install

We recommend uv to manage an analysis project's Python version, dependencies and commands. CPython 3.10–3.14 is supported; start a Python 3.12 project with the stable package:

uv init --python 3.12 rexafs-analysis
cd rexafs-analysis
uv add rexafs==0.2.5 numpy
uv run python -c "import rexafs; print(rexafs.__version__)"

uv add records dependencies in pyproject.toml and resolves their versions in uv.lock. uv run uses the project's .venv automatically; no shell activation is needed. Commit pyproject.toml, .python-version and uv.lock with your analysis, and leave .venv out of version control. NumPy is listed explicitly because the examples import it. See uv's project guide.

PyPI provides platform wheels. Rust is needed only when building rexafs from source.

Version note: keyword constructors, direct configuration setters and XrayFFTR were added in 0.2.5. The basic example also works in 0.2.4.

Load and process a spectrum

Save this as analyze.py inside the project, beside your data file:

import numpy as np
from rexafs import Spectrum

# Two whitespace-delimited columns: energy in eV and absorption mu.
data = np.loadtxt("spectrum.dat")
spectrum = Spectrum(data[:, 0], data[:, 1]).fft()
r, magnitude = spectrum.r(), spectrum.chir_mag()
print(spectrum.e0(), r, magnitude)

Run it with uv run python analyze.py. In your editor, select the Python interpreter from this project's .venv for completion and hover help.

For QAS transmission files, rexafs.io.read_qas_transmission(path) returns a Spectrum using mu = ln(I0 / It), the natural logarithm of incident intensity I0 divided by transmitted intensity It. Both intensities must be positive and in matching units; this produces optical depth rather than an absolute absorption coefficient. The first three whitespace-delimited columns are energy, I0 and It; # starts a comment and additional columns are ignored. Do not use this reader on a file that already contains mu. It accepts str or pathlib.Path:

from rexafs import io
spectrum = io.read_qas_transmission("Ru_QAS.dat").fft()

File/parse failures raise RuntimeError. The reader sorts energy and calculated mu together when needed, retaining duplicate energy rows. It does not enforce positive intensities or finite ratios; processing rejects non-finite data, and duplicates can require cleanup for the selected numerical stage. Check the raw intensities before analysis. The array constructor instead rejects unordered or duplicate energy values.

normalize(), calc_background(), fft() and ifft() return the same spectrum. Missing prerequisite stages run automatically. Results are copied NumPy float64 arrays; a result is None until its stage has run. Check for None when using a result in typed code. Input lists, arrays and strided views are accepted and copied; input energy must be finite and strictly increasing, with matching mu.

Configure only the parameters you need

from rexafs import AUTOBK, PrePostEdge, XrayFFTF, XrayFFTR

spectrum.set_normalization_method(PrePostEdge(pre_edge_end=-30.0))
spectrum.set_background_method(AUTOBK(rbkg=1.0))
spectrum.set_fft(XrayFFTF(kmin=2.0, kmax=12.0, kweight=2.0)).fft()
# Optional: isolate an R range and back-transform it.
spectrum.set_ifft(XrayFFTR(rmin=1.0, rmax=3.0, dr=0.5)).ifft()
q, filtered_chi = spectrum.q(), spectrum.chiq()

These ranges are examples; choose them for your data. Constructors take named arguments, which editors can complete. Fields remain mutable, so existing code such as background = AUTOBK(); background.rbkg = 1.2 still works. BackgroundMethod.AUTOBK(background) and NormalizationMethod.PrePostEdge(parameters) remain available for Rust-style algorithm selection; passing settings directly is the shorter equivalent.

Settings	Recommended starting values
`PrePostEdge()`	Automatically choose E0, fit ranges and polynomial order from the spectrum
`AUTOBK()`	`rbkg=1.0`, `kstep=0.05`, `kweight=1`, `window="Hanning"`, `solver="LinearDirect"`, `clamp_scale_policy="FixedPenalty"`, `clamp_lambda=0.001`
`XrayFFTF()`	`kmin=2.0`, `kmax=15.0`, `kweight=2.0`, `dk=1.0`, `window="KaiserBessel"`, `nfft=2048`, `grid="Input"`
`XrayFFTR()`	`rmin=0.0`, `rmax=20.0`, `dr=1.0`, `rweight=0.0`, `qmax_out=10.0`, `nfft=2048`, automatic `kstep`

Omitting an argument preserves its constructor default. None requests automatic resolution for optional fields; this can differ from the constructor default (for example, XrayFFTF(kmax=None) uses the measured upper k limit). PrePostEdge() follows Rust PrePostEdge::new(), whose ranges are automatic, rather than the fixed ranges in Rust PrePostEdge::default().

Setters copy settings and clear affected results. Editing the original settings later requires calling the setter again. set_e0(eV) clears normalization and all later results; set_fft() preserves chi(k); set_ifft() preserves chi(R). Calling a stage explicitly recomputes it. See the shared API guide.

Some automatic values, including fit ranges and FFT spacings, are retained inside the spectrum on subsequent calls; clearing results does not reset them to None. AUTOBK's automatic kmax and nknots are instead calculated from each input. The original settings object remains unchanged. For example, after changing the background k spacing, reassign forward and inverse settings so their automatic spacings are inferred again:

# Requires 0.2.5 or later; continue with the spectrum above.
spectrum.set_background_method(AUTOBK(kstep=0.1))
spectrum.set_fft(XrayFFTF(kstep=None))
spectrum.set_ifft(XrayFFTR(kstep=None)).ifft()

Apply the same principle to a new scan with a different normalization range: reassign fresh automatic normalization settings instead of retaining the prior scan's resolved bounds. The generated member help explains these choices.

What the calculations mean

Normalization subtracts a fitted pre-edge baseline and divides absorption by its edge step. AUTOBK estimates the smooth background to obtain the EXAFS oscillations, chi(k). The Fourier transform weights and windows those oscillations to display them against R; its peaks are not automatically phase-corrected bond lengths. An inverse transform filters selected R contributions back into q space.

The forward code multiplies an unnormalized, negative-exponent FFT by kstep / sqrt(pi), with no extra division by FFT length. For dimensionless chi and forward exponent w, chi(R) has units Å⁻⁽ʷ⁺¹⁾; the default w=2 gives Å⁻³. The real inverse retains the forward weighting and window, so with default inverse rweight=0, its signal has units Å⁻ʷ and is generally not the original unweighted chi(k).

The processing theory guide explains the equations, symbols, units, assumptions and implementation choices, with scientific references. Use the fitting-statistics guide when interpreting structural fits and uncertainties.

Completion, hover help and errors

The installed package includes py.typed, annotated .pyi files and native runtime docstrings. Select the project's .venv interpreter in your editor (Pylance/Pyright, for example). Hover over parameters for units, defaults and behavior; help(AUTOBK) and help(Spectrum.fft) also work in a terminal. FTWindow, FFTGrid, AUTOBKSolver and AUTOBKClampScalePolicy are Literal type aliases, so editors suggest supported strings and flag invalid choices.

Invalid inputs/normalization raise ValueError; background/FFT failures raise RuntimeError. Stages release the GIL during Rust computation. The public API is rexafs; _core is an implementation detail.

Build from source

From the repository root, with the pinned Rust toolchain installed, use an isolated build environment. This development workflow uses uv venv and uv run --no-project so installing the local extension does not change the library's dependency manifest with analysis-project dependencies:

uv venv --python 3.14
uv pip install maturin numpy
uv run --no-project maturin develop --release --locked
uv run --no-project python py-rexafs/tests/test_api.py

Fitting, groups, structures, plotting and direct ReFEFF calculation remain Rust/desktop APIs. MBack and ILPBkg selectors are unimplemented placeholders and raise errors when processed. See AUTOBK defaults and FFT grid compatibility. Licensed under MIT OR Apache-2.0.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rexafs-0.2.5.tar.gz (3.6 MB view details)

Uploaded Sep 13, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

rexafs-0.2.5-cp314-cp314-win_amd64.whl (2.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.14Windows x86-64

rexafs-0.2.5-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.3 MB view details)

Uploaded Sep 13, 2026 CPython 3.14manylinux: glibc 2.17+ x86-64

rexafs-0.2.5-cp314-cp314-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded Sep 13, 2026 CPython 3.14macOS 11.0+ ARM64

rexafs-0.2.5-cp314-cp314-macosx_10_12_x86_64.whl (1.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.14macOS 10.12+ x86-64

rexafs-0.2.5-cp313-cp313-win_amd64.whl (2.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.13Windows x86-64

rexafs-0.2.5-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.3 MB view details)

Uploaded Sep 13, 2026 CPython 3.13manylinux: glibc 2.17+ x86-64

rexafs-0.2.5-cp313-cp313-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded Sep 13, 2026 CPython 3.13macOS 11.0+ ARM64

rexafs-0.2.5-cp313-cp313-macosx_10_12_x86_64.whl (1.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.13macOS 10.12+ x86-64

rexafs-0.2.5-cp312-cp312-win_amd64.whl (2.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.12Windows x86-64

rexafs-0.2.5-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.3 MB view details)

Uploaded Sep 13, 2026 CPython 3.12manylinux: glibc 2.17+ x86-64

rexafs-0.2.5-cp312-cp312-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded Sep 13, 2026 CPython 3.12macOS 11.0+ ARM64

rexafs-0.2.5-cp312-cp312-macosx_10_12_x86_64.whl (1.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.12macOS 10.12+ x86-64

rexafs-0.2.5-cp311-cp311-win_amd64.whl (2.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.11Windows x86-64

rexafs-0.2.5-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.3 MB view details)

Uploaded Sep 13, 2026 CPython 3.11manylinux: glibc 2.17+ x86-64

rexafs-0.2.5-cp311-cp311-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded Sep 13, 2026 CPython 3.11macOS 11.0+ ARM64

rexafs-0.2.5-cp311-cp311-macosx_10_12_x86_64.whl (1.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.11macOS 10.12+ x86-64

rexafs-0.2.5-cp310-cp310-win_amd64.whl (2.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.10Windows x86-64

rexafs-0.2.5-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.3 MB view details)

Uploaded Sep 13, 2026 CPython 3.10manylinux: glibc 2.17+ x86-64

rexafs-0.2.5-cp310-cp310-macosx_11_0_arm64.whl (1.4 MB view details)

Uploaded Sep 13, 2026 CPython 3.10macOS 11.0+ ARM64

rexafs-0.2.5-cp310-cp310-macosx_10_12_x86_64.whl (1.9 MB view details)

Uploaded Sep 13, 2026 CPython 3.10macOS 10.12+ x86-64

File details

Details for the file rexafs-0.2.5.tar.gz.

File metadata

Download URL: rexafs-0.2.5.tar.gz
Upload date: Sep 13, 2026
Size: 3.6 MB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for rexafs-0.2.5.tar.gz
Algorithm	Hash digest
SHA256	`4a015da63f9f0e7aeae8459ab7b926ddf981add1f81e9b8fabb2737abdb7b7e6`
MD5	`116dfe4d988e3f8a64410941b45ef6ec`
BLAKE2b-256	`449351b51431b2e0c7c754b39f2dd67459a19ddffc10c2b9b2757486632eb6ca`

Algorithm	Hash digest
SHA256	`8f4a0dfef2189f51374fea2004dd8f885dca545a71266e8a7e78f3523d426c11`
MD5	`c3eef52476e7d1bb5e397ab22d9c19ae`
BLAKE2b-256	`b94e04c29f12a4f0ada05ff3d6106acd6379cc0c8dcc62d908ab820093ed072c`

Algorithm	Hash digest
SHA256	`e467a3d21a23724577f162e6f6de2ffd53011c0ec1f4dd4bfb32f9fac17fbdfa`
MD5	`fcffd1d9709a391e8b95a3f9e8d96a19`
BLAKE2b-256	`7bcc8d8bfcd5c68b520b59816d4e0bf49a884ba3637def90061b0fd2f4ecf9a7`

Algorithm	Hash digest
SHA256	`d6f23f40ac926553c2f5167695eab596dfeda64069d0d6f22e2d276b0b35f312`
MD5	`c53acbb52de6100c0fa2fe8fabed67a4`
BLAKE2b-256	`23e3038e3a61441f9940364dc384e36e6ed78887845a5553ee052c972b15912c`

Algorithm	Hash digest
SHA256	`613f97ada5267265896ed9db527ddeb17bfb891ff66eb4bd7b8ba34df02717ea`
MD5	`b43feb7f86b467fb059655ef3805b3a4`
BLAKE2b-256	`ddb6d59b13c54849381ab2d803cfee6d0ddb272f29f1ff064f886c909badc71c`

Algorithm	Hash digest
SHA256	`278210d24c717a6cb021413d00af5a6d67de443feb7a2c513a6d1fd22209881e`
MD5	`99aa415bfbbb0497ef9504098f07830c`
BLAKE2b-256	`8c6befe511727f5a4c31aa84c402d3a66d60124df841a96c10579fe474472f94`

Algorithm	Hash digest
SHA256	`fb54cdb826e48f188b55b9b3d2e3923c21fbe4e83d470a7c3dfb3be64afa6efc`
MD5	`38176048333392deb6875557d5f56e75`
BLAKE2b-256	`a936bd684699b40e0a4e218f717c5eb84245fc01743fe13668a8a120724d5a55`

Algorithm	Hash digest
SHA256	`bb87bf460bb0161889e42dfbed489a1cd7639983ba43d05d9567831ed6dd1f51`
MD5	`ae057af9ececa1453dfc911454d2912d`
BLAKE2b-256	`53e0bfb75e90e114ff98a9c702ad40cefbdd80422f7d81548736d60abe1e854f`

Algorithm	Hash digest
SHA256	`334d2d3451c6a8e66bbd68de293c28d3584d7169f16adda0ce4f9f50db92b01e`
MD5	`a1992b0beba09e1b369989dd946cc56b`
BLAKE2b-256	`e02b8fbb8071355ed56a2c14086a061403ffafd52f61670fab05fd0f25d7ddf4`

Algorithm	Hash digest
SHA256	`ce8c8c4c31678902434e4260122afeb4832b5f8ea29c2e71b619b96395502272`
MD5	`0dbb23d1b30db5046ff46bbd169bf0cb`
BLAKE2b-256	`7e5602dbc408ddbf84f43361dcfd032314be44765fc48af4f34e7fab370400b9`

Algorithm	Hash digest
SHA256	`6f69b86122609e599410990f5fe38a581366f85d201884fcc595856588a65434`
MD5	`55b687ae207c6585590c84d36ac7918d`
BLAKE2b-256	`49a56e4126c585482aac99ee0d0621d5895c16fb653884686cb69bdc5280455a`

Algorithm	Hash digest
SHA256	`4bbbb497c21d540978f1fc5a87324f4a5b66fedcd854f1daffa835d5f2b44981`
MD5	`050789438d4652c581779adbda1aefdf`
BLAKE2b-256	`1012351d153da2dad24d711afd7a759f5fe495335b1719a8ac2fac5d32bd9ca6`

Algorithm	Hash digest
SHA256	`95678ae7702b4514815ddb0c47ee1f34ea043171d013a267f89ec4b6479abbe2`
MD5	`b1081e0cac65e6f8eda9407373a2acbf`
BLAKE2b-256	`54f8a09d45619d6ab8502990743176cbfc0f6521b5ae8cee4928533715a3ffb8`

Algorithm	Hash digest
SHA256	`7329c54fd0e3558848fb2d0e174b9a222dd502a37d55bf4763f253847cebacf0`
MD5	`7b1e09ebe305e9a2a8c4df06e63bc645`
BLAKE2b-256	`f68c4be263fa7ee2e5561de1dff334aea3f084e195c6513d3e31491941ba452c`

Algorithm	Hash digest
SHA256	`f197578a958b3258b76040c753c3afc54818f40362d9cd3768cb0b2dab38df6f`
MD5	`4ba31742f425de4cd8b6a79fd19d2c22`
BLAKE2b-256	`9a70863cc397a95b921ee4d0027e73f2822ffe4efda479effef1fc3e32014164`

Algorithm	Hash digest
SHA256	`fa4ae12e9dc4b373536536eb994e577e939b4fb65824c992c192014c9afd831c`
MD5	`2cea1603cbf40e78447265a35ff65f2a`
BLAKE2b-256	`dc5219812dc4daef1a1a261cbdc1c5920448770da03c3f176d925b1f497c4ee4`

Algorithm	Hash digest
SHA256	`fe169a457e4a702e1825e77c473e39113a420be5209d02f78a77c9ba86a02aa5`
MD5	`f7c0eb935d97418cb7a2377282a9b93c`
BLAKE2b-256	`cd32495f24452990da5f913b33f780b19a248cb790725b70863eb8afd47559cc`

Algorithm	Hash digest
SHA256	`f61166122b84eac0d293432eee12db2f4eae8b6a9aad797486b87195b7757111`
MD5	`a2024a057117b0a4a9c53f61da5a4fd6`
BLAKE2b-256	`cf7a0260f7e78c15504a3d629bc9673ed2183660168596d1ba146855ca01ebd7`

Algorithm	Hash digest
SHA256	`bb438111da65a01f21855ff7d0e34e9d91a5d3456c88215218b3deca9d452a24`
MD5	`7646aa1beff75a1b417a06013f691692`
BLAKE2b-256	`4cacda79f40e897ab04a33648adb639a2de3b827db0350f51836fee631c0786c`

Algorithm	Hash digest
SHA256	`8f301b5fe578cf7a37f57d528c9f0eb89c784c6e6f75c43d71b1f5b7e0024889`
MD5	`2b961bc2f9bb45d2f4f5f74de76f51b3`
BLAKE2b-256	`531881ada55b38d3babdf4ebcd4176eb29cbca47bd561859e90cf52bcc0cb9f6`

Algorithm	Hash digest
SHA256	`462676a0b6d64a24a247a48f16d14dc8b19fb50c3d1f31af731a20fae6311601`
MD5	`0b9bebc4507d03aab9fb312ed78a53de`
BLAKE2b-256	`c4753273748e885efff946950ec455b4ac3050488a77e93631b42355cabb1494`

rexafs 0.2.5

rexafs for Python

Install

Load and process a spectrum

Configure only the parameters you need

What the calculations mean

Completion, hover help and errors

Build from source

Project links

Key dates

1 maintainer

Credits

GitHub Statistics

License

Requires

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details

File metadata

File hashes

Provenance

File details