diffpy.srmise 0.8.0

Peak extraction and peak fitting tool for atomic pair distribution functions.

Implementation of the ParSCAPE algorithm for peak extraction from atomic pair distribution functions (PDFs)

SrMise is an implementation of the ParSCAPE algorithm for peak extraction from atomic pair distribution functions (PDFs). It is designed to function even when a priori knowledge of the physical sample is limited, utilizing the Akaike Information Criterion (AIC) to estimate whether peaks are statistically justified relative to alternate models. Three basic use cases are anticipated for SrMise. The first is peak fitting a user-supplied collections of peaks. The second is peak extraction from a PDF with no (or only partial) user-supplied peaks. The third is an AIC-driven multimodeling analysis where the output of multiple SrMise trials are ranked.

The framework for peak extraction defines peak-like clusters within the data, extracts a single peak within each cluster, and iteratively combines nearby clusters while performing a recursive search on the residual to identify occluded peaks. Eventually this results in a single global cluster containing many peaks fit over all the data. Over- and underfitting are discouraged by use of the AIC when adding or, during a pruning step, removing peaks. Termination effects, which can lead to physically spurious peaks in the PDF, are incorporated in the mathematical peak model and the pruning step attempts to remove peaks which are fit better as termination ripples due to another peak.

Where possible, SrMise provides physically reasonable default values for extraction parameters. However, the PDF baseline should be estimated by the user before extraction, or by performing provisional peak extraction with varying baseline parameters. The package defines a linear (crystalline) baseline, arbitrary polynomial baseline, a spherical nanoparticle baseline, and an arbitrary baseline interpolated from a list of user-supplied values. In addition, PDFs with accurate experimentally-determined uncertainties are necessary to provide the most reliable results, but historically such PDFs are rare. In the absence of accurate uncertainties an ad hoc uncertainty must be specified.

For more information about the diffpy.srmise library, please consult our online documentation.

Citation

If you use this program for a scientific research that leads to publication, we ask that you acknowledge use of the program by citing the following paper in your publication:

L. Granlund, S. J. L. Billinge and P. M. Duxbury, Algorithm for systematic peak extraction from atomic pair distribution functions, Acta Crystallogr. A 4, 392-409 (2015).

Installation

The preferred method is to use Miniconda Python and install from the “conda-forge” channel of Conda packages.

To add “conda-forge” to the conda channels, run the following in a terminal.

conda config --add channels conda-forge

We want to install our packages in a suitable conda environment. The following creates and activates a new environment named diffpy.srmise_env

conda create -n diffpy.srmise_env diffpy.srmise
conda activate diffpy.srmise_env

You can also install diffpy.srmise in another activated environment with

conda install -c conda-forge diffpy.srmise

If the above does not work, you can use pip to download and install the latest release from Python Package Index. To install using pip into your diffpy.srmise_env environment, type

pip install diffpy.srmise

If you prefer to install from sources, after installing the dependencies, obtain the source archive from GitHub. Once installed, cd into your diffpy.srmise directory and run the following

pip install .

This package also provides command-line utilities. To check the software has been installed correctly, type

diffpy.srmise --version

The output should print the latest version displayed on the badges above. You can also type the following command to verify the installation.

python -c "import diffpy.srmise; print(diffpy.srmise.__version__)"

To view the basic usage and available commands, type

diffpy.srmise -h

Getting Started

You may consult our online documentation for tutorials and API references.

Support and Contribute

If you see a bug or want to request a feature, please report it as an issue and/or submit a fix as a PR.

Feel free to fork the project and contribute. To install diffpy.srmise in a development mode, with its sources being directly used by Python rather than copied to a package directory, use the following in the root directory

pip install -e .

To ensure code quality and to prevent accidental commits into the default branch, please set up the use of our pre-commit hooks.

1. Install pre-commit in your working environment by running conda install pre-commit.

2. Initialize pre-commit (one time only) pre-commit install.

Thereafter your code will be linted by black and isort and checked against flake8 before you can commit. If it fails by black or isort, just rerun and it should pass (black and isort will modify the files so should pass after they are modified). If the flake8 test fails please see the error messages and fix them manually before trying to commit again.

Improvements and fixes are always appreciated.

Before contributing, please read our Code of Conduct.

Contact

For more information on diffpy.srmise please visit the project web-page or email the maintainers Simon Billinge([e-mail]).

Acknowledgements

diffpy.srmise is built and maintained with scikit-package.