SpectraFit

---

Data Analysis Tool for All Kinds of Spectra

SpectraFit is a Python tool for quick data fitting based on the regular expression of distribution and linear functions via the command line (CMD) or Jupyter Notebook It is designed to be easy to use and supports all common ASCII data formats. SpectraFit runs on Linux, Windows, and MacOS.

Scope

• Fitting of 2D data, also with multiple columns as global fitting
• Using established and advanced solver methods
• Extensibility of the fitting function
• Guarantee traceability of the fitting results
• Saving all results in a SQL-like-format (CSV) for publications
• Saving all results in a NoSQL-like-format (JSON) for project management
• Having an API interface for Graph-databases

SpectraFit is a tool designed for researchers and scientists who require immediate data fitting to a model. It proves to be especially beneficial for individuals working with vast datasets or who need to conduct numerous fits within a limited time frame. SpectraFit's adaptability to various platforms and data formats makes it a versatile tool that caters to a broad spectrum of scientific applications.

Installation

via pip:

pip install spectrafit

# with support for Jupyter Notebook

pip install spectrafit[jupyter]

# with support for the dashboard in the Jupyter Notebook

pip install spectrafit[jupyter-dash]

# with support to visualize pkl-files as graph

pip install spectrafit[graph]

# with all upcomming features

pip install spectrafit[all]

# Upgrade

pip install spectrafit --upgrade

via conda, see also conda-forge:

conda install -c conda-forge spectrafit

# with support for Jupyter Notebook

conda install -c conda-forge spectrafit-jupyter

# with all upcomming features

conda install -c conda-forge spectrafit-all

Usage

SpectraFit needs as command line tool only two things:

1. The reference data, which should be fitted.
2. The input file, which contains the initial model.

As model files json, toml, and yaml are supported. By making use of the python **kwargs feature, the input file can call most of the following functions of LMFIT. LMFIT is the workhorse for the fit optimization, which is macro wrapper based on:

1. NumPy
2. SciPy
3. uncertainties

In case of SpectraFit, we have further extend the package by:

1. Pandas
2. statsmodels
3. numdifftools
4. Matplotlib in combination with Seaborn

spectrafit data_file.txt -i input_file.json

usage: spectrafit [-h] [-o OUTFILE] [-i INPUT] [-ov] [-e0 ENERGY_START]
                  [-e1 ENERGY_STOP] [-s SMOOTH] [-sh SHIFT] [-c COLUMN COLUMN]
                  [-sep {       ,,,;,:,|, ,s+}] [-dec {.,,}] [-hd HEADER]
                  [-g {0,1,2}] [-auto] [-np] [-v] [-vb {0,1,2}]
                  infile

Fast Fitting Program for ascii txt files.

positional arguments:
  infile                Filename of the spectra data

optional arguments:
  -h, --help            show this help message and exit
  -o OUTFILE, --outfile OUTFILE
                        Filename for the export, default to set to
                        'spectrafit_results'.
  -i INPUT, --input INPUT
                        Filename for the input parameter, default to set to
                        'fitting_input.toml'.Supported fileformats are:
                        '*.json', '*.yml', '*.yaml', and '*.toml'
  -ov, --oversampling   Oversampling the spectra by using factor of 5;
                        default to False.
  -e0 ENERGY_START, --energy_start ENERGY_START
                        Starting energy in eV; default to start of energy.
  -e1 ENERGY_STOP, --energy_stop ENERGY_STOP
                        Ending energy in eV; default to end of energy.
  -s SMOOTH, --smooth SMOOTH
                        Number of smooth points for lmfit; default to 0.
  -sh SHIFT, --shift SHIFT
                        Constant applied energy shift; default to 0.0.
  -c COLUMN COLUMN, --column COLUMN COLUMN
                        Selected columns for the energy- and intensity-values;
                        default to '0' for energy (x-axis) and '1' for intensity
                        (y-axis). In case of working with header, the column
                        should be set to the column names as 'str'; default
                        to 0 and 1.
  -sep { ,,,;,:,|, ,s+}, --separator { ,,,;,:,|, ,s+}
                        Redefine the type of separator; default to ' '.
  -dec {.,,}, --decimal {.,,}
                        Type of decimal separator; default to '.'.
  -hd HEADER, --header HEADER
                        Selected the header for the dataframe; default to None.
  -cm COMMENT, --comment COMMENT
                        Lines with comment characters like '#' should not be
                        parsed; default to None.
  -g {0,1,2}, --global_ {0,1,2}
                        Perform a global fit over the complete dataframe. The
                        options are '0' for classic fit (default). The
                        option '1' for global fitting with auto-definition
                        of the peaks depending on the column size and '2'
                        for self-defined global fitting routines.
  -auto, --autopeak     Auto detection of peaks in the spectra based on `SciPy`.
                        The position, height, and width are used as estimation
                        for the `Gaussian` models.The default option is 'False'
                        for  manual peak definition.
  -np, --noplot         No plotting the spectra and the fit of `SpectraFit`.
  -v, --version         Display the current version of `SpectraFit`.
  -vb {0,1,2}, --verbose {0,1,2}
                        Display the initial configuration parameters and fit
                        results, as a table '1', as a dictionary '2', or not in
                        the terminal '0'. The default option is set to 1 for
                        table `printout`.

Jupyter Notebook

Open the Jupyter Notebook and run the following code:

spectrafit-jupyter

or via Docker Image for <cpu> with amd64 and arm64:

docker pull ghcr.io/anselmoo/spectrafit-<cpu>:latest
docker run -it -p 8888:8888 spectrafit-<cpu>:latest

or just:

docker run -p 8888:8888 ghcr.io/anselmoo/spectrafit-<cpu>:latest

Next define your initial model and the reference data:

from spectrafit.plugins.notebook import SpectraFitNotebook
import pandas as pd

df = pd.read_csv(
    "https://raw.githubusercontent.com/Anselmoo/spectrafit/main/Examples/data.csv"
)

initial_model = [
    {
        "pseudovoigt": {
            "amplitude": {"max": 2, "min": 0, "vary": True, "value": 1},
            "center": {"max": 2, "min": -2, "vary": True, "value": 0},
            "fwhmg": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
            "fwhml": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
        }
    },
    {
        "pseudovoigt": {
            "amplitude": {"max": 2, "min": 0, "vary": True, "value": 1},
            "center": {"max": 2, "min": -2, "vary": True, "value": 1},
            "fwhmg": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
            "fwhml": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
        }
    },
    {
        "pseudovoigt": {
            "amplitude": {"max": 2, "min": 0, "vary": True, "value": 1},
            "center": {"max": 2, "min": -2, "vary": True, "value": 1},
            "fwhmg": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
            "fwhml": {"max": 0.4, "min": 0.1, "vary": True, "value": 0.21},
        }
    },
]
spf = SpectraFitNotebook(df=df, x_column="Energy", y_column="Noisy")
spf.solver_model(initial_model)

Which results in the following output:

Documentation

Please see the extended documentation for the full usage of SpectraFit.

The documentation is generated by .