Workflows

The workflow module in EPRpy provides high-level processing pipelines for common pulsed-EPR experiments. It is designed to take an EprData object as input and return a fully processed EprData object, with all intermediate results and processing steps preserved for reproducibility in the data_dict method.

Currently, the module supports workflows for:

  • HYSCORE (Hyperfine Sublevel Correlation) experiments

  • ESEEM (Electron Spin Echo Envelope Modulation) experiments

These workflows include baseline correction, windowing, zero-filling, Fourier transformation, and frequency axis generation. Optionally, symmetrisation of HYSCORE spectra and background fitting strategies for ESEEM are provided.

A workflow can be simply used by loading experimental data with a compatible pulse program :

# import EPRpy
import eprpy as epr

# Load data by providing path to .DSC or .DTA file
hyscore_data = epr.load('hyscore_exp.DSC')

# process using a workflow, zero fill 1024 points
hyscore_processed = hyscore_data.workflow(zf=1024)

Have a look at workflow examples.

Internally, the EprWorkflow Class is used when EprData.workflow() is called. In the example above, hyscore_processed is an EprData object, so all methods of EprData object are also accessible. Other parameters which are supported for a workflow is explained in the section EprWorkflow Class. In the following, the workflow processing steps are elaborated.

HYSCORE processing

The workflow supports pulse programs:

  • HYSCORE

HYSCORE datasets are procssed by by:

  • Applying 2D polynomial baseline correction.

  • Windowing with Hamming functions in both dimensions.

  • Zero-filling (optional).

  • Performing a 2D FFT and shifting the result.

  • Symmetrising the FFT data (optional).

  • Returning all results (raw, intermediate, final) in a dictionary.

The output contains frequency axes in MHz and processed spectra suitable for plotting and further analysis.

If you have a processed EprData object from a HYSCORE data set, the processed arrays can be accessed using EprData.data_dict:

- 'proc_param': Processing parameters used.
- 'raw_data': Original raw data.
- 'baseline_dim1', 'baseline_dim2': Baseline correction results for both dimensions.
- 'bc_data': Baseline-corrected data.
- 'window_dim1', 'window_dim2': Window functions applied.
- 'bc_w_data': Baseline-corrected and windowed data.
- 'time_axis1', 'time_axis2': Time axes (extended if zero-filled).
- 'bc_w_zf_data': Zero-filled, baseline-corrected, windowed data.
- 'frequency_axis1', 'frequency_axis2': Frequency axes for FFT.
- 'FFT_data': 2D FFT of processed data.
- 'FFT_shifted_data': FFT data after shift.
- 'data': Absolute value of shifted FFT (final processed data, optionally symmeterised).
- 'dims': Frequency axes for plotting.
- 'is_complex': Boolean indicating if data is complex (always False).
- 'history': List of processing steps.

ESEEM processing

The workflow supports pulse programs:

  • 2P ESEEM

  • 3P ESEEM

  • 2P ESEEM vs. B0

  • 3P ESEEM vs. B0

  • 3P ESEEM vs tau

ESEEM datasets are processed by:

  • Applying background correction (exponential or polynomial).

  • Windowing with a Hamming function.

  • Zero-filling (optional).

  • Performing a 1D FFT (or along the last axis for 2D data).

  • Returning absolute, shifted FFT spectra along with all intermediate steps.

Results include the processed frequency-domain spectra in MHz, with dimensionality preserved for field-dependent ESEEM maps.

If you have a processed EprData object from a ESEEM data set, the processed arrays can be accessed using EprData.data_dict:

- 'proc_param': Processing parameters used.
- 'raw_data': Original raw data.
- 'baseline_dim1': Baseline correction results.
- 'bc_data': Baseline-corrected data.
- 'window_dim1': Window function applied.
- 'bc_w_data': Baseline-corrected and windowed data.
- 'time_axis1': Time axis (extended if zero-filled).
- 'bc_w_zf_data': Zero-filled, baseline-corrected, windowed data.
- 'frequency_axis1': Frequency axis for FFT.
- 'FFT_data': FFT of processed data.
- 'FFT_shifted_data': FFT data after shift.
- 'data': Absolute value of shifted FFT (final processed data).
- 'dims': Frequency axis (and second axis if 2D data).
- 'is_complex': Boolean indicating if data is complex (always False).
- 'history': List of processing steps.

EprWorkflow Class

The EprWorkflow class encapsulates complete workflows for HYSCORE and ESEEM data processing. It manages parameters like zero-filling, polynomial order for baseline correction, and optional symmetrisation. All results, including raw data, corrected data, intermediate steps, and frequency-domain spectra, are stored in a dictionary that is updated step by step.

Parameters

  • eprdata (EprData) – Instance of the EprData class.

  • zf (int, optional) – Number of points to add via zero filling.

  • poly_order (int, optional, default=3) – Order of the polynomial for baseline correction in case of 3P ESEEM experiments

  • x_max (float, optional) – Upper bound of the Hamming window.

  • pick_eseem_points (bool, default=False) – Whether to pick points along the ESEEM decay curve for background calculation. If False, all data points are used.

  • symmeterise (bool or str, default=False) – If set, symmetrises HYSCORE FFT results across diagonal or anti-diagonal.

Both hyscore() and eseem() return a dictionary containing raw data, intermediate results, processing parameters, and final spectra. Some keys are primarily useful for debugging, while others are essential for end-users who wish to analyse or visualise results.