You're reading the documentation for a development version. For the latest released version, please have a look at v0.2.

trepr.processing module

Data processing functionality.

Key to reproducible science is automatic documentation of each processing step applied to the data of a dataset. Each processing step is self-contained, meaning it contains every necessary information to perform the processing task on a given dataset.

Processing steps, in contrast to analysis steps (see trepr.analysis for details), not only operate on data of a trepr.dataset.Dataset, but change its data. The information necessary to reproduce each processing step gets added to the trepr.dataset.Dataset.history attribute of a dataset.

Due to the inheritance from the aspecd.processing module, all processing steps provided are fully self-documenting, i.e. they add all necessary information to reproduce each processing step to the trepr.dataset.ExperimentalDataset.history attribute of the dataset.

Concrete processing steps

This module provides a series of processing steps that can be divided into those specific for tr-EPR data and those generally applicable to spectroscopic data and directly inherited from the ASpecD framework.

What follows is a list as a first overview. For details, see the detailed documentation of each of the classes, readily accessible by the link.

Processing steps specific for tr-EPR data

A number of processing steps are rather specific for tr-EPR data namely correcting DC offsets, background, and microwave frequency:

General processing steps contained in the ASpecD framework

Besides the processing steps specific for tr-EPR data, all processing steps of the underlying ASpecD framework are available. To list those most relevant for tr-EPR spectroscopy:

Implementing own processing steps is rather straight-forward. For details, see the documentation of the aspecd.processing module.

Module documentation

class trepr.processing.PretriggerOffsetCompensation

Bases: SingleProcessingStep

Correct for DC offsets of tr-EPR data.

Usually the first processing step after recording tr-EPR data is to compensate for DC offsets due to experimental instabilities. This is done by setting the average of the pretrigger part of the time trace to zero (pretrigger offset compensation). At the same time, this will remove any background signals of stable paramagnetic species, as they would appear as DC offset as well.

parameters

All parameters necessary for this step.

zeropoint_indexint

Index of the time axis corresponding to t = 0

Will be automatically detected during processing.

Type:

dict

Examples

For convenience, a series of examples in recipe style (for details of the recipe-driven data analysis, see aspecd.tasks) is given below for how to make use of this class. The examples focus each on a single aspect.

In the simplest case, just invoke the pretrigger offset compensation with default values:

- kind: processing
  type: PretriggerOffsetCompensation

This will correct your data accordingly and should always be the first step when processing and analysing tr-EPR data.

class trepr.processing.BackgroundCorrection

Bases: SingleProcessingStep

Subtract background, mainly laser-induced field-independent background.

When the laser hits the EPR cavity, this usually introduces a field-independent absorptive background signal that needs to be subtracted from the data.

Depending on the spectrometer control and measurement software used, this background signal can get automatically subtracted already during the measurement. More often, it needs to be done afterwards, and therefore, it is crucial to record the tr-EPR data with sufficient baseline at both ends of the magnetic field range to allow for reliable background correction.

parameters

All parameters necessary for this step.

num_profileslist

Number of time profiles (transients) to use from lower and upper end of the magnetic field axis.

If two values are provided, a linear regression will be performed between lower and upper end and the background subtracted accordingly. If only a scalar (or a list with one element) is provided, the background traces from the lower magnetic field position are used.

Default: [5, 5]

Type:

dict

Examples

For convenience, a series of examples in recipe style (for details of the recipe-driven data analysis, see aspecd.tasks) is given below for how to make use of this class. The examples focus each on a single aspect.

In the simplest case, just invoke the background correction with default values:

- kind: processing
  type: BackgroundCorrection

This will correct your data accordingly.

If you would like to control more carefully the transients (time profiles) used to obtain the background signal, you can set the respective parameters. Suppose you would want to use only the first 10 transients from the lower end of the magnetic field:

- kind: processing
  type: BackgroundCorrection
  properties:
    parameters:
      num_profiles: 10

Similarly, if you would want to use only the last 10 transients from the lower end of the magnetic field:

- kind: processing
  type: BackgroundCorrection
  properties:
    parameters:
      num_profiles: -10

And finally, if you would like to use the first 5 and the last 10 transients, you would write:

- kind: processing
  type: BackgroundCorrection
  properties:
    parameters:
      num_profiles: [5, 10]
static applicable(dataset)

Check whether processing step is applicable to the given dataset.

Background correction is only applicable to 2D datasets.

Parameters:

dataset (aspecd.dataset.Dataset) – dataset to check

Returns:

applicable – True if successful, False otherwise.

Return type:

bool

class trepr.processing.FrequencyCorrection

Bases: SingleProcessingStep

Convert data to a given microwave frequency.

To compare EPR spectra, it is necessary to first correct them for the same microwave frequency, i.e. to adjust the magnetic field axis accordingly. Note that each individual measurement will have its own microwave frequency. Particularly for tr-EPR data with their usually quite large steps of the magnetic field axis, one could first check whether the difference in microwave frequency is reasonably large compared to the magnetic field steps, and only in this case correct for the same frequency.

parameters

All parameters necessary for this step.

frequencyfloat

Microwave frequency to correct for in GHz.

Default: 9.5

Type:

dict

Examples

For convenience, a series of examples in recipe style (for details of the recipe-driven data analysis, see aspecd.tasks) is given below for how to make use of this class. The examples focus each on a single aspect.

In the simplest case, just invoke the frequency correction with default values:

- kind: processing
  type: FrequencyCorrection

This will correct your data accordingly.

If you would like to set the target microwave frequency explicitly, this can be done as well:

- kind: processing
  type: BackgroundCorrection
  properties:
    parameters:
      frequency: 9.8

In this case, the data would be corrected for a microwave frequency of 9.8 GHz.

Code author: Mirjam Schröder

class trepr.processing.TriggerAutodetection

Bases: SingleProcessingStep

Automatically detect trigger position for time axis.

Depending on the setup used for recording tr-EPR data, either the trigger position (i.e. index of the zero value of the time axis) is set directly in the transient recorder, or it is not set at all (in case the start of recording data of the transient recorder can not be set to times prior to the trigger pulse).

However, a valid trigger position is a prerequisite for pretrigger offset compensation (via PregriggerOffsetCompensation). Therefore, in cases no pretrigger has been set, automatically detecting this position would be helpful.

Note

Auto-detecting the trigger position depends on the statistics of the time trace used. Hence the position will never be the same as if you synchronise it to the laser pulse (or flash lamp trigger, or else). Nevertheless, keep in mind that even in case of triggering the transient recorder by the laser flash detected by a fast photodiode, signal travel time within your cables will be in the range of tens of nanoseconds (for both, the cable connecting the photodiode with the recorder as well as for the signal path within the EPR bridge). Therefore, trigger positions are always somewhat arbitrary and can never be used to obtain accurate delays between laser flash and raise/maximum of the tr-EPR signal.

Important

If you try to auto-detect the trigger position, do not perform a background subtraction (via BackgroundCorrection) before, as this will remove the laser background signal used to auto-detect the trigger position.

Notes

Automatically detecting the trigger position relies on a number of assumptions regarding the shape of a time trace and the underlying statistics:

  • The time trace should be dominated by the laser-induced background resulting in an absorptive signal or otherwise have an absorptive signal.

  • The time trace needs to be recorded starting before the actual laser flash and hence the signal raise. Currently, at least 50 points need to be recorded before the actual laser flash.

  • The trigger position is the position of the time trace where the signal (positively) deviates by a threshold from the value before.

In case of two-dimensional data, the first time trace will be used to detect the trigger position.

The algorithm used for auto-detecting the trigger position works basically as follows:

  • Compute differences of adjacent points of the time trace

  • Smooth the differences by applying a boxcar filter

  • Obtain the first point where the difference is larger than a given threshold

The window length used for smoothing is set to 1/20 of the length of the time trace, and the threshold is computed as the standard deviation of the first 50 points of the time trace multiplied by a factor (n_sigma) that can be set (for details, see below).

parameters

All parameters necessary for this step.

n_sigmaint or float

Threshold used to detect the raise of the signal.

The trigger position is detected by obtaining the first point of the smoothed differences of the time trace whose value is above a threshold. This threshold is calculated as n times the standard deviation (sigma).

Default: 4

Type:

dict

Examples

For convenience, a series of examples in recipe style (for details of the recipe-driven data analysis, see aspecd.tasks) is given below for how to make use of this class. The examples focus each on a single aspect.

Suppose you have recorded a tr-EPR dataset with a setup that does not allow to set the trigger position to somewhere near the actual laser flash. In this case, you will end up with a time axis starting at zero. To perform steps such as pretrigger offset compensation (via PretriggerOffetCompensation), you need to set the trigger position first. Automatically detecting the trigger position and afterwards performing the routine processing steps for tr-EPR data may look like this:

- kind: processing
  type: TriggerAutodetection
- kind: processing
  type: PretriggerOffsetCompensation
- kind: processing
  type: BackgroundCorrection
- kind: singleplot
  type: SinglePlotter2D
  properties:
    filename: overview_poc_bgc.pdf

If you like to adjust parameters, simply provide them in the recipe:

- kind: processing
  type: TriggerAutodetection
  properties:
    parameters:
      n_sigma: 3

In this case, the threshold would be set to 3 times the standard deviation (sigma). Note that you are not limited to integer values, but can give floats as well here.

Added in version 0.2.

static applicable(dataset)

Check whether the processing step is applicable to the dataset.

Trigger autodetection is only possible if the first axis (in case of 1D data) or the second axis (in case of 2D data) is a time axis.