Upload details

Contents

  • AiiDA-exported archives for the two sample batches discussed in the paper:

    • eclab.aiida - cycling experiments generated by the importer script from results obtained via EC-lab
    • aurora.aiida - cycling experiments submitted in the AiiDAlab-Aurora app and executed/tracked by AiiDA

  • Robot outputs for the two sample batches discussed in the paper:

    • eclab_robot_output.csv - cycled without AiiDA
    • aurora_robot_output.csv - cycled with AiiDA

  • Two zipped archives as part of the EC-lab batch:

    • eclab_mpr.zip - raw .mpr files produced by EC-lab
    • eclab_json.zip - the above raw .mpr post-processed into .json files by the same tools used by tomato

  • This README.md file

Robot output

Each robot output is a .csv file containing information on the 36 samples of the respective batch, including compositions (e.g. electrodes, electrolyte, separator, etc.), theoretical capacities, electrode masses, and more. This information is loaded into the AiiDAlab-Aurora app and stored as digital twins used when submitting experiments through the app.

EC-lab data

The raw .mpr EC-lab output provided here in eclab_mpr.zip can be processed using yadg, the parsing package used by tomato (see instructions here and here). The resulting .json files include additional cycling metadata that has been stripped from the .json files provided in eclab_json.zip, which is otherwise identical. This was done to reduce file size when used in the AiiDAlab-Aurora app during analysis.

Note for Mac users

On Mac, you may need to unzip the files from terminal with the command unzip <archive>

Usage

For the EC-lab batch, the post-processed .json files can be queried directly.
To query the AiiDA datasets, you will first need to set up an AiiDA environment. You may then proceed to follow the instructions below to explore the data.

Load the archive

  1. Upload the archive file to your AiiDA environment
  2. Import the archive into your AiiDA instance with verdi archive import <archive>.aiida
    • replace <archive> with your archive of interest

Exploring the data

We recommend exploring the data from the built-in AiiDA shell, as it provided convenient auto-completion for dot notation. For example, hitting tab after workflow.inputs. will yield a list of all inputs provided to the workflow.

To start, open the shell with verdi shell and run the following:

workflows = QueryBuilder().append(WorkChainNode, filters={"label": {"like": "%<label>%"}}).all(flat=True)

where <label> is 230511 for archive 1, or 231012 for archive 2.

This will store the archive's workflows in a local variable. You can load the first (or any other) workflow with

wf = workflows[0]  # replace 0 with any index smaller than the result of running `len(workflows)`

Inputs

wf.inputs.battery_sample.attributes
wf.inputs.protocols.<protocol>.attributes
wf.inputs.protocol_order.attributes
wf.inputs.control_settings.<protocol>.attributes

where <protocol> is any protocol given on tab-completion.

Outputs

Exploring the outputs defers slightly between the datasets.

Provenance imported from EC-lab results (eclab_data)

The results node was generated by the importer script and is available at

wf.outputs.results.cycling.attributes

Provenance auto-generated by AiiDA (aurora_data)

The results node is generated automatically when a workflow terminates normally. However, the available workflows in this dataset were terminated prematurely. Nevertheless, the reader may explore the results of completed protocols, including the final protocol terminated by the monitoring.

wf.called_descendants[#].outputs.results

where # is the index of the protocol of interest. Note that descendants do not necessarily follow the execution order defined in wf.inputs.protocol_order.

Advanced querying

More advanced querying may be achieved but is outside the scope of this document. To learn more, please visit the official AiiDA documentation on querying.