Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
55 changes: 51 additions & 4 deletions 01_intro.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,11 +3,58 @@ title: Introduction
numbering:
enumerator: 1.%s
---
In four-dimensional scanning transmission electron microscopy (4-D STEM) a STEM microscope rasters a small probe with
a diameter of several nm to several Å across a thin sample. At each real space position (r$\_x$, r$\_y$) a pixelated
detector acquires a fully resolved diffraction pattern (k$\_x$, k$\_y$). The resulting dataset, a 4-dimensional
data cube with dimensions (r$\_x$, r$\_y$, k$\_x$, k$\_y$), provides very detailed information about the local structure
of the sample. The high dimensionality of these datasets, however, provide unique challenges related to processing,
scaling, and visualization(Ophus, 2019). The large size of the data requires the development of semi-automated methods
and tools for analysis with associated powerful and fast data pipelines.

The Introduction section of your scientific paper serves as a crucial foundation for your research work. It presents the context, background, and objectives of your study, setting the stage for readers to understand the significance of your contributions. This section should be concise yet comprehensive, providing essential information to the audience.
Next generation datasets from recent development of high-speed detectors, continually push the boundaries for what
experiments are possible. Current cameras operate at 1,000's - 100,000's frames per second. At the low end of these
speeds, a 4-D STEM experiment with 512x512 probe positions will be acquired in less than 5 minutes while at the high
end that same experiment can be done in under 3 seconds. This advancement in speed has driven insitu 4-D STEM
measurements where the structure is measured as a function of temperature(Bi et al., 2023), tilt angle(Bi et al., 2023)
or time. Additionally, high area and high spatially resolved 4-D STEM maps have reduced previous concerns
that 4-D STEM provides too small of an area to adequately determine good statistics about the structure(Levin, 2021).

Provide the broader context in which your research is situated. Clearly articulate the problem or research question your work seeks to address. Highlight the relevance and significance of your study within the scope of interactive and educational content. This is an opportunity to engage readers and emphasize the importance of your research to the field.
In response to these challenges multiple open-source software exist for analyzing 4-D STEM datasets. These include
general efforts like py4dSTEM(Savitzky et al., 2021), pycroscopy(Vasudevan et al., n.d.), and
LiberTEM(Clausen et al., 2020). More specific packages like pyptychostem(Pennycook et al., 2021), and
abtem(Madsen & Susi, 2021), also exist which preform more specialized functions in 4-D STEM. The large number of
these packages helps to drive new technique development and provides method validation as well as helpful comparison.

Make sure you review and summarize the existing literature and relevant prior research related to your study. Identify key papers, theories, methodologies, and technological advancements that have influenced this research field. Discuss the strengths and limitations of previous studies
The HyperSpy project (originally eels-lab) additionally, provides scalable, visualization, machine learning,
computing and metadata handling to a wide range of fields. HyperSpy supports extensions into Kikuchi diffraction
(kikuchipy)(Ånes et al., 2023), luminescence (lumiSpy)(Lähnemann et al., 2023), atom resolution imaging (Atomap)
(Nord et al., 2017), particle analysis (particleSpy)(Slater et al., 2021), holography (holoSpy)(Prestat, de la Peña,
Migunov, et al., 2023), eels and eds measurements (exSpy)(de la Peña et al., 2023) and the package described here
for 4-D STEM (pyxem). This growing ecosystem of packages provides shared resources for data loading, parallel
processing, and visualization. By sharing much of the base code and functionality, methods developed for one domain
transfer easily to another domain. Additionally, shared syntax between different packages allows for a smooth
transition from one package to another.

In the final part of the introduction, clearly outline the specific objectives and research questions that your study aims to address. Describe the original contributions your work brings to the field of interactive education. Highlight any novel methodologies, innovative tools, or unique datasets that you have employed to advance the state of the art.
Interoperability between the specific 4-D STEM packages, however, still leaves something to be desired. Projects like
rosettasciio(Prestat, de la Peña, Lähnemann, et al., 2023) translate and load many different file types. Pyxem uses
this ability to easily load and save many different file types. This helps to bridge the gap between commercial
software and open source. Use of standard packages like dask for parallel and out of memory computing in Abtem,
LiberTEM and pyxem/ HyperSpy allow for easy transfer of data without saving between many of these projects allowing
for hybrid workflows where multiple packages can be used to process one dataset. Additional work to improve
compatibility is a key focus for the pyxem project.

Here we introduce the python package pyxem. Originally started in 2016 by Duncan Johnstone, pyxem has been under
development continually since. Pyxem’s goals are to:

1. Provide easy to use scalable tools for reproducible 4-D STEM to the scientific community.
2. Extend, integrate with, and support existing python packages, fostering collaboration and free exchange of ideas between domains.
3. Adhere to a consistent syntax defined by hyperspy and the scientific python ecosystem.
4. Create a project that is scalable from small to large datasets and future proof for next generation datasets from next generation detectors.
5. Create tools to visualize and interact with datasets while providing consistent and reproducible workflows which can be published in a manner consistent with FAIR practices.
6. Develop of community of users and developers which promote open and shared science.

Pyxem offers both stable and fast processing routines for doing 4-D STEM (with flexibility to handle N-D STEM).
In this publication, we will discuss and showcase the different workflows present in pyxem as well as discuss the
processing speed and general design of the python package. For more details on specific implementation for each method,
cited publications give more information. Additionally, tutorials in the form of jupyter notebooks provide full
workflows for common tasks in pyxem and the pyxem documentation provides detailed information.