MTPy

Overview

A modular, extensible, and modifiable python based tool for Meltpool Tomography. The aim of this tool is to create a core library of PBF tools that can be applied in the many areas of our research on the topic.

Current Features

• Streaming of larger-than-RAM datasets from HDD for:
  • analysis
  • plotting
  • Data mining
• No need for binning or other data reduction methods
• Distributed computing, allowing for analysis to take place on compute clusters and larger distributed systems
• Extensive type hinting and structurally subtyped composition. This design ensures that:
  • New modules can easily be added
  • Documentation can be automatically generated
  • Users will get LSP tooltips to help them develop quickly
  • Static analysis can be used to catch errors quickly
  • Modules can easily and seamlessly be implemented in more performant, statically typed languages (e.g: the read_aconity_layers submodule here, written in rust)
• Detailed, togglable user feedback and debug logs, allowing for quick diagnosis of problems when they occur
• Test coverage for core, IO interfaces (as they are most likely parts to error) with plans to expand coverage in the future

Planned Features

• A CLI to avoid the need for scripting when doing common analyses
• A new GUI to allow non-programming researchers to use the tool (previous GUI Melter was a good demonstrator but isn't really fit for our current purposes)
• Reimplementation of the previouslt removed interactive 3d plot feature
• GPU support to further accelerate analysis of extremely large datasets
• Better sample detection via mathematically driven raster detection

ToDo

• implement a CLI for basic functions
• rewrite experimental GUI using flet
• implement Dask GPU support
• Add raster detection component as processor
• fix interactive 3d tomography visualizations (in-progress project)
• Carve out save/load functionality into a separate module