Metadata-Version: 2.3
Name: TDAtaframe
Version: 0.2.4
Requires-Dist: polars >=1.5.0
Requires-Dist: pydantic >=2.8.0
Summary: A package to handle shape-matching for meshes
Author-email: Jack Ruder <jackruder@montana.edu>
Requires-Python: >=3.12
Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM

# TDAtaframe
TDA brought to dataframes.

# Install
TDAtaframe is available on pypi as a source distribution. 
The backend requires that you have the appropriate libtorch 2.5.1 libraries 
installed on your system in the standard location.

On Linux/macOS, this is typically /usr/lib, /usr/include, /usr/share for 
library, include, and share files respectively. Symlinking the files will not 
work.

After ensuring that these libraries are present, all that is needed is
```pip install tdataframe```

## Manual libtorch install instructions
It is recommended to use your package manager to install libtorch.
If not readily available for your system, follow the below instructions 
as a workaround.

Download libtorch from [here](https://pytorch.org/).
Then, unzip the files.
`unzip libtorch-*.zip && cd libtorch`
Next, copy all the files to the system. Make sure
you understand what these commands do as they are difficult to undo.
You will likely need to run these commands with `sudo`.
`cp -r lib/* /usr/lib && cp -r include/* /usr/include && cp -r share /usr/share`
Then, update your linker cache with 
`ldconfig`.

## Alternative libtorch locations
todo. see tch-rs docs

